University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA (All Rights Reserved) COLLEGE OF BASIC AND APPLIED SCIENCES DEVELOPMENT OF NATIONAL INDICATION-BASED DIAGNOSTIC REFERENCE LEVELS AND OPTIMISATION METHODS FOR COMPUTED TOMOGRAPHY EXAMINATIONS IN GHANA BENARD BOTWE DEPARTMENT OF NUCLEAR SAFETY AND SECURITY JULY, 2020 University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA (All Rights Reserved) COLLEGE OF BASIC AND APPLIED SCIENCES DEVELOPMENT OF NATIONAL INDICATION-BASED DIAGNOSTIC REFERENCE LEVELS AND OPTIMISATION METHODS FOR COMPUTED TOMOGRAPHY EXAMINATIONS IN GHANA BY BENARD BOTWE (10600444) A DISSERTATION SUBMITTED TO THE SCHOOL OF GRADUATE STUDIES IN PARTIAL FULFILMENT OF THE AWARD OF DEGREE OF DOCTOR OF PHILOSOPHY IN RADIATION PROTECTION DEPARTMENT OF NUCLEAR SAFETY AND SECURITY JULY, 2020 University of Ghana http://ugspace.ug.edu.gh DECLARATION I hereby, declare that this dissertation is the result of original research work undertaken by Benard Botwe in the School of Nuclear and Allied Sciences (SNAS), University of Ghana (UG), under the supervision of Professor Cyril Schandorf (SNAS UG-Legon), Dr. Stephen Inkoom (SNAS UG-Legon) and Professor Augustine Faanu (SNAS UG-Legon). This work has not been previously accepted in substance for any degree, and neither is it being concurrently submitted in candidature for any degree. Sign: ……………………….. Date: ………………….. Benard Botwe (Candidate) Sign: ……………………….. Date: ………………….. Prof. Cyril Schandorf (Principal Supervisor) Sign: ……………………….. Date: ………………….. Dr. Stephen Inkoom (Co-Supervisor) Sign: ……………………….. Date: ………………….. Prof. Augustine Faanu (Co-Supervisor) I University of Ghana http://ugspace.ug.edu.gh ABSTRACT Background: Diagnostic reference levels (DRLs) and dose optimisation methods are crucial for effective radiation dose management. Countries utilising ionising radiation for medical purposes are required to develop and implement them, taking into consideration their clinical situations, infrastructure, population characteristics as well as social, technical and economic factors. However, in Ghana, there is no established national indication-based DRL. Main Objective: The main objective of this study was to develop national indication-based DRL values for common and prioritised indications of the adult human body for clinical application in Ghana. It was also to assess the risk of undertaking each indication-based CT examination, and also propose some steps for dose optimisation. Materials and Methods: The methodological approach recommended by the International Commission for Radiological Protection (ICRP), publication 135, for the development of DRLs, was employed. Studies on CT infrastructure and common indications as well as quality management systems (QMS) were conducted. Quality control (QC) tests were undertaken using a CT dose profiler, barracuda set, uniform water phantom and an ImageQC software v.1.43. Radiologists were mainly requested to define the basic diagnostic requirement of each indication. Dose descriptors such as volume weighted CT dose index (CTDIvol) and dose length product (DLP) of reported CT scans were retrieved from the picture archiving and communication system (PACS) of scanners, constituting 71.4% of the total CT scanners in Ghana. Overall, 3,960 data sets were collected for all the common and prioritised indications which included: cerebrovascular accident (CVA) or stroke, head trauma/injury, brain tumour/space occupying lesion (SOL), lung tumour/cancer, chest lesion with chronic kidney disease, abdominopelvic lesion, kidney stone, urothelial malignancy/CT intravenous urography (CT-IVU) and pulmonary embolism (PE). II University of Ghana http://ugspace.ug.edu.gh ImageJ software version 1.52 was used to analyse the objective image qualities. Statistical Package for the Social Sciences (SPSS) version 23.0 was used to extract the DRL values for the common indications in CT examinations. Microsoft excel version 2013 was used to pictorially project the results and also develop a tool (BOTB) for dose monitoring. Lifetime Attributable Risk (LAR) of cancer incidence and mortality were estimated for various organs using a Monte Carlo-based software (National Cancer Institute Dosimetry System CT software version 2.1) and the Biological Effects of Ionising Radiation (BEIR) VII model. An anthropomorphic Alderson RANDO phantom and patients’ clinical data were used to explore an optimisation method for cerebrovascular accident (CVA) imaging. Regression analyses were further used to model equations for organ doses in CVA imaging. CT phantom PBU-60 was also used to evaluate automatic exposure control (AEC) dose impact in facilities operating without AEC systems. In all inferential analyses, a p- value of ≤ 0.05 was used to interpret the findings as statistically significant. Main Results: The various indications and their respective projected DRL values in terms of CTDIvol (mGy) and DLP (mGy.cm) were CVA/stroke (77 mGy; 1313 mGy.cm), head trauma/injury (76 mGy; 1596 mGy.cm), brain tumour/SOL (77 mGy; 2696 mGy.cm), lung tumour/cancer (12 mGy; 828 mGy.cm) and chest lesion with chronic kidney disease (13 mGy; 467 mGy.cm). Others were abdominopelvic lesion (17 mGy; 1299 mGy.cm), kidney stone (15 mGy; 731 mGy.cm), urothelial malignancy/CT-IVU (11 mGy; 1449 mGy.cm) and pulmonary embolism (14 mGy; 942 mGy.cm). The risk of PE radiation-induced breast cancer ranged from 6-115.8 people in 100,000 procedures. Moreover, CT-IVU radiation-induced colon cancer risks ranged from 53.3-66.4 people in 100,000 procedures. About 1 in 38,462 to 1 in 14,706 patients were also likely to develop ovarian cancer due to CT-IVU examinations in Ghana. A novel examination protocol was further developed in the study that could be used to scan CVA related conditions III University of Ghana http://ugspace.ug.edu.gh with optimal image quality, while reducing the mean effective dose of the facilities by 23.8%, and organ doses by 32% (lens), 70.7% (spinal cord), 57.2% (thyroid) and 75.6% (oral cavity). Moreover, eight organ dose equations were developed to aid in dose management. Finally, if AEC are used in facilities operating without such systems, radiation dose levels could also be reduced by a range of 46.4-58.3% without any significant compromise on image quality. Conclusion: The projected indication-based DRL values and optimisation methods could be used to manage CT radiation dose in Ghana. KEYWORDS Diagnostic Reference Levels (DRLs); Optimisation; Computed Tomography; Indication-Based DRLs; Cancer Risk. IV University of Ghana http://ugspace.ug.edu.gh DEDICATION This research is dedicated to the Ohene-Botwe, Kwarteng, Obeng-Nkansah and Andoh families. V University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENTS I will forever thank the almighty God for His guidance, mercies, protection and love. My special thanks go to my supervisors: Professor Cyril Schandorf, Dr. Stephen Inkoom, and Professor Augustine Faanu for their kind advice, suggestions and constant help throughout the research period. I also thank Prof. Pål Erik Goa, Linn Rolstadaas, Atle Hegge and Tommy Berglund for their scholarly advice during my exchange experience at the Norwegian University of Science and Technology (NTNU). I am extremely grateful to Dr. D. Okoh Kpeglo (HOD) and faculty members of School of Nuclear and Allied Sciences for their support. I thank the Norwegian Partnership Programme for Global Academic Cooperation (NORPART), particularly, the Ghana-Norway NORPART project facilitators, especially, Prof. J.J. Fletcher and Dr. M. Afadzi. The student exchange opportunity I had with the help of the Ghana-Norway NORPART project was a great learning opportunity for me. My special thanks also go to Dr. S. Anim-Sampong, Dr. W.K. Antwi and Dr. F. Hasford and Dr. I. Shirazu for their encouragement and scholarly advice. Further, I express my heartfelt gratitude to Prof. Mary Boadu for her support. In addition, I wish to thank the Technical Heads of CT facilities who helped in this study. Particularly, I thank Mr. B.B. Ofori- Manteaw, Mr. Evan Tettey, Rev. D. Atawone, Dr. B. D. Sarkodie, Mr. N.O. Amoah, Mr F. Botwe, Mr. Kofi Antwi, Mr Atta Osei and Mr. T. Ntiri for their assistance. I am also grateful to Mr. C.E. Kokah, Miss J.A. Mensah-Doku, Miss R.N.A. Smillie, Mr. A.A. Adi and Mr. A-B Owusu. To Mr. N.S Korto, Mr. S. Basoa, Mr V.F. Dacosta, Mr. K. Etete, Mr. S. Fanusah, Madam A. Frimpong and all who helped in diverse ways, I say may God bless you. I am extremely thankful to the Sweden Ghana Medical Centre and the Nuclear Regulatory Authority (NRA), for their support. Finally, I thank the BaNGA-Africa project funded by the Carnegie Corporation of New York with University of Ghana and the ISRRT-Chesney research fund managers for their financial support. VI University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENTS DECLARATION ............................................................................................................................. I ABSTRACT .................................................................................................................................... II DEDICATION ................................................................................................................................ V ACKNOWLEDGEMENT ............................................................................................................ VI TABLE OF CONTENTS ............................................................................................................. VII LISTS OF FIGURES ................................................................................................................. XIV LISTS OF TABLES ................................................................................................................. XVIII ABBREVIATIONS ................................................................................................................. XXIII CHAPTER ONE ............................................................................................................................. 1 INTRODUCTION .......................................................................................................................... 1 1.1 BACKGROUND ........................................................................................................................ 1 1.2 STATEMENT OF THE PROBLEM ................................................................................................ 4 1.3 SCOPE OF THE STUDY ............................................................................................................. 7 1.4 MAIN OBJECTIVE .................................................................................................................... 7 1.5 SPECIFIC OBJECTIVES ............................................................................................................. 8 1.6 RELEVANCE AND JUSTIFICATIONS FOR THE STUDY ................................................................ 9 1.6.1 New Research Area in Ghana ........................................................................................ 9 1.6.2 Policy Implications ......................................................................................................... 9 1.6.3 Health Implications ...................................................................................................... 10 1.6.4 Research Implications .................................................................................................. 11 1.7 ORGANISATION OF THESIS .................................................................................................... 11 CHAPTER TWO .......................................................................................................................... 12 LITERATURE REVIEW ............................................................................................................. 12 VII University of Ghana http://ugspace.ug.edu.gh 2.0 OVERVIEW ............................................................................................................................ 12 2.1 COMPUTED TOMOGRAPHY IMAGING .................................................................................... 12 2.2 BASIC APPLICATION OF COMPUTED TOMOGRAPHY IN MEDICINE ......................................... 15 2.3 COMPUTED TOMOGRAPHY DOSIMETRY ................................................................................ 17 2.3.1 Computed Tomography Dose Index (CTDI) ................................................................ 17 2.3.2 CTDIFDA ........................................................................................................................ 18 2.3.3 CTDI100 ......................................................................................................................... 19 2.3.4 Weighted CT Dose Index (CTDIw)................................................................................ 19 2.3.5 Volume Weighted CT dose Index (CTDIvol) .................................................................. 20 2.3.6 Dose Length Product .................................................................................................... 21 2.3.7 Effective Dose ............................................................................................................... 22 2.3.8 Size-Specific Dose Estimate (SSDE) ............................................................................. 23 2.4 TYPICAL RADIATION DOSES IN COMPUTED TOMOGRAPHY .................................................. 24 2.5 RADIATION RISKS ................................................................................................................. 26 2.6 DOSE MANAGEMENT AND OPTIMISATION IN COMPUTED TOMOGRAPHY IMAGING ............... 32 2.6.1 Quality Management Systems ....................................................................................... 32 2.6.2 Justification .................................................................................................................. 33 2.6.3 Optimisation ................................................................................................................. 33 2.6.3.1 Tube Current and Tube Loading (mAs) ................................................................. 34 2.6.3.2 Tube Voltage (kVp) ............................................................................................... 35 2.6.3.3 Pitch ....................................................................................................................... 35 2.6.3.4 Scan Thickness....................................................................................................... 36 2.6.3.5 X-ray Beam-Shaping Filter .................................................................................... 36 2.6.3.6 Scan Length and Z-Axis Overscan ........................................................................ 36 VIII University of Ghana http://ugspace.ug.edu.gh 2.6.3.7 Detector Configuration .......................................................................................... 37 2.6.3.8 Automatic Exposure Control/Tube Current Modulation ....................................... 37 2.6.3.9 Reconstruction Algorithm ...................................................................................... 40 2.6.3.10 Patient Positioning ............................................................................................... 40 2.7 DIAGNOSTIC REFERENCE LEVEL .......................................................................................... 41 2.7.1 Diagnostic Reference Level in Medicine ...................................................................... 41 2.7.2 Role of DRLs in Computed Tomography dose Optimisation ....................................... 44 2.7.3 Anatomical DRLs in Computed Tomography ............................................................... 46 2.7.4 Indication-Based Diagnostic Reference Levels ............................................................ 51 2.7.5 Existing Indication-Based DRLs in Computed Tomography ....................................... 54 2.7.6 Theories Guiding DRL Development in Computed Tomography ................................ 59 CHAPTER THREE ...................................................................................................................... 67 MATERIALS AND METHODS .................................................................................................. 67 3.0 OVERVIEW ............................................................................................................................ 67 3.1 STUDY DESIGN ..................................................................................................................... 67 3.2 STUDY AREA AND COMPONENTS.......................................................................................... 68 3.3 PHASE 1 STUDY: CT TECHNICAL DATA, COMMON INDICATIONS AND IMAGING REQUIREMENTS ................................................................................................................................................... 71 3.3.1 Study Site and Population ............................................................................................. 71 3.3.2 Sample Size, Inclusion and Exclusion Criteria ............................................................ 73 3.3.3 Data Collection Tool .................................................................................................... 73 3.3.4 Questionnaire Validity and Reliability ......................................................................... 75 3.3.5 Data Collection Procedure ........................................................................................... 76 3.3.6 Data Analysis ................................................................................................................ 77 IX University of Ghana http://ugspace.ug.edu.gh 3.4 PHASE 2 STUDY: SCANNERS’ PERFORMANCE CHARACTERISTICS (QC TESTS) ...................... 78 3.4.1 Outline .......................................................................................................................... 78 3.4.2 Materials ....................................................................................................................... 80 3.4.3 Data Collection Procedure ........................................................................................... 82 3.4.3.1 CT Dose Delivery Accuracy .................................................................................. 82 3.4.3.2 CT Dose Delivery Reproducibility ........................................................................ 85 3.4.3.3 Geometric Efficiency ............................................................................................. 86 3.4.3.4 Tube Voltage Accuracy and Half Value Layer ...................................................... 87 3.4.3.5 CT Number (Water), Image Noise Testing and Homogeneity .............................. 88 3.5 PHASE 3 STUDY: CT DOSE DATA AND IMAGE QUALITY ASSESSMENT ................................. 92 3.5.1 Outline .......................................................................................................................... 92 3.5.2 Study Population........................................................................................................... 92 3.5.3 Type of Data Set Collected ........................................................................................... 93 3.5.4 Patients’ Data Size ....................................................................................................... 94 3.5.5 Data Acquisition Tool and Process .............................................................................. 94 3.5.6 Image Quality Assessment of the Collected Image Data .............................................. 95 3.5.7 Statistical Control of Collected Dose Descriptors ....................................................... 97 3.6 PHASE 4 STUDY: DRL VALUES ESTIMATION AND DOSE MONITORING TOOL ...................... 99 3.6.1 Outline .......................................................................................................................... 99 3.6.2 DRL Values Estimation ................................................................................................ 99 3.6.3 Tool for Dose Monitoring ............................................................................................. 99 3.7 PHASE 5 STUDY: DOSE IMPACT ON PATIENTS AND CANCER RISK ...................................... 101 3.7.1 Outline ........................................................................................................................ 101 3.7.2 Effective Dose ............................................................................................................. 101 X University of Ghana http://ugspace.ug.edu.gh 3.7.3 Cancer Risk................................................................................................................. 102 3.8 PHASE 6 STUDY: OPTIMISATION ......................................................................................... 107 3.8.1 Optimisation Method 1: Management of Scan Length ............................................... 107 3.8.1.1 Phantom-Based Optimisation Study .................................................................... 110 3.8.1.1.1 Materials ........................................................................................................ 110 3.8.1.1.2 Procedure ....................................................................................................... 111 3.8.1.1.3 Image Quality Assessment ............................................................................ 113 3.8.1.1.4 Data Analysis ................................................................................................ 113 3.8.1.2 Patient-Based Optimisation Study ....................................................................... 114 3.8.1.2.1 Materials ........................................................................................................ 114 3.8.1.2.2 Sample Size Determination ........................................................................... 115 3.8.1.2.3 Inclusion and Exclusion Criteria ................................................................... 116 3.8.1.2.4 Procedure ....................................................................................................... 116 3.8.1.2.5 Radiation Dose Assessment .......................................................................... 120 3.8.1.2.6 Image Quality Assessment ............................................................................ 120 3.8.1.2.7 Data Analysis ................................................................................................ 120 3.8.2 Optimisation Method 2: The Role of AEC Utilisation ................................................ 122 3.8.2.1 Overview .............................................................................................................. 122 3.8.2.2 Materials .............................................................................................................. 122 3.8.2.3 Methods and Procedure ........................................................................................ 124 3.8.2.4 Image Quality and Data Analysis ........................................................................ 125 3.8.2.5 Data analysis ........................................................................................................ 125 3.9 ETHICAL CONSIDERATION AND DATA HANDLING .............................................................. 126 CHAPTER FOUR ....................................................................................................................... 128 XI University of Ghana http://ugspace.ug.edu.gh RESULTS AND DISCUSSION ................................................................................................. 128 4.1 OVERVIEW .......................................................................................................................... 128 4.2 PHASE 1: CT TECHNICAL DATA, COMMON INDICATIONS AND IMAGING REQUIREMENTS 128 4.2.1 CT Technical Data and Common Indications ............................................................ 128 4.2.2 Definition of Basic Imaging Requirements for Indications ........................................ 146 4.3 PHASE 2: PERFORMANCE CHARACTERISTICS DATA ON CT SCANNERS ............................ 153 4.4 PHASES 3 & 4: CT DOSE AND IMAGE QUALITY ASSESSMENT AND ESTIMATION OF DRLS ................................................................................................................................................. 162 4.5 PHASE 5: DOSE IMPACT AND ESTIMATION OF CANCER RISK ASSOCIATED WITH THE DOSES ................................................................................................................................................. 194 4.6 PHASE 6: OPTIMISATION METHODS .................................................................................. 205 4.6.1 Optimisation method 1: Dose reduction through optimisation of scan length........... 205 4.6.1.1 Phantom-Based Optimisation study ..................................................................... 207 4.6.1.2 Patient-Based Optimisation Study ....................................................................... 208 4.6.2 Regression Modelled Equations ................................................................................. 214 4.6.3 Optimisation Method 2: The role of AEC Utilisation................................................. 216 CHAPTER FIVE ........................................................................................................................ 220 SUMMARY, CONCLUSION AND RECOMMENDATIONS ................................................. 220 5.0 OVERVIEW .......................................................................................................................... 220 5.1 SUMMARY AND CONCLUSION ............................................................................................. 220 5.2 CHALLENGES/LIMITATIONS ................................................................................................ 222 5.3 RECOMMENDATIONS .......................................................................................................... 223 5.3.1 Nuclear Regulatory Authority .................................................................................... 223 5.3.2 Managers and Health Professionals in CT Imaging .................................................. 224 5.3.3 Future Research ......................................................................................................... 224 XII University of Ghana http://ugspace.ug.edu.gh REFERENCES ........................................................................................................................... 225 APPENDIX I: Number of Scanners on NRA’s Records ............................................................ 262 APPENDIX II: Introductory Letter From NRA ......................................................................... 263 APPENDIX III: Participant Information Sheet .......................................................................... 264 APPENDIX IV: Consent Form................................................................................................... 267 APPENDIX V: Questionnaire A ................................................................................................ 268 APPENDIX VI: Questionnaire B ............................................................................................... 274 APPENDIX VII: CT Dose Parameters Data Sheet .................................................................... 277 APPENDIX VIII: Control Chart for All Indication Data Sets (CTDIVOL and DLP) ................. 286 APPENDIX IX: Distances Covered above and below Upper Targets ....................................... 289 APPENDIX X: Testing of Normality ......................................................................................... 290 APPENDIX XI: Testing for Outlies ........................................................................................... 291 APPENDIX XII: Test for Multicollinearity ............................................................................... 292 APPENDIX XIII: Ethical Approval-University of Ghana Ethics Committee for Basic and Applied Sciences (ECDAS) ........................................................................................................ 293 APPENDIX XIV: Ethical Approval-Ghana Health Service Ethics Review Committee ............ 294 APPENDIX XV: Ethical Approval -The Korle Bu Teaching Hospital Institutional Review Board (KBTH) ....................................................................................................................................... 295 APPENDIX XVI: Other Permission Letters .............................................................................. 297 APPENDIX XVII: Supporting Document Showing an Expanded Version of the Methodological or Conceptual Framework (Figure 3.1) used for Conducting Phases 1 To 4 Studies. ................ 301 APPENDIX XVIII: Publication Lists ......................................................................................... 302 APPENDIX XIX: Conference and Poster Presentation ............................................................. 303 APPENDIX XX: Awards Associated with the Study ................................................................ 304 APPENDIX XXI: Copies of Published Articles ......................................................................... 306 XIII University of Ghana http://ugspace.ug.edu.gh LISTS OF FIGURES Chapter Two Figure 2.1: Schematic diagram of an x-ray tube and production (Plate A), bremsstrahlung interactions (Plate B), “K-shell” emission (Plate C) and a typical CT imaging setup consisting the and x-ray beam spectrum and its relationship with the patient and detectors for CT imaging (Plate D) (Radiologycafe, 2019; Disher et al., 2006) ............................................................ 13 Figure 2.2: A simple schematic diagram of CT imaging process (Sprawls, 2019). ..................... 14 Figure 2.3: AP and LAT diameter measurements ........................................................................ 23 Figure 2.4: Schematic diagram of radiation impact on DNA, where plates A and B show direct and indirect impacts of radiation, respectively (Teach Nuclear, 2019). ...................................... 27 Figure 2.5: Diagram of deterministic (a) and stochastic (b) effect-dose response curves (Rahman, 2018). ..................................................................................................................................... 28 Figure 2.6: A diagram showing the 75th percentile value of a dose distribution (Vock and Frija, 2016). ..................................................................................................................................... 65 Chapter Three Figure 3.1 Methodological framework and flow chart. ................................................................ 70 Figure 3.2: Regional distribution of CT scanners in Ghana as at December 2017 ....................... 72 Figure 3.3: Flowchart of scanner selection ................................................................................... 79 Figure 3.4: RTI CT dose profiler (RTI Group, 2016). .................................................................. 80 Figure 3.5: Barracuda set (A= Cabinet, and B=Multipurpose Detector, MPD) (RTI Group, 2016). ............................................................................................................................................... 80 Figure 3.6: Standard PMMA Phantoms (A; head, A and B fused together to form body phantom) (RTI Group, 2016). ................................................................................................................ 81 Figure 3.7: Experimental set up for dose delivery accuracy tests................................................. 84 XIV University of Ghana http://ugspace.ug.edu.gh Figure 3.8: CT Dose profiler dose output for dose delivery accuracy. ......................................... 85 Figure 3.9: Water-filled phantom for CT number (water), homogeneity and image noise testing. ............................................................................................................................................... 88 Figure 3.10 : CT number (arrow A), image noise (arrow B) analysis using ImageQC v.1.43. .... 90 Figure 3.11: Uniformity/homogeneity analysis using ImageQC v.1.43. ...................................... 91 Figure 3.12: Positioning of ROI in chest (A), abdomen and pelvis (B) and head (C) examinations ............................................................................................................................................... 96 Figure 3.13: Graphical user interface (GUI) of the NCICT program showing an example of entered patient- and scan-specific parameters and estimated organ dose calculation for adult male head (brain tumour/SOL) scan. .................................................................................................... 105 Figure 3.14: The measurements of DAUT and DBLT for each indication ................................ 109 Figure 3.15: An anthropomorphic Alderson RANDO phantom as shown in Plate A, while Plate B shows the position of the phantom in a Siemens CT Somatom Emotion scanner. ............. 111 Figure 3.16: Scan coverages used in scanning the Alderson RANDO phantom. ....................... 112 Figure 3.17: A typical scan coverage (grid areas) for a “Full range” CVA CT. Plate A shows the last two slices from the base of skull (caudally) and Plate B shows the last two slices cranially which contain no information for CVA diagnosis. The red arrow shows a typical mean DAUT and the black arrow shows a typical mean DBLT used across the facilities. ...................... 117 Figure 3.18: A typical scan coverage (grid areas) for a “Reduced range” CT for CVA. Plate I shows the last two slices from the skull base (caudally) and Plate II shows the last two slices cranially. The red arrow shows a 1 cm allowable distance from the vertex of the skull that can be used to optimise radiation in CVA examinations without compromising on image quality. ...... 118 Figure 3.19: Unnecessary areas (grid lines) along z-axis when Reduced range CT area (red arrow) was used and their corresponding images. .......................................................................... 119 Figure 3.20: Positioning of PBU-60 phantom for AEC studies ................................................. 124 XV University of Ghana http://ugspace.ug.edu.gh Chapter Four Figure 4.1: Equipment ability to display CTDIvol and DLP parameters (a); the scan mode systems incorporated in the CT scanners (b); availability of AEC systems on the CT scanners (c). 132 Figure 4.2: Regularity of routine QC (N=20) ............................................................................. 137 Figure 4.3: Responses on (A) whether or not CT facilities benchmark their dose information with internationally established indication-based DRLs and (B) whether or not indication-based DRLs were needed in Ghana. .............................................................................................. 141 Figure 4.4: Percentage distribution of CT examinations per region (Plate A), anatomical part (Plate B) and hospital type (C)....................................................................................................... 142 Figure 4.5: Grade of the respondents .......................................................................................... 146 Figure 4.6: Distribution of representative CTDIvol values for CVA CT. .................................... 166 Figure 4.7: Distribution of representative DLP values for CVA CT .......................................... 167 Figure 4.8:Distribution of representative CTDIvol values for head trauma or injury CT ............ 169 Figure 4. 9: Distribution of representative DLP values for head trauma/injury CT ................... 170 Figure 4.10: Distribution of representative CTDIvol values for brain tumour/SOL CT indication ............................................................................................................................................. 172 Figure 4.11: Distribution of representative DLP values for brain tumour/SOL CT indication .. 173 Figure 4. 12: Distribution of representative CTDIvol values for lung tumour/cancer CT indication ............................................................................................................................................. 175 Figure 4.13: Distribution of DLP values forlung tumour/cancer CT indication ......................... 176 Figure 4.14: Distribution of representative CTDIvol values for CT of lung lesion with CKD indication. ............................................................................................................................ 178 Figure 4.15: Distribution of DLP values for CT of lung lesion with CKD indication ............... 179 Figure 4.16: Distribution of representative CTDIvol values for CT of AP lesion indication ...... 180 Figure 4.17: Distribution of DLP values for CT of AP lesion indication ................................... 181 Figure 4.18: Distribution of CTDIvol values for CT of kidney stone indication ......................... 183 XVI University of Ghana http://ugspace.ug.edu.gh Figure 4.19: Distribution of DLP values for CT of kidney stone indication .............................. 184 Figure 4.20: Distribution of CTDIvol values for CT of urothelial malignancy (CT-IVU) indication ............................................................................................................................................. 186 Figure 4.21: Distribution of DLP values for CT of urothelial malignancy indication ............... 187 Figure 4.22: Distribution of CTDIvol values for CT of pulmonary embolism indication ........... 189 Figure 4.23: Distribution of DLP values for CT of pulmonary embolism indication ................ 190 Figure 4.24: Interphase of a Microsoft Excel-based developed tool (BOTB) for inspection and monitoring of DRLs compliance purposes. ......................................................................... 192 Figure 4.25: Pictorial presentation of the organ doses for the reduced and full range CVA CT scans ............................................................................................................................................. 212 Figure 4.26: AEC modulated tube current and fixed tube current profiles across different z- positions ............................................................................................................................... 217 XVII University of Ghana http://ugspace.ug.edu.gh LISTS OF TABLES Chapter Two Table 2.1: Lifetime attributable risk of cancer incidence coefficients from Table 12D-1 of BEIR Report (National Research Council, 2006) ............................................................................ 30 Table 2.2: Lifetime attributable risk of cancer mortality coefficients from Table 12D-2 of BEIR report (National Research Council, 2006) ............................................................................. 31 Table 2.3: AEC techniques available in modern CT systems ....................................................... 38 Table 2.4: The AEC systems from major manufacturers ............................................................. 39 Table 2.5: Reviewed existing adult national anatomical DRLs .................................................... 49 Table 2.6: Reviewed existing paediatric national anatomical DRLs ............................................ 50 Table 2.7: Adult head CT DRLs, based on clinical indications.................................................... 55 Table 2.8: Adult chest CT DRLs, based on clinical indications ................................................... 56 Table 2.9: Adult abdomen and pelvic region CT DRLs, based on clinical indications ................ 57 Table 2.10: Adult cervical CT DRLs, based on clinical indications ............................................ 58 Table 2.11 Examination selection and ICRP’s recommended assessment method for DRLs...... 62 Chapter Three Table 3.1: ICRP Publication 103 recommended region-specific DLP-ED conversion factors ... 101 Table 3.2: Indication-related organs whose doses were estimated ............................................. 106 Table 3.3: Parameters used for performing the experimental study for scan coverage for routine CVA ..................................................................................................................................... 111 Table 3.4: CT equipment and scanning parameters .................................................................... 114 Table 3.5: The DAUT and DBLT used in the patient-based study ............................................ 117 Table 3.6: characteristics of CT phantom PBU-60 ..................................................................... 122 Table 3.7: Characteristics/specifications of CT scanners ........................................................... 123 XVIII University of Ghana http://ugspace.ug.edu.gh Chapter Four Table 4.1: Demographics of study respondents (Technical Heads) ........................................... 129 Table 4.2: CT models, manufacturers, years of manufacture, installation and number of detector rows/slices (N=31) ............................................................................................................... 130 Table 4.3: Geographical location, ownership status and functionality of the CT scanners (N=31) ............................................................................................................................................. 133 Table 4.4: Number of CT scanning radiographers, reporting radiologists and attending medical physicists ............................................................................................................................. 134 Table 4.5: QA infrastructure availability at the CT facilities (N=31) in Ghana ......................... 135 Table 4.6: Basic quality improvement (QI) structures in the CT facilities (N=31) in Ghana .... 138 Table 4.7: Policy infrastructure and their availability in Ghana ................................................. 139 Table 4.8: Duration to repair broken down equipment (down time) .......................................... 140 Table 4.9: CT examinations undertaken in a year ...................................................................... 142 Table 4.10: Identified common CT indications for adult head, chest and abdomino-pelvic regions ............................................................................................................................................. 144 Table 4.11: Basic diagnostic imaging requirements for CVA/stroke, head injury and brain tumour/SOL procedures in Ghana. ...................................................................................... 147 Table 4.12: Basic diagnostic imaging requirements for lung tumour, chest lesion with CKD and PE CT procedures in Ghana ................................................................................................ 149 Table 4.13: Basic diagnostic imaging requirements for abdomino-pelvic lesion, kidney stone and urothelial malignancy indication (CT-IVU) examinations in Ghana .................................. 151 Table 4.14: Comparison of measured and console displayed CTDIvol values for head phantom153 Table 4.15: Comparison of measured and displayed CTDIvol values for body phantom .......... 154 Table 4.16: CTDIvol dose delivery reproducibility for both head and body phantoms ............... 155 Table 4.17: Measured geometric efficiency values for CT scanners .......................................... 156 Table 4.18: Measured kVp accuracy values for CT scanners ..................................................... 157 XIX University of Ghana http://ugspace.ug.edu.gh Table 4.19: Measured HVL values for CT scanners ................................................................... 158 Table 4.20: Measured CT number for water and image noise in scanners ................................. 159 Table 4.21: Image uniformity (homogeneity) findings at different locations ............................ 161 Table 4.22: Demographic characteristics of patients’ data ......................................................... 162 Table 4.23: Descriptive statistics of scanning parameters used to acquire images .................... 163 Table 4.24: Descriptive statistics of scanning mode and contrast and AEC utilisation ............. 165 Table 4.25: Descriptive statistics of representative CTDIvol values for CVA CT ...................... 166 Table 4.26: Descriptive statistics of representative DLP values for CVA CT ........................... 167 Table 4.27: A comparison of CVA indication-based DRL values with international values. .... 167 Table 4.28: Descriptive statistics of representative CTDIvol values for head trauma/injury CT 169 Table 4.29: Descriptive statistics of representative DLP values for head trauma/injury CT ..... 170 Table 4.30: A comparison of indication-based DRL values for head trauma/injury CT, with international values. ............................................................................................................. 170 Table 4.31: Descriptive statistics of representative CTDIvol values for brain tumour/SOL CT indication ............................................................................................................................. 172 Table 4.32: Descriptive statistics of representative DLP values for brain tumour/SOL CT indication ............................................................................................................................. 173 Table 4.33: A comparison of indication-based DRL values for brain tumour/SOL CT, with international values. ............................................................................................................. 173 Table 4.34: Descriptive statistics of representative CTDIvol values for lung tumour/cancer CT indication ............................................................................................................................. 175 Table 4.35: Descriptive statistics of DLP values for lung tumour/cancer CT indication ........... 176 Table 4.36: A comparison of indication-based DRL values for lung tumour/cancer CT, with international values. ............................................................................................................. 176 Table 4.37: Descriptive statistics of representative CTDIvol values for CT of lung lesion with CKD indication ............................................................................................................................. 178 XX University of Ghana http://ugspace.ug.edu.gh Table 4.38: Descriptive statistics of DLP (mGy.cm) values for CT of lung lesion with CKD . 179 Table 4.39: Descriptive statistics of CTDIvol (mGy) values for CT of AP lesion indication ..... 180 Table 4.40: Descriptive statistics of representative DLP values for CT of AP lesion indication 181 Table 4.41: A comparison of indication-based DRL values for CT of AP lesion, with international values. .................................................................................................................................. 181 Table 4.42: Descriptive statistics of CTDIvol values for CT of kidney stone indication ............ 183 Table 4.43: Descriptive statistics of representative DLP values for CT of kidney stone indication ............................................................................................................................................. 184 Table 4.44: A comparison of indication-based DRL values for CT of kidney stone indication with international values. ............................................................................................................. 184 Table 4.45: Descriptive statistics of representative CTDIvol values for CT of urothelial malignancy indication ............................................................................................................................. 186 Table 4.46: Descriptive statistics of representative DLP values for CT of urothelial malignancy indication ............................................................................................................................. 187 Table 4.47: A comparison of indication-based DRL values for CT of urothelial malignancy indication with international values. .................................................................................... 187 Table 4.48: Descriptive statistics of representative CTDIvol values for CT of pulmonary embolism indication ............................................................................................................................. 189 Table 4.49: Descriptive statistics of representative DLP values for CT of pulmonary embolism indication ............................................................................................................................. 190 Table 4.50: A comparison of indication-based DRL values for CT CT of pulmonary embolism indication with international values. .................................................................................... 190 Table 4.51: Indication-specific effective doses and their equivalent background radiation levels ............................................................................................................................................. 194 Table 4.52: Various organ doses associated with the CT imaging for the CVA, head trauma/Injury and brain tumour/SOL indications ...................................................................................... 195 XXI University of Ghana http://ugspace.ug.edu.gh Table 4.53: Various organ doses associated with CT imaging for lung tumour/cancer, lung lesion with CKD and PE indications .............................................................................................. 197 Table 4.54: Various organ doses associated with CT imaging for AP lesion, kidney stones and urothelial malignancy indications ........................................................................................ 199 Table 4.55: Cancer and mortality risks associated with CT doses for lung tumour/cancer, lung lesion with CKD and PE indications. .................................................................................. 201 Table 4.56: Cancer and mortality risks associated with CT doses for AP lesion, kidney stones and urothelial malignancy indication ......................................................................................... 203 Table 4.57: Descriptive statistics of average extra scan length allowed above and below the target anatomic regions/areas. ....................................................................................................... 205 Table 4.58: Extra scan length (z-axis) evaluated, mean scores for subjective image quality analysis, and level of agreement between raters in head region examination; routine CVA ............. 207 Table 4.59: Demographics and clinical history of the patients ................................................... 209 Table 4.60: Measured SNR and subjective image quality scores for full range and reduced range CVA CTprocedures ............................................................................................................. 210 Table 4.61: Comparison of dose impact of full range and reduced range CVA CT procedures 211 Table 4.62: Diagnoses based on radiologists’ reports ................................................................ 213 Table 4.63: Comparison of dose output for abdomino-pelvic procedures undertaken with and without AEC systems. ......................................................................................................... 218 XXII University of Ghana http://ugspace.ug.edu.gh ABBREVIATIONS AAPM American Association of Physicists in Medicine ACS Automatic Current Selection AD Achievable Dose ADMIRE Advanced Model-Based Iterative Reconstruction AEC Automatic Exposure Control AIDR Adaptive Iterative Dose Reduction ALARA As Low as Reasonably Achievable AP Anterior-Posterior ASIR™ Adaptive Statistical Iterative Reconstruction BEIR Biological Effect of Ionising Radiation CAT Computed Axial Tomography CdWO4 Cadmium Tungstate CKD Chronic Kidney Disease CNR Contrast to Noise Ratio CoV Coefficient of Variation CT Angiography CT CT Computed Tomography CTDI CT Dose Index CTDIvol Volume Weighted CT Dose Index CTDIw Weighted CT Dose Index XXIII University of Ghana http://ugspace.ug.edu.gh CVA Cerebrovascular Accident DAUT Distance Below Lower Target DBUT Distance Above Upper Target DLP Dose Length Product DNA Deoxyribonucleic Acid DRLs Diagnostic Reference Levels EANM European Association of Nuclear Medicine EAR Excess Absolute Risk Deff Effective Diameter ED Effective Dose EFOM European Federation of Organisations for Medical Physics EFRS European Federation of Radiographer Societies EPA United State Environmental Protection Agency ERR Excess Relative Risk ESR European Society of Radiology ESTRO European Society for Radiotherapy and Oncology FDA United States Food and Drug Authority FOV Field of View Gd2O2S Gadolinium Oxysulphide GE Geometric Efficiency Gy Gray HU Hounsfield Unit HVL Half Value Layer XXIV University of Ghana http://ugspace.ug.edu.gh IAEA International Atomic Energy Agency ICC Intraclass Correlation Coefficients ICRP International Commission on Radiological Protection IMR Iterative Model Reconstruction IRIS Iterative Reconstruction in Image Space IVU Intravenous Urogram LAR Lifetime Attributable Risk LARi Lifetime Attributable Risk of Cancer Incidence LARm Lifetime Attributable Risk of Cancer Mortality LAT Lateral Patient LCL Lower Control Limits LDRLs Local Diagnostic Reference Levels LNT Linear No-Threshold Model MBIR Model-Based Iterative Reconstruction MDCT Multi-Detector-Row Technology MPD Multipurpose Detector MSAD Multiple Scan Average Dose NaI Sodium Iodide NCICT National Cancer Institute Dosimetry System For CT NCRP National Council on Radiation Protection and Measurements NDRLs National Diagnostic Reference Levels NRA Nuclear Regulatory Authority OSL Optically Stimulated Luminescence XXV University of Ghana http://ugspace.ug.edu.gh PACS Picture Archiving and Communication System PE Pulmonary Embolism PET-CT Positron Emission Tomography-CT PMMA Polymethyl Methacrylate QA Quality Assurance QC Quality Control QI Quality Improvement QMS Quality Management System RDSRs Radiation Dose Structured Reports RSN Reactive Nitrogen Species SD Standard Deviation SNR Signal to Noise Ratio SOL Space-Occupying Lesion SPECT-CT Photon Emission Computed Tomography-CT SPSS Statistical Package for The Social Sciences SSDE Size-Specific Dose Estimate TLD Thermoluminescence Dosimeters UCL Upper Control Limits UK United Kingdom UNSCEAR United Nations Scientific Committee on The Effect of Atomic Radiation USA United States of America VIF Variance Inflator Factor XXVI University of Ghana http://ugspace.ug.edu.gh WHO World Health Organisation YGdO Yttrium gadolinium oxide YoI Year of Installation YoM Year of Manufacture XXVII University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION 1.1 Background Computed tomography (CT) is a special radiological imaging modality which utilises x-ray photons to create cross-sectional images of the human body (Imai et al., 2018). It was invented in the early 1970s by a British engineer, Godfrey Hounsfield, and a South-African born physicist, Allan Cormack (Imaginis, 2017). CT imaging modality is considered an invaluable diagnostic tool, which is applicable for an extensive range of clinical work and research activities (Gao et al., 2017). Some unique benefits of CT include very fast scanning abilities, the production of images of good resolution, assistance in disease staging, as well as planning and treatment (National Cancer Institute, 2019). It is also relatively cheaper than other radiological imaging modalities like magnetic resonance imaging (MRI) (Gao et al., 2017). Studies (Imai et al., 2018; Gao et al., 2017; Pandharipande et al., 2016) indicate that CT impacts enormously and positively on patient management, and it has assisted in saving many lives. In spite of the enormous benefits of CT to patients, the radiation doses associated with it are high (Pyfferoen et al., 2017) and comparatively, far higher than those utilised in conventional x- ray examinations (Vawda et al., 2015). A publication (Storrs, 2013) argued that it would take about 150 to 1,100 conventional x-rays to obtain a similar radiation dose level in the range of a single CT scan. The International Commission on Radiological Protection (ICRP) and the World Health Organisation (WHO) have also stated that worldwide CT imaging constituted approximately 6% of all medical imaging examinations. However, out of the total dose of radiation acquired through imaging, CT examination-associated doses accounted for about 43% (ICRP, 2017; WHO, 2016). 1 University of Ghana http://ugspace.ug.edu.gh These dose statistics make CT one of the largest contributors in the medical environment (Yeh et al., 2016). The need to solve a plethora of clinical problems, as well as easy accessibility of the modality, has currently led to an upsurge in CT usage globally (WHO, 2018; Bellolio et al., 2017). Unfortunately, the relatively high radiation doses related with CT could cause genetic changes, and induce hereditary and cancerous effects in patients (Lim et al., 2018; Goodman, 2018; Liang et al., 2017; MacGregor et al., 2015). The cancer risk (probability) also increases linearly with dose (Goodman, 2018; Vassileva and Rehani, 2011; de Gonzalez et al., 2009). Some cancer predictive models used in the United States of America (USA) have even estimated that CT examinations and their associated radiations could generate up to about 2% of future neoplasms among residents, if the radiation levels are not optimised (Brenner and Hall, 2007). The ICRP has suggested the use of the principles of justification and optimisation to control and optimise medical exposures which include CT examinations (ICRP, 2007). The justification principle implies that the utilisation of radiation in medical imaging needs to be justified at the practice, procedure and patient levels. In particular, it requires that medical decisions involving the usage of any ionising radiation to diagnose patient conditions, should produce more benefits than harmful effects (Cannata et al., 2017). Dose limitation does not apply in medical exposures, and the optimisation procedure involves that a justified procedure, having weighed the good and the harmful effects, should be performed with the least possible dose (Cannata et al., 2017; Council of the European Union; 2014; ICRP, 2007). The optimised dose should be as low as reasonably achievable (ALARA) to ensure that the clinical objectives are met, while considering the current state of societal, technical and economic factors (Cannata et al., 2017; ICRP, 2007). One of the commonly recognised tools in the optimisation method is the development and implementation of 2 University of Ghana http://ugspace.ug.edu.gh diagnostic reference levels (DRLs), which was formally introduced in healthcare in 1996 by the ICRP (ICRP, 2017). The International Atomic Energy Agency (IAEA, 2014 p. 387) defines a DRL as: “A level used in medical imaging to indicate whether, in routine conditions, the dose to the patient or the amount of radiopharmaceuticals administered in a specified radiological procedure for medical imaging is unusually high or unusually low for that procedure” It is a numerically established optimal dose parameter that, in routine practice, practitioners should aim at not exceeding, following adherence to standards, good practices, and technical performances (Tsukamoto, 2017). Hence, it is employed as a trigger to detect radiological facilities, utilising unusually high doses (outliers) in a particular radiological task, for which optimisation measures are required (Martin and Vano, 2018; Kanal et al., 2017; Vassileva and Rehani, 2015; Vawda et al., 2015). In fact, the implementation and usage of DRLs in healthcare have been found to decrease the total scale of radiation doses observed in medical environments in many countries (Gao et al., 2017; Brink and Miller, 2015; MacGregor et al., 2015). Currently, the two CT dose descriptor values employed in establishing DRLs, in CT imaging are the volume weighted CT dose index (CTDIvol), and the dose length product (DLP) (ICRP, 2017; Tsukamoto, 2017; Bauhs et al., 2008). The CTDIvol characterises the average dose per slice, whereas the DLP represents the cumulative energy absorbed along the scan length (Vawda et al, 2015). A few researchers are working in the area of clinical indication-based DRLs, which is a new paradigm in DRL establishment (Jarvinen et al., 2015; Lajunen, 2015). This study seeks to provide a national reference data for DRLs, based upon the clinical indication for some frequently 3 University of Ghana http://ugspace.ug.edu.gh performed CT examinations, and also to extrapolate other relevant data which could help in dose optimisation in clinical practice in Ghana. 1.2 Statement of the Problem It is well documented that the high doses linked with CT procedures are a major global health concern in medicine (Liang et al., 2017; Gao et al., 2017; Kanal et al., 2017; WHO, 2016; Yeh et al., 2016; IAEA, 2013; ICRP, 2007). Radiation dose from CT imaging represents about 43% of the collective dose resulting from all radiological imaging examinations in medicine worldwide and could induce cancer in patients (WHO, 2016). According to a study (Pyfferoen et al., 2017), the radiation doses utilised to achieve comparable CT examinations of relatively similar diagnostic quality across facilities have to be within a relatively smaller range. However, national and multinational surveys continue to demonstrate very wide-ranging dose levels across facilities and countries (Liang et al., 2017; Lajunen, 2015; Vassileva and Rehani, 2015). In particular, dose parameter variations by way of a factor of 20 or more, have been found in anatomical regions across certain CT facilities (Liang et al., 2017). The ICRP has proposed DRL as one of the important means to optimising the high radiation dose levels in CT imaging (Vassileva and Rehani, 2015). Further, it has been recommended that each country ought to develop its own national diagnostic reference levels (NDRLs) because, among countries, the health care structure, socio-economic conditions, procedure guidelines and protocols, as well as the anatomical structures of their people, vary (ICRP, 2017). The WHO and the IAEA have also included in their Bonn Call for Action 2, the need for the establishment, usage of, and the frequent update of DRLs for radiological examinations (WHO, 2016; IAEA, 2013, 4 University of Ghana http://ugspace.ug.edu.gh IAEA GSR Part 3, 2014), while the European Society of Radiology (ESR) has recently included it in its EuroSafe Imaging Call for Action 2018 (European Society of Radiology, 2018). However, in Ghana there are no established NDRLs against which dose parameters could be compared with, to ensure effective dose monitoring and optimisation, despite the installation and utilisation of several CT scanners in the country. Although some studies (Anim-Sampong et al., 2016; Addo, 2016; Akyea-Larbi, 2015; Hasford et al., 2015; Inkoom et al., 2014; Muhogora et al., 2010, Muhogora et al., 2009) have provided some dosimetric data (usually a single or few centre-studies) on the anatomical parts of the body in some facilities in Ghana, no anatomical- based NDRLs were proposed from these studies. This is largely due to the lack of adequate data on which an NDRL could be built. Furthermore, the needed collaboration between the regulatory authority, relevant professional bodies and interested parties, to establish the relevant DRLs has not been exploited. Currently, reports (Ridley, 2016; Vock and Frija, 2016) have suggested that indication- based CT scanner protocols and DRLs, instead of anatomical references, decrease patient radiation dose considerably. The European Society of Radiology (2017), has also indicated that NDRLs based on the anatomical region do not reflect clinical practice, and should be re-examined using clinical indications. It is further argued that image quality and dose requirements vary among the different indications for an anatomical region (European Society of Radiology, 2017; Vock and Frija, 2016). Hence, the clinical indications dictate the main parameters (e.g. collimation, contrast usage, scan length) which contribute to patient radiation doses from CT (European Society of Radiology, 2017). Therefore, DRLs in CT should be indication-based (Vock and Frija, 2016; Jarvinen et al., 2015; Lajunen, 2015). Vock and Frija (2016) have again argued that, it is not logical 5 University of Ghana http://ugspace.ug.edu.gh to use, for instance, one abdomen DRL (anatomical) as a dose benchmark for diagnosis of liver metastases and kidney stones, as is presently the norm in many organisations. As a result, there have been recent calls for CT indication-based (indication specific) DRL development and utilisation, for optimisation purposes by several international bodies, including the European Association of Nuclear Medicine (EANM), European Society of Radiology (ESR), European Federation of Organisations for Medical Physics (EFOM), European Federation of Radiographer Societies (EFRS), and the European Society for Radiotherapy and Oncology (ESTRO) (European Association of Nuclear Medicine et al., 2017). In CT imaging, an indication is the medical condition for which a particular CT procedure is undertaken. As such, an indication-based DRL implies a DRL that is particular to a specific disease indication (Jarvinen et al., 2015). It is expected that the establishment and implementation of indication-based DRLs, has the potential to narrow down dose-monitoring and optimisation in CT examinations, and offer another level of accountability regarding the application of radiation in medicine. Also, an indication-based DRL offers the chance to utilise a DRL that adapts to a specific medical condition, and reduces the variability and ambiguities in using anatomical regions. For instance, a DRL for lung tumour could provide the chance to relate the dose parameters to CT chest procedures, with the suspicion of a lung tumour disease, and not just chest DRLs (Lajunen, 2015). Therefore, the absence of indication-based DRLs in Ghana prevents or limits the effort of effective dose monitoring and optimisation in certain clinical tasks, especially those which require diverse levels of exposure. This current study seeks to fill this knowledge gap by developing national indication-based DRLs, based on clinical data for dose optimisation in CT imaging in Ghana. 6 University of Ghana http://ugspace.ug.edu.gh 1.3 Scope of the Study This study was directed at developing indication-based DRLs in adult CT examinations undertaken in Ghana. It also identifies areas and steps for dose optimisations. The ICRP (ICRP, 1996) has suggested that DRLs should be developed for the common and prioritised diagnostic procedures which contribute significantly to the population dose. Hence, this study was limited to common indications relating to the head, chest, and abdomino-pelvic regions for adult patients as they account for the highest number of CT cases in Ghana (Inkoom et al, 2014). The dose descriptors of interest in the DRLs development were the CTDIvol and the DLP. Moreover, to achieve the aims and objectives of the study, CT technical data, scanner performance characteristics, common indications, image quality and diagnostic imaging requirements, were all considered. Furthermore, to quantify the dose impact associated with undertaking these indication- based CT procedures, and for optimisation steps to be recommended, the lifetime attributable risk (LAR) of cancer incidence and cancer mortality for each indication were estimated, while different steps for dose optimisation were also evaluated. 1.4 Main Objective The main objective of this study was to develop national indication-based DRL values for common and prioritised indications of the adult human body for clinical application in Ghana. It was also to assess the risk of undertaking each indication-based CT examination, and also propose some steps for dose optimisation. 7 University of Ghana http://ugspace.ug.edu.gh 1.5 Specific Objectives The specific objectives included: 1. To obtain technical data on CT scanners in the country and information on common indications for adult CT examinations, and as well define the basic diagnostic imaging requirements for each common indication in Ghana; 2. To obtain performance characteristic data on CT scanners in the country; 3. To evaluate CT dose descriptors/quantities and image qualities of CT images of the indications prioritised for DRL development; 4. To establish the national indication-based DRL values for common and prioritised indications; 5. To estimate the dose impact and risk of cancer associated with undergoing CT imaging for each of the indications in Ghana; 6. To propose measures for dose optimisation for the protection of patients in CT examinations in Ghana; and 8 University of Ghana http://ugspace.ug.edu.gh 1.6 Relevance and Justifications for the Study 1.6.1 New Research Area in Ghana CT application in healthcare in Ghana has increased dramatically in recent times (Inkoom et al, 2014). Although, CT scans have positively contributed immensely to patient care, it delivers a high radiation dose to patients compared to conventional x-rays in the diagnosis of diseases (Storrs, 2013; Pearce et al, 2012). Studies (Anim-Sampong et al, 2016; Inkoom et al, 2014), have established that there are appreciably high CT radiation dose levels in the few facilities they have studied in Ghana. As part of the optimisation process, there is a need to guarantee that the doses used in CT facilities are judiciously and adequately accounted for, so that the harmful effects connected with CT imaging are minimised. The development of national indication-based DRLs in this study would be the first study in Ghana to characterise and benchmark the dose needed for specific indications for CT investigations of body parts. This development and subsequent implementation of the indication-based DRLs is expected to decrease the range of doses observed in clinical practice (ICRP, 2017; European Society of Radiology, 2017; Lajunen, 2015; IAEA, 2013). 1.6.2 Policy Implications The use of ionising radiation in medicine requires some binding policies and regulations. Many international recommendations (ICRP, 2017; European Society of Radiology, 2017; WHO, 2016; European Union, 2014; IAEA, 2013; American College of Radiology; 2013) have suggested the development or adoption and implementation of DRLs, and recently, indication-based DRLs in respective countries to further regulate radiation usage in diagnostic radiology. Ghana, (a member of IAEA and WHO), is yet to accomplish this by the national regulatory authorities. It is 9 University of Ghana http://ugspace.ug.edu.gh believed that the lack of extensive research-based data to support the facilitation of this has been a drawback to the establishment of this policy. This study is timely, and its anticipated outcome should offer a data reference and foundation on which such policies and regulations can be built. In addition, health authorities in partnership with medical professional organisations must be responsible for establishing or adopting DRLs, and ensuring that as much as possible, the doses from medical imaging remain within the set levels (IAEA, 2019). The results of this study will go a long way to provide a reliable basis on which such organisations can work together to promote sustainable programmes to ensure optimisation of patient protection and safety. 1.6.3 Health Implications A report (Ridley, 2016) argued that indication-based DRLs from seven CT facilities in Switzerland have proven to significantly lower doses to patients. Therefore, the establishment and implementation of indication-based DRLs could also narrow down dose monitoring in CT examinations in Ghana, and offer another level of dose accountability in healthcare. Studies (Liang et al, 2017; MacGregor et al, 2015; Vassileva and Rehani, 2015; de González et al, 2009) have proven cancer risks associated with CT exposures. The DRLs proposed for Ghana could also be utilised to assess if a facility's dose index is remarkably high, and hence prompt corrective measures. This, in turn, would help to reduce adverse impacts such as radiation-induced cancer in patients. Moreover, there is currently no national indication-based DRL developed for Ghanaian healthcare which radiographers, radiologists, medical physicists, etc., may use as a country-based guide to regulate and optimise their practices in Ghana. It is envisaged that the outcomes of this study would offer a required tool to resolve these inadequacies. 10 University of Ghana http://ugspace.ug.edu.gh 1.6.4 Research Implications In Ghana, research activities on indication-based DRLs have just began with this study. It is anticipated that the outcome of this thesis would form the foundation on which future research studies could be based, and other DRLs developed for other pathological cases. This would not only serve Ghana, but would also serve as a model for other African countries that may like to develop a research focus in this area since there is no published work in the African sub-region on indication-based DRLs. 1.7 Organisation of Thesis This dissertation consists of five chapters. Chapter One deals with the introduction, which outlines and explains the background, problem statement, objectives, relevance and scope of the thesis. Chapter Two deals with the literature review relevant to existing information on the study area under investigation. Chapter Three deals with the materials and methods used for undertaking the study. Chapter Four presents the results and the relevant discussions using relevant literature. Chapter Five covers the conclusions and recommendations to recognised stakeholders. 11 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO LITERATURE REVIEW 2.0 Overview This Chapter presents information on the current knowledge and developments available in the literature on CT imaging, basic application of CT in medicine, CT dosimetry, typical radiation doses in CT, radiation risks, dose management, and optimisation in CT imaging and DRLs. Topics under DRLs include DRL in medicine, the role of DRLs in CT optimisation, anatomical DRLs in CT, indication-based DRLs, existing indication-based DRLs in CT, and theories guiding the DRL development in CT are discussed. 2.1 Computed Tomography Imaging CT, also referred to as computed axial tomography (CAT) scanner is a specialised radiological equipment which employs x-rays to generate cross-sectional radiological images and diagnostic information of the human body, and was invented in the early 1970s by Godfrey Hounsfield and Allan Cormack (Imaginis, 2017). A typical CT scanner is made up of an x-ray tube which revolves round a patient producing x-ray beams which transmit through a cross section of the body. It also features an array of detectors which receive the transmitted x-ray beam intensities, and are linked to a mathematical algorithm for image processing (Duncan and Panahipour, 2014). The x-ray tube is made up of a cathode from which an applied current heats up the filament, releasing electrons through thermionic emission (Radiologycafe, 2019). When a large potential difference is applied to the anode, the electrons are accelerated towards the anode, and collide with heavy atoms in the metal anode (Disher et al., 2006). A sudden slowdown of electrons as a result of the bombardment, 12 University of Ghana http://ugspace.ug.edu.gh causes the electrons to lose their energies, while part of it is converted into x-rays through the bremsstrahlung effect. Alternatively, an incoming electron could eject an inner shell electron, which causes an outer shell electron to move down to replace the ejected one, thereby creating an x-ray called a “K-shell” emission (Radiologycafe, 2019). The generated photons are used in CT imaging. A schematic diagram of an x-ray tube, the interactions leading to the emission of x-rays, as well as a typical CT imaging setup consisting of the x-ray beam spectrum and its relationship with the patient and detectors for CT imaging, is shown in Figure 2.1. Figure 2.1: Schematic diagram of an x-ray tube and production (Plate A), bremsstrahlung interactions (Plate B), “K-shell” emission (Plate C) and a typical CT imaging setup consisting of the x-ray beam spectrum and its relationship with the patient and detectors for CT imaging (Plate D) (Radiologycafe, 2019; Disher et al., 2006). 13 University of Ghana http://ugspace.ug.edu.gh In the CT setup, the basic principle of imaging involves the production of a photon attenuation map (image) of a body, as a result of the variable attenuation of an x-ray beam as it goes through the body of a patient, exposing a given volume of tissue from all directions (Bauhs et al., 2008). The philosophy underpinning CT imaging is grounded on the premise that the structural formation of an object obtained by a detector, can be reconstructed from multiple projections of the object, by utilising computer algorithms to create attenuation profiles of the variation of tissue radiodensity (Duncan and Panahipour, 2014). To create an image, several hundreds of attenuation coefficient spectra are formed in each rotation of the x-ray tube across the body of the patient under an examination (IAEA, 2012). These attenuation coefficients are then used to calculate the CT number values in Hounsfield units (HU) and subsequently, into a visible image characterised by diverse shades of grey (IAEA, 2012). Older scanners used back projection techniques to generate images (Padole et al., 2015). Filtered back projection has been used on modern scanners for a long time. Currently, iterative reconstruction is also used to reconstruct millions of data points when radiation transmits through a tissue in CT imaging (Ozdiev et al., 2018). Figure 2.2 presents a simple schematic diagram of CT imaging process. Figure 2.2: A simple schematic diagram of CT imaging process (Sprawls, 2019). 14 University of Ghana http://ugspace.ug.edu.gh CT scanners have undergone a number of design changes, and there have been seven generations since the discovery (Ali, 2015). The early generation CT scanners generated an image per one complete tube rotation in about 5 minutes (Ali, 2015). These machines were called single- slice CT scanners (Ali, 2015). In modern scanners, multiple slices (presently reaching 640) are produced instantaneously in a rotation within 0.5 second and are referred to as multidetector CT scanners (MDCT) or multislice CT scanners (MSCT) (IAEA, 2012). Early CT machines utilised scintillation detectors including sodium iodide (NaI) or cadmium tungstate (CdWO4) (IAEA, 2012). Subsequently, these early detectors were largely replaced by high-pressure xenon systems. Scintillator doped ceramics like gadolinium oxysulphide (Gd2O2S) or yttrium gadolinium oxide (YGdO) (IAEA, 2012) have been utilised in recent years. High dynamic range, superior quantum absorption efficiency, and a fast temporal response with low afterglow, are the essential specifications and factors in their development that make them very reliable detectors (IAEA, 2012). 2.2 Basic Application of Computed Tomography in Medicine CT modality is considered a very valuable diagnostic tool, which is applicable for an extensive range of clinical and research purposes (Gao et al., 2017). Faster scanning speed, isotropic spatial resolution, shorter computer reconstruction times, affordability compared to other imaging modalities like MRI, applications in staging, and the planning and treatment of cancer, are some of its exceptional benefits to mankind (de Basea et al., 2016; Abdulkadir et al., 2016; Yu et al., 2009; Tsukagoshi et al, 2007). It is presently one of the extensively utilised medical imaging machines in clinical applications, because it is often the most appropriate modality for 15 University of Ghana http://ugspace.ug.edu.gh certain types of screening, diagnosis, treatment and the management of patients’ diseases (Gao et al., 2017; Dauer and Hricak, 2014). In emergency departments for instance, CT results substantially influence diagnostic confidence, and admission decisions (Pandharipande et al, 2016). Studies (Imai et al., 2018; Furugori et al., 2018; Cheung et al., 2018) have shown that CT utilisation has significantly reduced mortality levels in hospitals. The development of MDCT technology with sub-second acquisition and CT fluoroscopy (enabling interventional radiological procedures to be undertaken) have advanced CT applications in medicine (European Society of Radiology, 2017). Furthermore, the development of hybrid radiological modalities such as single photon emission computed tomography-CT (SPECT-CT), positron emission tomography-CT (PET-CT), and CT simulation machine in radiotherapy treatment planning amongst others, is strengthening CT utilisation and applications in radiotherapy, nuclear medicine, and other treatment planning workups in healthcare (Sergieva et al, 2014). The positive impact of CT in medicine, has created an upsurge in its utilisation globally. In the USA for example over 70 million CT scans are undertaken yearly, and over 5 million in the United Kingdom (UK) with an increasing annual rate of 10% (de Basea et al., 2016). In Ghana, the first CT scanner was installed in 1994 at the Korle Bu Teaching Hospital. Since then, there has been an increased number of installations in the public, private and private- government (quasi) organisations (Schandorf and Tetteh, 1998). In 2012, the Government of Ghana undertook a large-scale old equipment replacement exercise and also planned to provide at least one CT scanner in each teaching or regional hospital (Ministry of Health, 2014). These initiatives, coupled with the development of private health facilities, have resulted in the installation of a greater number of CT scanners in the country, with the intention of improving 16 University of Ghana http://ugspace.ug.edu.gh patients’ access to good healthcare (Ministry of Health, 2014). Although there is no documented data on the extent of the impact of CT scanners in Ghana, they are well known to impact positively on patients’ care, as they are used to diagnose, plan, stage disease processes, and carry out interventional radiological procedures. 2.3 Computed Tomography Dosimetry The principal CT dosimetry formalism is grounded on the CT dose index (CTDI) and its variants, which are measured in gray (Gy) (McLean and Chapple, 2015). Although it does not explicitly provide patients’ individual doses, it characterises the average absorbed dose output per slice, in a standard geometry, from which a patient’s dose estimation can be made (McCollough et al., 2011). 2.3.1 Computed Tomography Dose Index (CTDI) The term CTDI is “the integral of the single scan radiation dose profile along the z-axis, normalized to the thickness of the imaged section (slice thickness)” (Bauhs et al., 2008 p. 246). During CT imaging, the radiation energy deposition in the body extends outside the collimated area, owing to scatter radiation effects, divergence of the beam, and the penumbral and restricted efficiency of the collimator (Boone et al., 2012). CTDI accounts for the scatter tails to obtain correct integration limits in a standardised and convenient manner. Mathematically, the CTDI is defined in equation 2.1 as: + 1 CTDI = D(z)dz (2.1) NT − where 17 University of Ghana http://ugspace.ug.edu.gh D(z) is a radiation dose profile (air kerma in a phantom) along the z-axis, z, centred at z = 0, N is the number of detector channels and T is the width of each channel (Bujila et al., 2018). The quantity NT denotes the nominal total collimation width of the scan. In MDCT scanners, many detector elements could be organised to make one data channel, hence N > 1 (American Association of Physicists in Medicine [AAPM], 2007). In single-detector-row CT, N=1, hence the z-axis collimation (T) at the isocentre is the nominal scan width (Huda, 2011). 2.3.2 CTDIFDA The dose contribution from the scanner along the z-axis and penumbral radiation dose profile should be accounted for in CTDI measurements. However, the exact integration limits were not defined in CTDI (Boone et al., 2012). To normalise CTDI measurements (as infinity is not a possible measurement quantity), the United States Food and Drug Authority (FDA) introduced the integration limits of ± 7T (CTDIFDA over 14 slices), where T signified the nominal slice width (United States Food and Drug Administration, 1984). Mathematically, the CTDIFDA is calculated from equation 2.2 as: +7T 1 CTDI = D(z)dz (2.2) FDA NT −7T where D(z), z, N and T maintain their usual meanings as defined in equation 2.1, and +7T and -7T are the integration limits of radiation dose profile/output. However, the limits of integration in CTDIFDA were not equally estimated in terms of NT, permitting for the possible underestimation of the multiple scan average dose (MSAD) by the CTDI, especially in thin slices or narrow slice widths (Bauhs et al., 2008; Dixon and Ballard, 2007). 18 University of Ghana http://ugspace.ug.edu.gh 2.3.3 CTDI100 In order to standardise the measurement length and limits as observed in the CTDIFDA, the CTDI100 was defined with the integration limits fixed at the standard length of a commercially accessible pencil ionisation chamber (100 mm) (AAPM, 2008). It was introduced to produce an integration of the radiation dose profile of the accumulated scan dose distribution over a 100-mm pencil ionisation chamber from a single axial scan (McCollough et al., 2011). Mathematically, the CTDI100 is expressed from equation 2.3 as: +50mm 1 CTDI100 = D(z)dz (2.3) NT −50mm where +50 mm and -50 mm are the integration limits of radiation dose profile/output 100-mm using a pencil ionisation chamber and D(z), z, N and T maintain their usual meanings as defined in equation 2.1. However, CTDI100 only characterises the total multiple scan dose at the middle of a 100- mm scan and could underestimate the total dose for scan lengths greater that 100 mm (McCollough et al, 2011). 2.3.4 Weighted CT Dose Index (CTDIw) During CT imaging, radiation dose reaching the peripherals and the centre of a phantom under investigation are not exactly the same. To account for dose levels across varying fields of view, and not just at the centre, and avoid the limitations of other variants of CTDI, the weighted CT dose index (CTDIw) was developed. The CTDIw represents the weighted average radiation dose 19 University of Ghana http://ugspace.ug.edu.gh assessed at the centre (CTDI100c) and peripherals (CTDI100p) of a standard phantom, polymethyl methacrylate (PMMA), across a field of view (Baughs et al., 2008). It is expressed mathematically using equation 2.4 as: 1 2 CTDI = (CTDI (w 100,centre )+ CTDI )100, periphery (2.4) 3 3 where the symbols bear their defined meanings. However, CTDIw does not describe doses for helical scans. Moreover, it has a limitation in estimations of doses for non-adjacent slices frequently seen in helical scanning where the x-ray beams may overlap (common with MDCT), or where extended pitch leaves gaps between rotations. 2.3.5 Volume Weighted CT Dose Index (CTDIvol) Helical scan technology utilises continuous gantry-rotation and incremental table feed to provide for a longer anatomic coverage in a shorter time (Fukuda et al., 2014). In estimating the dose in a helical scan, it is crucial to consider gaps or overlays between the radiation beams from successive revolutions of the x-ray source, when helical scans are used (Bujila et al., 2018). The volume weighted CTDI (CTDIvol) embodies the average dose output in the central region of a multiple scan examination, and considers scan spacing and helical pitch (Huda, 2011; AAPM, 2008). Mathematically, it is written using equation 2.5 as: 1 CTDIvol = CTDI w (2.5) pitch where CTDIvol is volume weighted CTDI and CTDIw is weighted CTDI. 20 University of Ghana http://ugspace.ug.edu.gh CTDIvol is measured using a100-mm long ionisation chamber or linear array of point dose measurements utilising thermoluminescence dosimeters (TLDs), film, metal-oxide-semiconductor field-effect transistors (MOSFET), optically stimulated luminescence (OSL), or solid state (diode) detectors. The RTI Electronics Group (Mölndal, Sweden) in particular, makes a solid-state detector probe known as the CT dose profiler that has proven reliability for this application (RTI Group, 2016; Bauhs et al., 2008). 2.3.6 Dose Length Product To be able to easily calculate the absorbed radiation dose along the entire scan length of a CT examination, the dose length product (DLP), measured in milligray-centimetres (mGy.cm) was introduced in CT dosimetry (Fukuda et al., 2014; Boone et al., 2012; Huda, 2011; AAPM, 2008). The concept of DLP integrates both the mean CTDIvol and the total distance (Ls) scanned (IAEA, 2013). Since the DLP is not a satisfactory risk indicator, it requires a conversion factor to express in effective dose. Currently, it is mandatory that CT scanners display both CTDIvol and DLP on the operator’s console (ICRP, 2017; International Electrotechnical Commission, 2011). Mathematically, DLP is expressed in equation 2.6 as: DLP = CTDI . volLs (2.6) where Ls is the scan length and CTDIvol is the volume weighted computed tomography dose index. 21 University of Ghana http://ugspace.ug.edu.gh 2.3.7 Effective Dose Effective dose (ED) is “the tissue-weighted sum of the equivalent doses in all specified tissues and organs of the human body and represents the stochastic health risk to the whole body” (ICRP, 2007 p.23). It is a single radiation dose descriptor that expresses the risk of a non-uniform radiation exposure in respect of an equivalent whole-body exposure (Mullenders et al., 2009). ED has a unit of Sievert (Sv) and one Sv indicates a 5.5% risk of cancer development (ICRP, 2007). Generally, the ICRP (2007) expresses ED by the equation 2.7: 𝐸𝐷 = ∑ 𝑤𝑇 ∑ 𝑤𝑅 𝐷𝑇,𝑅 𝑜𝑟 𝐸𝐷 = ∑ [∑ 𝑤𝑇𝐻𝑇] (2.7) 𝑇 𝑅 𝑧 𝑇 where wT = the ICRP tissue-specified weighting factor for tissue or organ T under examination. wR = radiation weighting factor DT, R = absorbed dose in tissue T by radiation type R HT or wR DT, R = the corresponding equivalent dose to the specified tissue or organ, T (ICRP, 2007). However, in CT imaging, this dose quantity could be estimated via multiplying the DLP with derived conversion factors or appropriately normalised coefficients that are specific to age, scanned region and CT geometry. This is expressed by equation 2.8 as. ED = kxDLP = kLCTDI vol (2.8) where k is the conversion factor for changing DLP to effective dose (ED) and unit is mSv/mGy.cm. DLP is dose length product, CTDIvol is volume weighted CT dose index and L is scan length. 22 University of Ghana http://ugspace.ug.edu.gh 2.3.8 Size-Specific Dose Estimate (SSDE) Measured in milligray (mGy), the size-specific dose estimate (SSDE), is a newer CT- derived descriptor that integrates patient-size-adjusted correction factor (f) based on patient’s effective diameter (deff) to efficiently predict patient dose (AAPM, 2014). The effective diameter is defined as “the square root of the product of the anterior-posterior (AP) and lateral patient diameter (LAT)” (AAPM, 2014, p.4). With knowledge of the deff, the f factor could be derived from the AAPM Report 204 to estimate SSDE (Boone et al., 2011). A diagram of AP and LAT diameter is shown in Figure 2.3. Figure 2.3: AP and LAT diameter measurements Mathematically, deff and SSDE (from 16 or 32 phantom size) are expressed in equations 2.9 and 2.10, respectively as: deff = (APxLAT ) (2.9) SSDE = CTDI 16or32 16or32vol xf (2.10) where f factor is patient-size-adjusted correction factor, AP is antero-posterior, LAT is lateral patient diameter, deff is effective diameter and CTDIvol is volume weighted CT dose index. 23 University of Ghana http://ugspace.ug.edu.gh 2.4 Typical Radiation Doses in Computed Tomography According to the WHO, the global CT usage represented about 6% of all medical imaging procedures performed, but it accounted for about 43% of the total dose resulting from those procedures (WHO, 2016). Recent publications (Smith-Bindman et al., 2019; Canadian Nuclear Safety Commission, 2019; Pyfferoen et al., 2017; Liang et al., 2017) continue to suggest CT as the major contributor to patient radiation exposure in medicine. Particularly, national and multinational surveys continue to indicate wide-ranging CT radiation dose levels across facilities and countries (Smith-Bindman et al., 2019). Reviewed international publications, with emphasis on commonly performed procedures, indicate typical paediatric CT effective doses in the range of head (0.8-2.0 mSv), chest (1.1-5.1 mSv), abdomino/pelvic (0.6-23.5 mSv) and spine (0.6-10 mSv) (Strauss et al., 2019; Ekpo et al., 2019; Saravanakumar et al., 2017; Kanal et al., 2017; Public Health England, 2016; European Union, 2014; Miglioretti et al., 2013; Brady et al., 2012; Muhogora et al., 2010). Adult CT effective doses also fall in the range of head (1.5- 5.8 mSv), abdomen (8.0- 39 mSv), chest (2.7-24 mSv), thorax high resolution (13.4-25 mSv), CT pulmonary angiogram (2.7-30 mSv), abdomen low-dose for kidney stones (3.3-11 mSv), CT urography (2.5-45 mSv) and thorax-abdomen (11.5-40 mSv) (Canadian Nuclear Safety Commission, 2019; IAEA, 2019; Ekpo et al., 2018; United States Food and Drug Administration, 2017; Public Health England, 2016; van der Molen et al., 2013; Korir, 2012; Muhogora et al., 2009; Smith-Bindman et al., 2009; ICRP, 2000). A recent study (Smith- Bindman et al., 2019) which evaluated the variation in adult radiation dose for CT procedures among countries also found a large variation of about a 4-fold range (7.0-25.7 mSv) in the mean effective dose for adult abdominal CT procedures.. The mean effective dose for chest CT examination also ranged from 1.7-6.4 mSv, while that of combined chest and abdomen CT 24 University of Ghana http://ugspace.ug.edu.gh examinations varied from 10.0-37.9 mSv. Head and pulmonary embolism CT effective doses also varied from 1.4-1.9 mSv, 8-27% and 5. 0- 33.2 mSv, respectively. In Ghana, there is a paucity of studies on national radiation dose levels in CT. However, available studies, mainly among few facilities, provide pertinent information on the dose levels across some facilities in the country. In particular, studies (Sackey et al., 2018; Akyea-Larbi et al., 2016; Anim-sampong et al., 2015; Hasford et al., 2015; Inkoom et al., 2014) reported adult CT effective doses in the range of head (1.1-5.2 mSv), chest (2.7 - 13.5 mSv), abdomen (5.3-24.0 mSv), neck (2.58-4.5 mSv), pelvis (4.82-9.1 mSv), and lumbar spine (9.2-11.2 mSv). Another study (Shirazu et al., 2017a) has quite recently reported mean renal doses of 10.36 mSv for females and 11.58 mSv for males in Ghana. With respect to paediatrics, a couple of studies (Addo, 2016; Gedel and Gablah, 2014) have reported CT effective doses in the range of head (0.9-3.2 mSv), chest (1.14-13.3 mSv) and abdomen (1.42-17.7 mSv). These evidences indicate that the radiation doses associated with CT imaging are high (more than 150 times) compared with conventional x- rays (Pyfferoen et al., 2017; Storrs, 2013). Therefore, there is the need for dose management and optimisation methods to reduce the potential health risks associated with CT radiation in Ghana and beyond (Pyfferoen et al., 2017; Storrs, 2013; Pearce et al., 2012). 25 University of Ghana http://ugspace.ug.edu.gh 2.5 Radiation Risks Ionising radiation could cause detrimental effects to living tissues and the mechanisms leading to these effects are complex. Basically, during exposure of living tissues to ionising radiation, energy is deposited in the tissues which could disrupt the organic molecules causing direct or indirect damage to the deoxyribonucleic acid (DNA) (Desouky et al., 2015). A DNA that is directly struck by ionising radiation could suffer direct damage, disrupting the genetic information. Indirect damage also arises when radiation interacts near the genetic material and releases electrons which react with a water molecule, forming a strong free radical (Pearce et al., 2012). The radiation interactions can create several highly reactive radicals that could react with DNA molecules to cause molecular structural injuries (Alkhorayef et al., 2017; Desouky et al., 2015; Mazrani et al., 2007). Due to the small volume of the DNA in relation to that of the cell, the probability of the occurrence of direct radiation damage of DNA is small (Goodman, 2018). Most (2 in 3) x-ray radiation induced damages are caused by indirect processes, as a cell is made up of nearly 70% water (Pearce et al., 2012). Apart from the effect of water radiolysis products, some evidence (Wardman, 2009) suggests that radiation activated reactive nitrogen species (RNS,) among others, could also cause molecular damage in the DNA. Damage to a DNA, through direct or indirect means, may be repaired correctly, or have several consequences. These include cell apoptosis or genetic mutations and alterations, which could induce health abnormalities and carcinogenesis (manifesting themselves over several years, even decades later) (Shah et al., 2012). The risk is much higher among children than in adults, because of the radiosensitivity of their developing tissues as a result of their rapidly dividing cells, high metabolic rate and developing cells (Goodman, 2018; Minami and Kudo, 2015). Figure 2.4 illustrates the direct and indirect impact of radiation dose on the DNA. 26 University of Ghana http://ugspace.ug.edu.gh Figure 2.4: Schematic diagram of radiation impact on DNA, where plates A and B show direct and indirect impacts of radiation, respectively (Teach Nuclear, 2019). The effect of radiation on humans has been classed into deterministic and stochastic effects. Deterministic effects are those that ensue when a threshold of dose has been exceeded, and the severity of the effect increases with radiation dose (Shah et al., 2012; Hall and Brenner, 2008). Some deterministic effects and their radiation threshold levels include erythema (2 Gy), epilation (3 Gy), cataract (2-10 Gy), sterility (2.5-3.5 Gy) and death (> 10 Gy to whole body) (Goodman, 2018). Any radiation effects (primarily cancer and genetic effects) which occur by chance, without exceeding a radiation threshold level are referred to as stochastic effects. The probability of its incidence increases with dose (Alkhorayef et al., 2017; Desouky et al., 2015). At low dose levels (below 100 mSv), including CT doses, stochastic effects could occur, though, there are considerable uncertainties in estimating the risk of developing a radiation-induced cancer (National Research Council, 2006). Statistical models indicate that higher radiation levels (as used in CT) may cause more stochastic effects than low levels (as in conventional x-rays) (National 27 University of Ghana http://ugspace.ug.edu.gh Research Council, 2006). Radiation-induced effect responses for both deterministic (a) and stochastic effects (b), are presented in Figure 2.5. Figure 2.5: Diagram of deterministic (a) and stochastic (b) effect-dose response curves (Rahman, 2018). Models used to predict radiation risks have been established by organisations such as the National Council on Radiation Protection and Measurements (NCRP, 1993), the United Nations Scientific Committee on the Effect of Atomic Radiation (UNSCEAR, 2000), the United States Environmental Protection Agency (1999), the Biological Effect of Ionising Radiation (BEIR VII) Committee (National Research Council, 2006), and the ICRP (ICRP, 2007). In literature, the BEIR VII Report of 2006 (phase 2) is considered the most widely used model for estimating risk of cancer, based on the magnitude of a single radiation exposure, and a patient’s age and sex at exposure (National Research Council, 2006). The model was developed based on the extensive studies on the survivors of Hiroshima and Nagasaki atomic bombs explosion, and is centred on the 28 University of Ghana http://ugspace.ug.edu.gh linear no-threshold model (LNT) concept, with the assumption that the smallest dose has the potential to cause a small increase in radiation risk to humans (National Research Council, 2006). Specific models used to express cancer risks include the excess relative risk (ERR), which is the proportional increase in risk over the background absolute risk (in the absence of exposure), and the excess absolute risk (EAR) which is the additional risk above the background absolute risk (Lim et al., 2018). The lifetime attributable risk (LAR) uses different final risk models such as EAR and ERR to predict risks for different organs (Lim et al., 2018). The LAR is defined as the probability of radiation-induced cancers or related mortality in a population of 100,000 who have been exposed to 100 mGy (Halid et al., 2018). The LAR at the age of “a” for a gender “g” given an exposure of “D” at an exposure age “e” can be expressed as follows (equation 2.11). a max Saj (a, g) LAR(D,e, g) = M (D,e,a, g) da (2.11) S e+L aj (e, g) where L = minimal latent period (5 year for solid cancers, 2 year for leukaemia), Saj(a, g) = a survival rate for gender “g” aged “a” unless the person suffers exposure to radiation, Saj(a,g)/Saj(e,g) = a conditional probability that a person survives from the exposure age “e” till an attained age, M(a,g) = baseline risk, ω = risk transport weight factor obtained from the BEIR VII report (Lim et al., 2018). In EAR model: M (D,e,a, g) = EAR(D,e,a, g) (2.12) ERR model: M (D,e,a, g) = ERR(D,e,a, g) (2.13) Combined model: M (D,e,a, g) =EAR(D,e,a, g )+ (1−)ERR(D,e,a, g)m(a, g) (2.14) 29 University of Ghana http://ugspace.ug.edu.gh For simplicity, the BEIR VII report has generated lifetime attributable risk of cancer incidence and mortality coefficients. They have been aggregated with cancer rates in 100,000 lives for various organs based on a single dose of 0.1 Gy at several specific ages, from which many explorations on radiation risks could be estimated (National Research Council, 2006). Tables 2.1 and 2.2 show the lifetime attributable risk of cancer and mortality coefficients, as aggregated by the BEIR model. Table 2.1: Lifetime attributable risk of cancer incidence coefficients from Table 12D-1 of BEIR Report (National Research Council, 2006) 30 University of Ghana http://ugspace.ug.edu.gh Table 2.2: Lifetime attributable risk of cancer mortality coefficients from Table 12D-2 of BEIR report (National Research Council, 2006) A number of publications have estimated the risk of cancer using the LAR model (Lim et al, 2018; Pai et al., 2018). In a recently published work, a group of scientists (Lim et al, 2018) found that CT examinations could induce LAR of cancer in 10 (bladder), 36 (colon), 48 (liver), 59 (lung), 81 (stomach), 32 (thyroid), 66 (breast) and 6 (leukaemia) per 100,000 people exposed to CT imaging. The associated LAR of radiation-induced mortality were 3 (bladder), 9 (colon), 33 (liver), 46 (lung), 23 (stomach), 1 (thyroid), 8 (breast) and 3 (leukaemia) per 100,000 population. These values present some worrying situations. Indeed, the ICRP models have long predicted the excessive relative risk of cancer mortality to be 5%/Sv. Therefore, there is a collective need for effective dose management and optimisation of radiation levels in medical exposures (ICRP, 2007). 31 University of Ghana http://ugspace.ug.edu.gh 2.6 Dose Management and Optimisation in Computed Tomography Imaging Due to the potential risks of radiation dose levels in medical exposures, there is the need for effective dose management and optimisation (IAEA Safety Standards Series, 2018; ICRP, 2017). The main mechanism and technical factors that play crucial roles in the management and optimisation of patients’ doses in CT are discussed in this section. 2.6.1 Quality Management Systems In CT imaging, quality management systems (QMS) are essential to ensure effective radiation protection and well-being of patients (Delis et al., 2017). QMS refers to the organisational structures, procedures and resources needed to implement quality management in an organisation (American Society for Quality, 2019). It includes quality assurance (QA) and quality control (QC), as well as quality improvement (QI) systems (Kruskal et al., 2009). The QA component is defined as an interdisciplinary management tool, including policies which offers a means for guaranteeing that all activities, including CT imaging, are effectively planned, correctly performed and assessed, while QC is a means of applying controls to the process to ensure that the product or service (CT imaging and care) consistently meets specifications such as providing optimal diagnostic information while delivering low doses (IAEA, 2006). QI, on the other hand, is a systematic, recognised approach to the assessment of practice performance and efforts to advance performance or a means of progressing from a particular level of performance, to a higher level of quality performance (Delis et al., 2017). A well implemented QMS, which encompasses appropriate policies, protocols, training and activities, enables the organisations to plan effectively, do the needed work, check that the system is functioning properly as they have been planned, and act by addressing any issues found in the check-up process that would ultimately lead to improved 32 University of Ghana http://ugspace.ug.edu.gh performance. It also involves collaboration between multiple stakeholders from referring physicians, radiologists, radiographers, medical physicists, and patients, to ensure effective patient care and optimal image quality for diagnosis, with the aim of preventing unnecessary radiation to people (Vano et al., 2014; Manghani, 2011). 2.6.2 Justification In the absence of dose limitation in medical exposures, the ICRP has long recommended the use of the principle of justification and optimisations to further manage and optimise radiation doses in healthcare (ICRP, 2007). The justification principle implies that the exposure of patients to radiation in medicine needs to be justified at the practice, procedure, and patient level (ICRP, 2007). In particular, it requires that medical exposures should be justified, and should produce more benefits than harmful effects (Cannata et al., 2017). A collaboration between the referring clinician and the radiological practitioners (radiologists and radiographers), is needed to justify procedures, and where appropriate, a non-ionising modality should be considered (IAEA Safety Standards Series, 2018). Referrer guidelines and clinical decision software solutions are known tools in the justification process, and a study (Vom and Williams, 2017) has indicated that, 58% of unjustified radiographic examinations could be prevented by using referral guidelines. 2.6.3 Optimisation Optimisation is the process of ensuring the radiation dose associated with imaging procedures are optimised as low as reasonably achievable (ALARA), while the clinical objectives are met with a consideration of societal, technical and economic factors (Cannata et al., 2017; ICRP, 2007). In CT imaging, this is usually applied at the design of equipment and facility level, 33 University of Ghana http://ugspace.ug.edu.gh as well as the day to day radiological practice level (ICRP, 2007). The design, construction of equipment, and installation of facilities play a crucial role in avoiding radiation leakages, scatter, and excessive exposure to patients and staff (Amaoui et al., 2020). The optimisation associated with the radiological practice requires that all the necessary tools needed for optimisation action in any justified radiological procedure, should be applied for radiation protection and safety (Amaoui et al., 2020). Various methods, with the majority involving the adjustment of exposure and technical factors, have been reported for radiation dose optimisation in CT imaging (Trattner et al., 2014); and they include the points indicated in Sections 2.6.3.1 to 2.6.3.10, and 2.7. 2.6.3.1 Tube Current and Tube Loading (mAs) There is a direct proportional relationship between radiation dose and tube load (the product of tube current and the exposure time per rotation), which determines primarily the intensity or the number of x-ray photons (photon fluence) (Raman et al., 2013) incident on a given area of the patient. Where all parameters remain constant, a 50% reduction in tube load will produce halve the dose (Aweda et al., 2007). However, since the image noise is inversely 1 proportional to the square root of the mAs (Noise ∝ = ), the noise will increase by a factor √mAs of √2 , (40% increase), if mAs is reduced by half of its original value (Guberina et al., 2016). This is because a reduction in mAs would result in fewer photons forming the image (Lira, et al., 2015). Therefore, it is important that image quality acceptability levels be considered when utilising mAs in optimisation (Guberina et al., 2016). 34 University of Ghana http://ugspace.ug.edu.gh 2.6.3.2 Tube Voltage (kVp) Tube potential measured in voltage or kilovoltage (kV) is defined “as the electrical “potential” difference between anode and cathode of the x-ray tube” (Lira et al., 2015 p.1). Increasing the tube voltage will increase the primary energy, and to some extent, the intensity of photons energy emitted, and dose (Raman et al., 2013). A recent study (Rusandu et al., 2018) has indicated that reducing the kilovoltage peak (kVp) factor from 100 to 80 during a pulmonary CT angiography could produce a desirable image quality, with approximately, 50% dose reduction. Another study (Yu et al., 2009) has also reported a dose reduction of 23% with an improved contrast and visualisation of mural stratification, when kVp was reduced from 120 to 100. However, the noise change is approximately inversely proportional to the change in voltage (Kalra et al., 2004). Therefore, there is always a need for a trade-off between kVp and image quality, to keep the patient’s dose optimised. 2.6.3.3 Pitch In modern CT scanners, pitch is referred to as the table distance travelled (as result of the incremental table movement) in one gantry scan revolution (360°) divided by combined thickness of all concurrently acquired slices (Guberina et al., 2016). Using a pitch of greater than (>) 1 or increasing table speed and table increment per gantry rotation (pitch factors) decreases scan duration and reduces the impacted radiation dose, although image quality can be compromised (Stradiott et al., 2009). Conversely, a pitch of less than (<) 1 could produce a better image quality with an associated higher patient dose, therefore, an appropriate trade-off would be needed to ensure optimisation (Stradiott et al., 2009). 35 University of Ghana http://ugspace.ug.edu.gh 2.6.3.4 Scan Thickness For a given anatomical scan coverage, decreasing the slice thickness also increases scan images, the spatial resolution along the z-axis and the dose within the given body (Raman et al., 2013; European Commission, 1999). The number of photons per a voxel decreases with decreasing thickness (Raman et al., 2013). This leads to increased levels of image noise, which will require an increased radiation dose to reduce the image noise levels (Raman et al., 2013). A prudent selection of slice thickness for a particular clinical task could, therefore, be used to optimise radiation doses in CT (Guberina et al., 2016). 2.6.3.5 X-ray Beam-Shaping Filter In CT imaging, x-ray beam-shaping filters are used to attenuate low energy x-rays emitted by CT x-ray tubes to allow hard enough energy to generate the needed diagnostic information (American College of Radiology, 2010). A bowtie filter compensates for this by attenuating the peripheral edges of the beam more than the centre to improve image quality and lower radiation dose, particularly to the skin and peripheral organs (Yu et al., 2009). 2.6.3.6 Scan Length and Z-Axis Overscan CT dose output (DLP) depends on the field coverage along the z-axis. Reducing the scan range to cover just the area of interest would reduce unnecessary radiation (Rahman et al., 2018). A study (Zinsse et al., 2019) demonstrated a dose reduction of 39.2% (without compromise on quality) after optimising the scan length for patients with suspected acute appendicitis. In helical imaging, overscanning (z-axis overscan) circumstances are possible where additional layers of tissues at the start and end positions (along the z-axis) are exposed. This is needed for interpolation 36 University of Ghana http://ugspace.ug.edu.gh of data from algorithms. Unfortunately, this could increase patients’ effective dose by 13.1% (in head and neck), 35.8% (in chest), and 29.0% (in abdomen) (Tzedakis et al., 2005). The use of dynamic z-axis collimation (collimator shutter action) decreases overscanning. 2.6.3.7 Detector Configuration Radiation dose could also be optimised by the detector configuration (Guberina et al., 2016). Due to the wider x-ray beam and penumbra effect (overbeaming), unnecessary radiation is common in detectors with smaller beam collimation and vice versa (Raman et al., 2013). Moreover, detectors with high quantum detection efficiency and geometric efficiency are known to produce optimised dose levels (Yu et al., 2009). 2.6.3.8 Automatic Exposure Control/Tube Current Modulation Contrary to a fixed mAs system which applies the same tube load across a body of varying thickness, the automatic exposure control (AEC) system modulates the tube current based on the different attenuation profiles of the various part of the object under consideration to achieve a target image quality through the scan (Merzan et al., 2017). By adjusting the mAs, doses vary according to the varying thicknesses of the object, unlike the fixed mAs system, where smaller and bigger volumes of a body receive similar doses. This optimisation process is reported to have reduced radiation to patients by about 20–40%, with an opportunity to further reduce more while ensuring good image quality (Higaki et al., 2019; Merzan et al., 2017). Modern CT scanners utilise three (AEC) techniques, namely the angular, longitudinal (z-axis) and combined modulations (Higaki et al., 2019). Table 2.3 presents the principles and modulation planes supporting each technique. 37 University of Ghana http://ugspace.ug.edu.gh Table 2.3: AEC techniques available in modern CT systems AEC Angular Longitudinal (z-axis) Combined modulation The tube current is The tube current is adjusted along the The tube Technique adjusted during each scanning direction of the patient, according to current is gantry rotation, the size and attenuation of the anatomic region adjusted both according to the size, being scanned and the predetermined image during each shape and attenuation quality gantry of body region being rotation and scanned. for each slice position. Modulation plane x, y z z, y, z Both Reference: (Higaki et al., 2019). Different principles of AEC are utilised by each manufacturer. In each, there is an image quality (IQ) reference parameter [such as noise index, standard deviation (SD), reference mAs, reference image etc.] by which the tube current is modulated to ensure its consistency across the body part being scanned (Higaki et al., 2019; Merzan et al., 2017). Moreover, some manufacturers (Siemens and GE Healthcare) have recently introduced organ dose modulation to decrease the radiation dose to radiosensitive anterior organs by effectively lowering the tube current once the x‐ray tube is situated in the anterior area of the human body (Dixon et al., 2016). A study (Fillon et al., 2018) found that organ dose modulation in combination with standard AEC could further reduce breast dose by 23% with negligible effect on image quality. Table 2.4 presents the major CT manufacturers and their reference parameter as well as their AEC systems. 38 University of Ghana http://ugspace.ug.edu.gh Table 2.4: The AEC systems from major manufacturers Siemens Philips GE Toshiba DoseRight: Consist of Automatic AEC type CareDose4D: A combined A combine AEC (AutomA 3D) SureExposure Current Selection (ACS) that and modulation together with which is made up of AutomA (z- 3D: A combined provides patient-based AEC, D- technique real time, online, controlled axis modulation) and SmartmA AEC DOM that provides angular AEC tube current modulation. (angular modulation). and Z-DOM that provides ΔX‐CARE similar as longitudinal AEC. All do not Δ Organ Dose CareDose4D but for work together except ACS which Modulation (ODM) similar as modulation doses to anterior combine with the other two. SmartmA but for modulation organs doses to anterior organs Noise index corresponding to a It provides desired IQ Quality reference mAs A reference image of a suitable required image quality is set. SD (or image reference (effective mAs) for a patient examination is stored. Based on patients’ quality level) of parameter reference patient (70-80kg) AEC modulates specific mAs characteristics, AEC modulates pixel values in is used as a reference, and values for each patient to patient-equivalent the tube current to preserve the subsequently modulates the achieve a constant image noise water phantom. In same level of noise in each mAs when different patient level as that of the reference a given body, the image. size is scanned. image. tube current is modulated to Δ ΔX‐CARE uses Quality ODM uses Noise index but achieve the stated reference mAs but reduces reduces mAs when beam is value throughout mAs when beam is facing facing anterior side at an arc of the scan. anterior side at an arc of 180° arc for body protocols and 1200 a 90° for head protocols Δ: are new AEC systems commonly referred to as organ dose modulation AEC; AEC: automatic exposure control, SD: standard deviation. Reference: (Higaki et al., 2019; Merzan et al., 2017). 39 University of Ghana http://ugspace.ug.edu.gh 2.6.3.9 Reconstruction Algorithm Older generations of CT scanners used filtered back projection to reconstruct CT images. However, because of its algorithm, images reconstructed at reduced doses present image noise and artifacts (Kim et al., 2014). Improvement in reconstruction algorithms such as iterative reconstruction (IR) techniques iterate CT image reconstruction several times, using mathematical algorithms to create images with lower noise (Padole et al., 2015). Different algorithmic approaches are used by CT manufacturers; Siemens Healthcare [Iterative Reconstruction in Image Space (IRIS), Sinogram-Affirmed Iterative Reconstruction (SAFIRE), and Advanced Model- Based Iterative Reconstruction (ADMIRE)], Toshiba America Medical Systems [Adaptive Iterative Dose Reduction (AIDR)], GE Healthcare [Adaptive Statistical Iterative Reconstruction (ASIR™) and Model-Based Iterative Reconstruction [MBIR or Veo™], Philips Healthcare [iDose™, iDose4™ and Iterative Model Reconstruction (IMR)] and Medic Vision [SafeCT] (Padole et al., 2015; Kim et al., 2014). Some authors (Greffier et al., 2016) have reported that the combination of AEC and IR with adequate optimisations could reduce radiation levels by 43-91% with good image quality. 2.6.3.10 Patient Positioning Bowtie filters and AEC systems in CT operate with the assumptions that the patient is isocentered during scanning (Kaasalainen et al., 2014). Off-centering of a patient away from the isocentre and closer to the tube leads to geometric magnification of the localiser radiograph and subsequently overestimation of the amount of attenuation in the field of view (AAPM, 2014). The reverse is also true. Unfortunately, the errors in size estimation result in wrong calculation and application of photon flux to the patient (AAPM, 2014). These could result in image noise and 40 University of Ghana http://ugspace.ug.edu.gh variation in CT number and dose (Kaasalainen et al., 2014). A study found that improper positioning of patients in the isocentre could increase breast and thyroid surface dose by 16% and 24%, respectively (Raman et al., 2013). 2.7 Diagnostic Reference Level DRL’s introduction in medicine, its optimisation role in CT, types (anatomical, indication- based) and existing values, and theories guiding DRL development in CT have been covered in Section 2.7.1- 2.7.6. 2.7.1 Diagnostic Reference Level in Medicine A Diagnostic Reference Level (DRL) is defined by the ICRP as; “a form of investigation level, applied to an easily measured quantity, usually the absorbed dose in air, or tissue-equivalent material at the surface of a simple phantom or a representative patient” (ICRP, 1996 p.2). The term “diagnostic reference level” (DRL) was first introduced in 1996 in the ICRP Publication 73 (ICRP, 1996), and then developed further in subsequently publications. The activities that preceded the development of the DRL concept were many. However, consensus in literature indicate that the commencement of national dose surveys promulgated the idea in medicine (ICRP, 2017; Vassileva and Rehani, 2015; Sutton et al., 2014; IAEA, 2013; ICRP, 1996; ICRP, 1991; Shrimpton et al., 1989). The ICRP noted that, proceedings from the first nationwide x-ray dose levels conducted in the 1970s in USA, and later in the UK during the 1980s were crucial in the DRL concept. The consequences of these and other surveys were the reason for proposals for 41 University of Ghana http://ugspace.ug.edu.gh radiographic technique and dose benchmarks, of which USA and UK were the pioneers (ICRP, 2017). The initial recommendations, which aimed at benchmarking radiation exposure level to prevent unnecessary radiation in medicine, have been mentioned variously as exposure guides, guideline doses, guidance levels, and reference doses (ICRP, 2017). In the UK specifically, the Royal College of Radiologists and the National Radiological Protection Board (NRPB) jointly agreed and introduced the concept of reference doses (Sutton et al., 2014). This was first introduced in general radiography, and later surveys indicated “some 1300-man Sv could be saved by persuading the 25% of hospitals with the higher doses for the six examinations they investigated to change their technique to fall in line with the remaining 75%” (Sutton et al., 2014 p.1). In the ICRP’s 1991 Publication 60, the concept of investigation levels (which was later referred to as diagnostic reference levels) for diagnostic medical exposures was first proposed by the ICRP (Hart et al., 2008). Particularly, the publication recommended the establishment of “investigation levels, selected by the appropriate professional or regulatory agency, for application in some common diagnostic procedures” (ICRP, 2017 p.138). Reference levels were referred to as values of measured dose descriptors. Beyond that, specified measures or decisions ought to be taken (ICRP, 1991; ICRP, 2007). They involved “recording levels, above which a result is recorded, lower values being reported; investigation levels, above which the cause of the implications of the results should be examined and interventional levels, above which some remedial action should be taken” (ICRP, 1991; ICRP, 2007). However, these levels were not decoupled from the principle of dose constraint, hence the need for it to be utilised with flexibility, to permit higher doses if required in accordance with sound clinical judgment (ICRP, 1991; ICRP, 2007). 42 University of Ghana http://ugspace.ug.edu.gh In the ICRP Publication 73 (ICRP, 1996), the Commission decoupled the two principles and introduced the term of diagnostic reference levels and discussed the concept in more details (ICRP, 1996). The European Commission subsequently included DRLs in a recommendation on diagnostic medical exposures in 1997, and the concept was advanced further, and practical advice was given in 2001 increasing the application of DRLs to interventional radiology and providing further guidance on flexibility in their development and implementation (ICRP, 2016). In the ICRP publication 103, the DRLs role in optimisations were further highlighted (ICRP, 2007). The current ICRP Publication 135 on DRLs (ICRP, 2017), was also introduced following the need for additional advice regarding “definitions of the terms used in previous guidance, determination of the values for DRLs, the appropriate interval for re-evaluating and updating these values, appropriate use of DRLs in clinical practice, methods for practical application of DRL process, and application of the DRL concept to newer imaging technologies” (ICRP, 2017 p.9). These technologies include PET-CT, cone beam CT, dual-energy computed tomography (CT), SPECT-CT, digital radiography, and tomosynthesis (ICRP, 2017). Recommendations to address the challenges in paediatric practice, especially due to the broad range in sizes, were also provided. The new definition introduced by the ICRP includes: i. DRL, which is a form of investigation level used as a tool to aid optimisation of protection, in the medical exposure of patients for diagnostic and interventional procedures. ii. DRL quantity, which is a commonly and easily measured or determined radiation metric that assesses the amount of ionising radiation used to perform a medical imaging task. 43 University of Ghana http://ugspace.ug.edu.gh iii. DRL value, which is an arbitrary notional value of a DRL quantity, set at the 75th percentile of the distribution of the medians of distributions of the DRL quantity obtained from surveys or other means. iv. DRL process, which is the cyclical process of establishing DRL values, using them as a tool for optimisation, and then determining updated DRL values as tools for further optimisation (ICRP, 2017). 2.7.2 Role of DRLs in Computed Tomography Dose Optimisation As compared to a screen film-radiograph, CT images do not easily give indication of over/underexposure, hence, images may appear good (often better) even if excessive and unnecessary radiation is used (Bauhs et al., 2008). The concept of DRLs was introduced in medicine as an additional tool to basically manage, optimise, monitor and control radiation use in medicine without compromising image quality or patient care (ICRP, 2017, Vassileva and Rehani, 2015). It should be noted that the application of patient dose management strategies requires the monitoring of low doses as well as high doses to ensure the preservation of the acceptable image quality (O’Connel, 2015). DRLs are critical in addressing the wide variations in doses often observed for same examination and similar patient groups (Kanal et al., 2017). They are not applied to occupational and public exposures but only medical exposure (McCollough, 2017). It is also not an ideal or a regulatory dose limit for a specific examination or an absolute upper limit for dose, but it represents the dose level at which an inquiry into the appropriateness of the dose ought to be initiated (McCollough, 2017). They are numerically established values expected not to be exceeded in routine practice when standards and good practices are adhered to (Tsukamoto, 2017). 44 University of Ghana http://ugspace.ug.edu.gh It is simply an indicative value for identifying circumstances where the level of patient dose or administered activity is remarkably high (Australian Government, 2016; EANM et al., 2017). If the dose delivered by an imaging facility consistently exceeds the established DRLs, it is a sign that the facility (including both the procedure and the equipment) should be reviewed locally for possible causes (ICRP, 2017). The regular review of possible causes using the DRLs as benchmarks at national, regional and local levels provides a feedback loop that ensures good practice (IAEA, 2013). Subsequently, a corrective action could be taken to further optimise the scanning protocols or whatever the problem might be unless clinically, the unusually high doses could be; justified or considered necessary for that particular task (Liang et al., 2017; Australian Government, 2016; Martin, 2016; Alessio et al., 2015; Vassileva and Rehani, 2015). Although DRLs are considered supplements to professional judgment, and therefore do not always offer a boundary between the ‘good’ and ‘bad’ medicine, they make a significant contribution to good radiological practice in healthcare (Do, 2016; Rehani, 2009). As such, a DRL is an essential tool for effective radiation dose management and QA/QC system (Tsukamoto, 2017; Edmonds, 2009). In jurisdictions where DRLs are well established, studies (Liang et al., 2017; MacGregor et al., 2015; Vassileva and Rehani, 2015; Brink and Miller, 2015; Hard and Wall, 2004) have shown that the usage of DRLs helped in reducing the range of doses in clinical practice for some procedures. In particular, the U.K national dose surveys established a 30% reduction in representative radiation doses for the period of 1984 to 1995, and a mean decline of approximately 50% for the period of 1985-2000 as a consequence of the establishment of DRLs and its variants (Hard and Wall, 2004). In Canada, MacGregor et al. (2015) reported a mean radiation dose reduction of 22% for CTDIvol and 13% for DLP during the period of 2010 - 2013 as a result of 45 University of Ghana http://ugspace.ug.edu.gh DRLs implementation. In Finland, a study (Lajunen, 2015) has also reported a decrease of approximately 20% in radiation dose levels between the year 2007 to 2013 following the implementation of DRLs and other optimisation programmes. The role of the DRL implementations in the above dose reductions is that they serve as a watch-mark for radiographers, radiologists, medical physicists, and other medical practitioners to be mindful of in order to optimise practices, in order to stay below the reference values. Owing to the benefits of DRL, adult-related DRLs have been instituted in 72% of the 36 European countries, and in 81% of European Union (EU) and European free trade association (EFTA) countries (Iceland, Norway and Switzerland) (European Commission, 2014). Moreover, 39% of the countries have instituted DRLs and 45% of EU and EFTA countries have done same for paediatric x-ray examinations (European Commission, 2014). The WHO and the IAEA have also included in their Bonn Call for Action 2, the need for the establishment, use of, and the frequent update of DRLs for radiological examinations (WHO, 2016; IAEA, 2013, IAEA GSR Part 3, 2014), while the European Society of Radiology (ESR) has recently included it in its EuroSafe Imaging Call for Action 2018 (European Society of Radiology, 2018). 2.7.3 Anatomical DRLs in Computed Tomography Anatomical CT DRLs are developed for a specific anatomical part/region of the body. The most common anatomical regions of the body for which DRLs are developed include the head, chest, abdomen, abdominopelvic, lumbar spine, pelvis, neck and a combined chest and abdomen. Anatomical DRLs are used to benchmark radiation dose levels for examinations undertaken around these body parts and are the most widely developed DRLs in CT (Damilakis et al., 2018). However, the limitation associated with this type of DRLs is that, only one DRL is used to account 46 University of Ghana http://ugspace.ug.edu.gh for benchmark and monitor all the procedures associated with the specific anatomical area. It is strongly argued that in CT imaging, the procedures are tailored tailed to a specific clinical question and different imaging protocols and image quality requirements are needed for each clinical question. Therefore, great variations in radiation dose exist among procedures performed under the same anatomical imaging (Vocks and Frija, 2016; European Society of Radiology, 2017). Hence, an anatomical DRL as a QA/QC tool presents some ambiguities which could lead to unnecessary radiations (European Society of Radiology, 2017; Lajunen, 2015; Järvinen et al., 2015). The European Society of Radiology (2017), has further argued that NDRLs based on anatomical regions do not reflect clinical practice and should be re-examined using clinical indications. A literature search produced no established NDRLs in Ghana. However, a couple of studies provided local anatomical reference levels. In particular, a study (Anim-Sampong, 2016) reported anatomical adult local DRLs (LDRLs) of head (63.29 mGy; 1008.26 mGy.cm), chest (5.92 mGy; 282.14 mGy.cm) and abdomen (6.79 mGy; 353.48 mGy.cm) for the Korle Bu Teaching Hospital. The paediatric values were head (23.87 mGy; 406.59 mGy), chest (1.70 mGy; 67.20 mGy.cm) and abdomen (2.70 mGy; 93 mGy.cm). The adult head examination especially was 5.5 % higher than the ICRP DRL value (ICRP, 2001). Another study (Addo, 2016) proposed a local paediatric DRLs for the Greater Accra Region in Ghana using only four facilities. For the head region, the age category and their DRLs were < 1 year (28 mGy; 395 mGy.cm), 1-5 years (38 mGy; 487 mGy.cm), 6-10 years (48 mGy; 601 mGy.cm) and 11-15 years (86 mGy; 1614 mGy.cm). In the chest region, DRLs for < 1 year and 1-5 years groupings were (1 mGy; 18 mGy.cm) and (5 mGy; 110 mGy.cm), respectively. Their abdominopelvic DRLs were (3 mGy; 71 mGy.cm, < 1 year), (3 mGy; 120 mGy.cm, 1-5 years) and (10 mGy; 494 mGy.cm, 6-10 years). 47 University of Ghana http://ugspace.ug.edu.gh Meanwhile, other studies (Inkoom et al., 2014; Muhogora et al., 2010; Muhogora et al., 2010) have tried to benchmark radiation doses in Ghana. However, the mean dose distribution, instead of the 75th percentile of the dose surveyed, were reported, and therefore could not be considered as DRLs values. Internationally, studies on anatomical DRLs are very vast. Since anatomical DRL is not the direct focus of the study, a few of the current studies (< 8 years) have been reviewed here to present a reflection of the current global trend on anatomical DRL values. Tables 2.5 and 2.6 present a summary of the reviewed national anatomical DRLs for adults and paediatrics, respectively. The NDRLs for adult head CTDIvol examinations ranged from 30.4-75 mGy, while the DLP ranged from 760 - 1358.6 mGy.cm. Other observed anatomical DRLs were in the range of chest (9.3-30 mGy; 270-735 mGy.cm); abdomen (13.3-18 mGy; 204-555 mGy.cm); abdominopelvic (13-35 mGy; 460-1486 mGy.cm); and a combined procedure of chest, abdomen and pelvis (11-32.8 mGy; 421-1322 mGy.cm). Paediatric head DRLs for age groupings < 1 year, 1–5 years, > 5–10 years and > 10–15 years were in a range of (20-31 mGy; 207-1060 mGy.cm); (35-48 mGy; 308-1493 mGy.cm); (40.3-54 mGy; 4467-1824 mGy.cm); and (35-53 mGy, 600-920 mGy.cm), respectively. The details of the reviewed DRLs are shown in Tables 2.5 and 2.6. 48 University of Ghana http://ugspace.ug.edu.gh Table 2.5: Reviewed existing adult national anatomical DRLs Abdomen A Head Chest Chest+abd+pelvis Authors/location /AbdominopelvicAP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP Ekpo et al., 61 1310 17 735 20 1486* - - (2018), NG. Simantirakis et al., (2015),GR. 66.7 1053 14.4 470 16.3 758* - - Salama et al., (2017), EG. 30.4 1358.6 - - 31 1323* 32.8 1322 Butler and Kanal, (2018), US. 57 1011 13 353 16 639* 15 779 Public Health England (2016), 63 973 12 610 13 745* 14 1000 UK, Ataç et al., 66.4 810 11.6 289 13.3 204# 19.4 421 (2015), TR. van der Molen et al., (2013), NL. - 813.7 - 320 - 567# - - Palorini et al., (2014), IT. 69 1312 15 569 18 555# 18 - Foley et al., (2012),IE 66.2 940 9.3 393 11.6 845* - - Zhou et al., (2019), CN. 51.7 906.5 10.3 412.6 18.2 886.9* - - Santos et al., 75 1010 14 470 18 800* - - (2014), PT. ARPANSA, 52 880 10 390 13 600* 11 940 (2018), AU American College of Radiology, 56 962 13 353 16 639* 15 779 (2018) European Union, 760- 270- 460- 50-75 10-30 13-35 - - (2014) 1300 700 1200* Key: A = abdomen (abd), AP = Abdominopelvic, NG= Nigeria, GR= Greece, EG= Egypt, US = United States of America, UK= United Kingdom, TR= Turkey, NL=Netherlands, IT= Italy, IE= Ireland, CN= China, PT= Portugal, AU= Australia, ARPANSA=Australian Radiation Protection and Nuclear Safety Agency; EU= European Union. CTDIvol =volume weighted CT dose index, DLP= dose length product. Unit; CTDIvol =mGy; DLP=mGy.cm 49 University of Ghana http://ugspace.ug.edu.gh Table 2.6: Reviewed existing paediatric national anatomical DRLs Paediatric CT < 1 year 1–5 years > 5–10 years >10–15 years part/authors CTDI vol DRL CTDIvol DLP CTDIvol DLP CTDIvol DLP Head Ataç et al., 31 288 33.4 368 40.3 467 51.3 625 (2015), TR. Head Ekpo et al., 27 1040 48 1493 54 1824 - - (2019), NG Head European 20-27 270- 35-40 470- 50 620- - 850-920 Union, 340 600 900 (2014), EU Head Vassileva et 26 440 36.0 540 43.0 690 53.0 840 al., (2015), IAEA Head ARPANSA, 30 470 - - - - 35 600 (2018), AU. Chest Ataç et al., 7.1 181 3.8 214 3.6 277 4.0 287 (2015),TR. Chest European - 12- - 55- - 105- - 200-205 Union, 200 230 370 (2014), EU Chest Vassileva et 5.2 130 6.0 140 6.8 170 7.3 300 al., (2015), IAEA Chest Santos et al., 2.42 42.75 5.6 138.5 5.65 186 7.19 194.5 (2014), PT Chest ARPANSA, 2 60 - - - - 5 110 (2018), AU. AP European - 27- - 125- - 240- - 400-500 Union, 130 230 400 (2014), EU AP Vassileva et 5.2 130 7.0 250 7.8 310 9.8 460 al., (2015), IAEA AP ARPANSA, 7 170 - - - - 10 390 (2018), AU AP Ataç et al., 3.1 104 2.5 125 2.7 179 3.1 210 (2015), TR. Key: AP= abdomen/pelvis, TR= Turkey, NG= Nigeria; EU= European Union. IAEA= International atomic energy agency, AU= Australia, CTDIvol =volume weighted CT dose index, DLP= dose length product. Unit; CTDIvol =mGy; DLP=mGy.cm 50 University of Ghana http://ugspace.ug.edu.gh The reviewed literature presents great variations across countries, and thus there is the need for optimisations to harmonise the dose levels. Studies (Smith-Bindman et al., 2019; Rehani, 2015; Vassileva and Rehani, 2015; Kalra et al., 2004) have linked the wide variations in doses, to the use of over-age CT scanners in some jurisdictions, differences in technique factors, variation in protocols, anatomical size and characteristics of a population, equipment technical factors, operationalisation of scanners, and a general lack of harmonisation and application of the same optimisation methods across countries, to improve radiation protection for patients. 2.7.4 Indication-Based Diagnostic Reference Levels In medicine, an indication is a medical reason or a condition that leads to the recommendation of a test, procedure or treatment (National Cancer Institute, 2017). In the diagnostic radiology department, patients are referred for various examinations, based on their indications. These CT indications are several in literature. Notwithstanding, studies (Fertikh, 2015; Marder et al., 2014) have indicated that head trauma (mostly through road traffic accidents), stroke or cerebrovascular accident (CVA) and headaches are the most prevalent head indications in CT imaging. In the case of stroke, a study (Marder et al., 2014) indicated that CT could differentiate infarction from haemorrhage, and distinguish from other causes like the extracerebral haemorrhage or glioma. Moreover, if CT is performed within 48-72 hours, it could diagnose subarachnoid haemorrhage in over 90% of cases (Marder et al., 2014). A publication (Hamed et al., 2015) explained that the suspicion of space occupying lesion (SOL) (usually because of malignancy, abscess or a haematoma) is the most common reason for headache related CT examinations. In the chest region, some literature (Purysko, 2016; Willson et al., 2014) indicated that lung cancer and other primary neoplastic conditions affecting the chest and metastatic diseases are the most 51 University of Ghana http://ugspace.ug.edu.gh common chest CT indications. Other authors (Skinner, 2015; Gould et al., 2013) reported that suspected pulmonary fibrosis, pneumonia, and solitary pulmonary nodule are the most indications in CT chest imaging. A study (Wilson et al., 2014) also found that in emergency CT scans, pulmonary embolism accounts for about 80% of the indications. Moreover, publications (Garcia et al., 2018; Taylor, 2017) suggest that the common CT indications of the abdomen and the pelvic regions include: abdominal pain, abdominal tumour or lesion, renal calculi or kidney stones and abdominal injury. Others included acute appendicitis, bowel obstruction, obstructive jaundice and metastasis (spread of cancer from the primary site) (Taylor, 2017). In CT imaging in particular, different imaging protocols are set for specific indications (Trattner et al., 2014). For instance, in the case of brain tumour, there are different sets of imaging protocols compared to CVA indications, although they are all head-related procedures. Each specific indication requires a specific diagnostic imaging requirement for its diagnosis. A DRL that is based upon a radiation dose data derived as a result of CT imaging for a specific indication, is referred to as an indication-based DRL (Järvinen et al., 2015; EANM et al., 2017). It is also referred to as a clinical DRL in some publications (Damilakis et al., 2018). The concept of indication-based DRLs was mentioned many years ago by the ICRP (ICRP, 2017), however, the vast majority of published DRLs have been on anatomical locations (Damilakis et al., 2018). The anatomical approach has bracketed all indications for a particular anatomical region into one DRL. For example, presently, an abdominal DRL is used as a dose benchmark for diagnosing kidney stones, appendicitis, lesions in the abdomen and liver metastases in many establishments (Vock and Frija, 2016). However, as mentioned earlier, the scanning protocols (such as collimation, contrast usage, scan length, series, etc.) which have a corresponding effect on image quality and dose, differ amongst the different indications for the anatomical part (European Society of 52 University of Ghana http://ugspace.ug.edu.gh Radiology, 2017). Some authors (García-Mónaco et al., 2019; Vock and Frija, 2016) have argued that it is not logical to use the same (e.g. abdominal) DRL for the diagnosis of kidney stones and of liver metastases. Hence, a single anatomical DRL for many indications of a body region is not a very reasonable indictor for dose optimisation, unless the clinical indication is considered (García-Mónaco et al., 2019). Reports (Ridley, 2016; Vock and Frija, 2016) have revealed that indication-based DRLs based on CT protocols could decrease patients’ dose considerably. A recent publication (EANM et al., 2017) further indicated that indication-based DRLs and its implementation could narrow down dose monitoring in CT examination and offer another level of accountability concerning the usage of radiation in medicine. It also argued that setting up indication–based DRLs would involve more classification of patient diseases where different dose outputs for one anatomical part are realistic, but this will help to reduce the overall dose to patients since, it could complement the other existing DRLs (Bujila, et al., 2018). Due to the many indications in CT, DRLs cannot be set simultaneously for all of them. Moreover, some indications share similar diagnostic requirement and protocols hence they could have common DRL values. However, it is advised that initial indication DRLs should centre on the most common ones contributing enormously to a population dose (Vock and Frija, 2016). These could include “acute head trauma, acute stroke, pulmonary embolus, pulmonary metastases, diffuse parenchymal disease, liver metastases, urinary calculus (stone), appendicitis, CT colonography and calcium coronary scoring” (Vock and Frija, 2016 p. 7-8). 53 University of Ghana http://ugspace.ug.edu.gh 2.7.5 Existing Indication-Based DRLs in Computed Tomography There was no literature on either local or national indication-based DRLs in Ghana. International publications on indication-based DRLs are few, unlike the anatomical DRLs. The available DRLs are largely from Europe, with none from Africa, to re-emphasise the importance of this current study as a model indication-based DRLs for Africa. Based on the literature, the most common adult indications and their DRLs were CVA/stroke (60-80 mGy; 970mGy.cm), brain haemorrhage (58-65 mGy, 930-1000 mGy.cm), sinusitis (9-25 mGy, 120-350 mGy.cm), lung cancer (9-16 mGy; 350-610 mGy.cm), pulmonary embolism (13-19 mGy, 300-557 mGy.cm) and Coronary CT angiogram (20-90 mGy; 173-1510 mGy.cm). Tables 2.7-2.10 summarise the available adult indication-based DRLs as obtained from the literature. 54 University of Ghana http://ugspace.ug.edu.gh Table 2.7: Adult head CT DRLs, based on clinical indications Ind. Acute Acute Acute Head trauma/ Haemorrhage, Brain Sinusitis cholesteatoma stroke/post stroke/ stroke/brain Injury aneurysms, metastases, fossa Cerebrum (whole) arteriovenous SOL Ref. malformations abscess CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP a I - - - - - - - - 58 930* - - - - - - IIa 80 - 60 - 60 970* - - - - - - - - - IIIa - - - - - - - - - - - - 9 120* - - IVa - - - - - - - - 65 1000* 65 1000 25 350* 50 250* Va - - - - - - - - - 936* - - - 133* - - VIa - - - - - - - - - - - - - 90* - - VIIa - - - - - - 60 950 - - - - - - - - Key: Ind.= indication, Ref. = Reference. SOL= space occupying lesion. CTDIvol =volume weighted CT dose index, DLP= dose length product. Unit; CTDIvol =mGy; DLP=mGy.cm. References: [I: Danish Health Authority, (2015), DK; II: Public Health England, (2016), UK; III: Schegerer et al., (2017), DE; IV: Treier et al., (2010), CH; V: Van der Molen et al., (2013), NL; VI: Wachabauer et al., (2017), AT; VII: Widmark, (2018), NO]. a Data with regulatory value implemented * single sequence procedure. 55 University of Ghana http://ugspace.ug.edu.gh Table 2.8: Adult chest CT DRLs, based on clinical indications Indication Lung cancer Interstitial Interstitial Pulmonary Coronary CT Calcium lung lung disease embolism angiogram Scoring disease (axial) (helical) (CCTA) Reference CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP a ? Danish Health Authority, (DK) (2015) 16 620 - - 13 500* - - 29 230* - - a ? Public Health England, (UK), (2016) 12 610 4 140* 12 350* 13 440* - - - - a Schegerer et al., (DE), (2017) - - - - - - 15 300* - - 8 119* Treier et al., (CH), (2010)a - - - - - - - - - 1000? 150* a Van der Molen et al., (NL), (2013) - - - - - 276* - 371* - 671* - 51* Wachabauer et al., (AT), (2017)a - - - - - - - 400* - - - Widmark, (NO), (2018)a 9 350* Castellano et al., (UK), (2017)a - - - - - - - - - 173* - - Foley et al., (IR), (2012) 7 276* 13 432* - - - - Fukushima et al., (JP), (2012) - - - - - - - - - 1510? - - German Federal Office for Radiation - - - - - - - - 20 330* - - Protection, (DE), (2016)a Hausleiter et al., (2009) - - - - - - - - - 1152? - - Japan Network for Research and Information - - - - - - - - 90 1400? - - on Medical Exposures, (JP), (2015)a Kanal et al., (USA), (2017) - - - - - - 19 557? - - - - Mafalanka et al., (FR), (2015)a - - - - - - - - - 870* - - Palorini et al., (IT), (2014) - - - - - - - - - 1208? - 131* Radiation and Nuclear Safety Authority (FI), 11 430* - - - - - - - - - - (2013)a and Lajunen, (2015), (FI) Salama et al., (EG), (2017) - - - - 22 421* - - - - - - Key: CTDIvol =volume weighted CT dose index, DLP= dose length product. Unit; CTDIvol =mGy; DLP=mGy.cm). a Data with regulatory value implemented * single sequence procedure, ?undeclared number of sequences. Letters in bracket eg. DK, UK are internationally accepted country abbreviations. a Data with regulatory value implemented. 56 University of Ghana http://ugspace.ug.edu.gh Table 2.9: Adult abdomen and pelvic region CT DRLs, based on clinical indications Indication Liver Abdomen- Kidney Kidney Acute Pancreas Metastases pelvic abscess stones/colic tumour/colic Abdomen Cancer CT-IVU Reference CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP Danish Health Authority, (DK), (2015)a - - - - - - - - 17 700 - - Public Health England, (UK), (2016)a 14 910 15 745 10 460* 13 1150 - - - - Radiation and Nuclear Safety Authority (FI) - - - - 7 330* - - - - - - (2013) a and Lajunen, (2015), FI. Salama et al., (EG), (2017) 31 1423 - - - - - - - - - - Treier et al., (CH), (2010)a 15 400 - - - - - - - - - - Van der Molen et al., (NL), (2013)a - - - - - 329* - 1371 - - - 1000 Wachabauer et al., (AT), (2017)a - 400 - - - - - - - - - - Widmark, (NO), (2018)a 11 800 - - - - 13 1300 - - - - Key: CTDIvol =volume weighted CT dose index, DLP= dose length product. Unit; CTDIvol =mGy; DLP=mGy.cm). CT-IVU= computed tomography intravenous urography. a Data with regulatory value implemented * single sequence procedure. Letters in bracket eg. DK, UK are internationally accepted country abbreviations. 57 University of Ghana http://ugspace.ug.edu.gh Cont'd Table 2.9: Adult abdomen and pelvic region CT DRLs based on clinical indications Indication. Abscess Virtual colonoscopy CT angiography for Suspicion of lymphoma Lymphadenopathy (VC) - polyps/tumour abdominal aortic Reference aneurysm (AAA) CTDIvol DLP CTDIvol DLP CTDIvol DLP CTDIvol DLP (mGy) (mGy.cm) (mGy) (mGy.cm) (mGy) (mGy.cm) (mGy) (mGy.cm) Public Health England, (UK), (2016)a 15 745 11 950 - - - - Treier et al., (CH), (2010)a 15 650 - - 15 650 - - Van der Molen et al., (NL) (2013)a - - - - - 727 - - Wachabauer et al., (AT), (2017)a - 650 - - - - - - Lajunen, (FI), (2015) - - 12 930 - - 17 970 Key: CTDIvol =volume weighted CT dose index, DLP= dose length product. Unit; CTDIvol =mGy; DLP=mGy.cm). a Data with regulatory value implemented * single sequence procedure. Letters in bracket eg. UK, CH are internationally accepted country abbreviations. Table 2.10: Adult cervical CT DRLs, based on clinical indications Indication. Fracture Disk pathology Adenopathy, abscesses Reference CTDIvol DLP CTDIvol DLP CTDIvol DLP (mGy) (mGy.cm) (mGy) (mGy.cm) (mGy) (mGy.cm) German Federal Office for Radiation 20* - 25* - - - Protection, (DE), (2016)a Public Health England (UK), (2016)a 26* 600* - - - - Treier et al., (CH), (2010)a - - - - 30* 600* Key: CTDIvol =volume weighted CTDI, DLP= dose length product. Unit; CTDIvol =mGy; DLP=mGy.cm). a Data with regulatory value implemented * single sequence procedure. Letters in bracket eg. DE, CH are internationally accepted country abbreviations. 58 University of Ghana http://ugspace.ug.edu.gh 2.7.6 Theories Guiding DRL Development in Computed Tomography Studies (American College of Radiology, 2018; ICRP, 2017; ICRP, 2007) show that the process leading to the establishment of DRLs requires co-operations among stakeholders within an organisation responsible for the establishment and implementations of DRLs. Other publications (Do, 2016; Korir et al., 2016) have suggested that although government, through the authorised regulatory body has a duty to ensure that NDRLs are established for a country, the development of DRL values could be conducted by an individual, institutions or professional medical bodies. However, such derived numerical values are advisory, and the implementation is supposed to be done by an authorised body (Do, 2016). According to the IAEA, in certain countries, a national governmental body manages the national patient dose database for DRLs whereas in other countries, this role is taken by a regulatory body or a professional body (IAEA, 2013). A publication (Järvinen et al., 2017) has further stressed that successful DRLs developments have largely been based on three factors, namely, national actions, efforts at the level of local institutions and efforts by individuals. The current ICRP recommendation on DRLs (ICRP, 2017) suggests that DRLs could also be adopted by institutions based upon published values that are suitable for local conditions, yet managements must take responsibilities for such DRLs. Same can also be done for countries where responsible bodies failed to establish DRLs based on their own dose registry or dosimetry (IAEA, 2013). A publication (European Commission, 2014), indicates that about 77% of adult DRLs in Europe are based on the countries’ own national dose surveys, while the remainder are adopted values. However, the disadvantage of adopting existing DRLs is that local levels may be far lower than the set DRLs and sometimes could defeat the purpose of dose optimisation (Rogers, 2014). 59 University of Ghana http://ugspace.ug.edu.gh Available literature (ICRP, 2017; IAEA, 2013) further indicates that, the DRL descriptors (quantity) on which DRLs are developed should be centred on metric that can easily and directly be measured in a practical way to reflect what is used in the clinical practice. Particularly, dose descriptors from a direct measurement for a procedure, or that obtainable from the imaging equipment which indicate the amount of radiation used are preferred (ICRP, 2017; IAEA, 2013). The IAEA has argued that effective dose, for instance, is a key indicator for estimating radiation risks statistically; however, in DRL it is not applicable as a key indicator because the quantity requires additional assumptions (IAEA, 2013). Currently, in MDCTs technology, dose quantities for DRL establishments are the CTDIvol of each sequence and cumulative DLP for the entire examination (ICRP, 2017), which are mandatory to display on the console of CT scanners, post examination (International Electrotechnical Commission, 2011; Institute of Physics and Engineering in Medicine, 2005). Size-specific dose estimate (SSDE) is thought to provide a more accurate estimate in the future, particularly, of paediatric patient doses than the aforementioned metrics, when scanner technology provides automatic calculation (ICRP, 2017; AAPM, 2014). Studies (ICRP, 2017; Public Health England, 2016; IAEA, 2013, Smith-Bindman and Miglioretti, 2011) have ascertained that CT-console-displayed quantities are reliable for establishing DRLs, provided equipment status and validation checks are performed as part of local QA/QC measures. This requires that scanner performance characteristics (in particular CTDIvol and DLP) shown on CT console are comparable to the results of a well calibrated CT dose quantity measuring device (such as ionisation chamber traceable to a primary or secondary standard laboratory) which has been scanned under same conditions (ICRP, 2017; Public Health England, 2016; IAEA, 2013; ICRP, 2007). In instances where a particular scanner fails a performance test, it could be excluded, or a correction factor could be applied in the DRL study (IAEA, 2012). 60 University of Ghana http://ugspace.ug.edu.gh There are two methods of measuring doses and developing DRLs in diagnostic imaging like CT examinations. These involve the use of phantom data (phantom-based dosimetry) and patient data (patient-based dosimetry) (National Council on Radiation Protection and Measurements, 2012; Vassileva and Rehani, 2015). An advantage of the phantom-based dosimetry is that it requires only one or two exposures for each procedure type, and for each radiologic facility. However, the drawback of this approach is that phantom-based dosimetry does not characterise actual clinical state (Vassileva and Rehani, 2015). In the case of patient-based dosimetry, patients’ data are generated during a clinical task to reflect practice. Many international bodies (ICRP, (2017; Public Health England, 2016; IAEA, 2013) strongly recommend patient- based dosimetry as the gold standard in CT-based DRLs. Table 2.11 shows the ICRP’s recommended method of dose assessment/survey for DRL development for various modalities in medical imaging (ICRP, 2017). 61 University of Ghana http://ugspace.ug.edu.gh Table 2.11 Examination selection and ICRP’s recommended assessment method for DRLs (ICRP, 2017). 62 University of Ghana http://ugspace.ug.edu.gh Studies (ICRP, 2017; IAEA, 2013; American College of Radiology, 2014; ICRP, 2001), have further suggested that if patient-based dosimetry or measurements are used in DRL developments, a large and representative patient sample, typically of 10 or more, should be selected from each of the facilities involved. However, the current ICRP recommendation on DRLs has suggested at least 20 patients, with 30 as its preferred sample size (ICRP, 2017). A report (Vassileva and Rehani, 2015) further indicated that the total inclusion of available scanners in DRLs development produces a generalisable result. However, this may not be possible in places where there are too many facilities. Many publications (ICRP, 2017; Kanal et al., 2017; Vock and Frija, 2016; IAEA, 2013) have also indicated that a national survey of dose quantities on which initial DRLs could be set should randomly cover about 20-30 of the facilities including medium- and large-sized healthcare facilities. They further suggested that the national survey should include 30-50% of facilities in small countries, while LDRLs should comprise much of the facilities. The ICRP Publication, 135 in particular, also indicated that the development of DRLs should take into consideration the populations’ pertinent characteristics like weight and height or the body mass indexes (ICRP, 2017). Many authors (Kanal et al., 2017; ICRP, 2017; Tsukamoto, 2017; Vassileva and Rehani, 2015, IAEA, 2013) agreed that x-ray beam attenuation depends on the amount of body tissue through which the beam has to penetrate. Therefore, it is important that DRLs are developed for a specific body size by the use of weight restrictions that reflect the common characteristics of a population of study. According to some publications (European Commission, 2014; IAEA, 2013), the standard mean body weight of an adult population is often defined as 70 ± 10 kg and this has been mostly used in setting DRLs. However, a study (Tsukamoto, 2017) noted that country populations vary in weight. Hence, using the 70 ± 10 kg as the reference weight for adult populations may not be necessarily appropriate for all countries. In 63 University of Ghana http://ugspace.ug.edu.gh the case of Japan for instance, it was noted that the weight category of 50-70 kg was a suitable standard weight for establishing DRLs for routine procedures for Japanese adults (Tsukamoto, 2017). Some authors (Vock and Frija, 2016) have also argued that, often patients vary in size from country to country, hence setting DRLs with a standardised sample may not cover very obese and thin patients, and thus, would have some adverse implications on its universal implementation. Consequently, the ICRP Publication 135 has recommended a mean weight of 70 ± 10 kg; or a range of 50–90 kg with 70-kg mean; or an appropriate weight standardisation/restriction set for each population of interest to be used in adult DRLs (ICRP, 2017). This is particularly, crucial if the number of patients per each facility for whom data is collected is limited (< 50), unless large samples (≥ 50) are used (ICRP, 2017). In the case of paediatrics, it has been suggested that the weight band should be categorised at <5 kg, 5–<15 kg, 15 –<30 kg, 30 –<50 kg, and 50–< 80 kg; and where ages are required, it should involve around the ages of 0, 1, 5, 10, and 15 years (ICRP, 2017). A study (Vock and Frija, 2016) has further suggested “age-grouping” for head related DRLs while trunk examinations should be based on “weight and body size groupings” if possible. There are three categories of DRLs namely, local (LDRL), national (NDRL) and regional (RDRL) (ICRP, 2017). Both LDRL and NDRL are established using the 75th percentile of the median values of dose distribution, although the 50th (also referred to as the achievable dose) and 25th percentiles among other values, are sometimes estimated (ICRP, 2017; Vassileva and Rehani, 2015). The LDRL is calculated on the dose distribution gained from radiology facilities in a single large medical unit or a group of medical units within a hospital, surveyed for standardised patient groupings, while the NDRL is calculated from a representative sample of radiology departments in a country (ICRP, 2017). The RDRL on the other hand, is based on median value of NDRLs of 64 University of Ghana http://ugspace.ug.edu.gh countries within a geographic region or a continent (ICRP, 2017). One key philosophy of a DRL is to offer the 25% of the facilities or the practices with median radiation dose outputs above the established DRL to work hard towards dose optimisation (American College of Radiology, 2018; Rogers, 2014). However, it is argued that the DRL on its own is no encouragement to the other 75% of facilities that attained dose outputs lower than the DRL but should guide them to further manage their dose and investigate those that are below the 25th percentile (Zira and Nzotta, 2016; Rogers, 2014). A pictorial view of how a DRL is estimated at the 75th percentile is presented in Figure 2.6. Figure 2.6: A diagram showing the 75th percentile value of a dose distribution (Vock and Frija, 2016). 65 University of Ghana http://ugspace.ug.edu.gh Reviewed older publications (Foley et al., 2012; McCollough et al., 2011; Watson and Coakley, 2010; Treier et al., 2010; Matthews and Brennan, 2009; Institute of Physics and Engineering in Medicine, 2005; ICRP, 1996) have statistically used the mean value of dose descriptors/quantities of each facility in the DRLs estimation. However, more recent studies (Razali et al., 2019; European Union, 2018; Wagner et al., 2018; ICRP, 2017; European commission, 2015; IAEA, 2013) recommend the use of statistical median value, as it is noted that a few outliers could significantly influence mean values. Studies (Ria et al., 2019; ICRP, 2017) further indicated that for effective development of DRLs, it is needful that all the images associated with the dose descriptors are within the acceptable diagnostic quality. They argued that the greatest importance for any diagnostic radiological procedure is to obtain image quality enough for the clinical task. Therefore, developing a DRL, particularly an indication-based DRL, without considering the image quality may significantly limit its objective. Structurally, the anatomical DRLs and indication-based DRLs are similar in their development approaches except that the former is set using radiation dose information associated with anatomical region of the body, while the latter is developed using data from specific clinical indications. Of particular importance in indication-based DRLs, is that the basic diagnostic imaging requirements and the image quality have to be taken into account during data collection (ICRP, 2017). Radiologists or approved documented protocols could be used to define these requirements to select the appropriate images and their corresponding dose descriptors deemed required without including the unnecessary ones (Brat et al., 2019; Healthmanagement, 2018; ICRP, 2017). 66 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE MATERIALS AND METHODS 3.0 Overview This Chapter presents the materials and methods used for conducting the study. It highlights the study design, study area and components, procedures, and data analyses for all the phases (1-6) of the study, as well as ethical consideration and data handling. 3.1 Study Design A research design is a blueprint adopted by a researcher to validly, objectively, accurately and economically answer a research question (Bowling, 2014). There are many available research designs. However, in DRL studies, designs based on numerical data are the gold standard (ICRP, 2017). Therefore, as the study was directed at using numerical data on the CT facilities across the country to develop indication-based DRLs, and propose optimisation approaches for radiation protection of patients in clinical practice in Ghana, it was necessary to use quantitative, prospective, cross-sectional, and experimental study designs in this study. A quantitative cross- sectional design allows a researcher to empirically gather information at one point in time to describe a population of interest (e.g. CT facilities), while experimental design allows interventions to be deliberately introduced, to observe their effects (Mitchell, 2015). In particular, the prospective component of these designs offers opportunities to undertake very crucial QC assessments of the CT scanners prior to the collection and manipulation of needed data. The decision to use these study designs is also supported by literature, as international recommendations (American College of Radiology, 2018; European Society of Radiology, 2018; 67 University of Ghana http://ugspace.ug.edu.gh ICRP, 2017, European Commission, 2015; IAEA, 2013; AAPM, 2008; ICRP, 2007; ICRP, 1996), on DRLs have all been grounded on this approach. 3.2 Study Area and Components In order to situate the study within a research frame and outline the major component of the project, a methodological framework was used (Figure 3.1). Many publications (American College of Radiology, 2018; European Society of Radiology, 2018; ICRP, 2017, Vassileva and Rehani; 2015; European Commission, 2015; IAEA, 2013; Treier et al., 2010; AAPM, 2008; ICRP, 2007; ICRP, 1996), provided the theoretical basis for the successful composition and development of the indication-based DRL methodological framework for this study. Specifically, it is recommended that data for determining NDRL values ought to be based on extensive CT data obtained from surveys or registries (ICRP, 2017). To ensure this, there was the need to first obtain technical data on the CT scanners in the country, since there was a very limited information on the status of the infrastructure that was expected to be used in the study. Secondly, indication-based DRLs require information on the common indications for CT examinations in the country (European Society of Radiology, 2018); therefore, this was part of the scope of the study. For the needed technical dose descriptors (quantity data) of the various indications to be appropriately selected, the basic diagnostic imaging requirements for each indication were also considered, and the concept defined. Moreover, scanner performance characteristics data were also considered. Since these data constitute a dosimetry standard that radiation dose (or quantity) assessments involving CT must relate to, they should be considered in order to account for any technical challenges that may be associated with the scanners in a study framework (IAEA, 2012; ICRP, 2007; European 68 University of Ghana http://ugspace.ug.edu.gh Commission, 1999). Subsequently, CT dose quantity surveys and the image quality assessments of the CT images of the prioritised indications were also considered crucial for the DRL development. In addition, radiation dose impact such as cancer risks associated with the indication-based CT examinations and steps for radiation dose optimisation for the protection of patients in CT examinations in Ghana were also explored. The study was accordingly conducted in phases as indicated in Figure 3.1, where P1-P6 denotes Phases 1 to 6. 69 University of Ghana http://ugspace.ug.edu.gh P1: Technical data on the CT infrastructure, identification of P2: QC assessment of CT common indications & definition scanners of the basic diagnostic imaging requirement P3: CT dose data & image P4: Estimation of DRL P5: Estimation of dose quality assessment values using the 75th impact & cancer risks percentile rule associated with the doses P6: Optimisation Figure 3.1 Methodological framework and flow chart. Key: CT; computed tomography, QC; quality control, DRL; diagnostic reference level 70 University of Ghana http://ugspace.ug.edu.gh 3.3 Phase 1 Study: CT Technical Data, Common Indications and Imaging Requirements The Phase 1 of the study centred on survey of technical data on the CT scanners, identification of common indications and definition of basic diagnostic imaging requirements of the common indications. The procedures for Phase 1 are outlined in Sections 3.3.1 to 3.3.6. 3.3.1 Study Site and Population According to the current ICRP recommendation 135 (ICRP, 2017) on DRLs, a NDRL should be set using a representative number of facilities across the country. According to the Nuclear Regulatory Authority (NRA) of Ghana (Appendix 1), there were 35 CT scanners in Ghana at the time of the study (December 2017). All the CT facilities were targeted for the study and the technical heads responsible for the CT scanners were also invited to participate in the study using an introductory letter from NRA (Appendix II) and an information sheet (Appendix III). The study sites were the CT Units in the Radiology Departments of all participating facilities which comprised public, private and quasi-government hospitals in Ghana. Figure 3.2 displays the regional distribution (number) of CT scanners in Ghana as at December 2017. 71 University of Ghana http://ugspace.ug.edu.gh Figure 3.2: Regional distribution of CT scanners in Ghana as at December 2017 Note: Number of administrative regions in Ghana changed from 10 to 16 since 2019. This occurred after data on regional distribution of CT for this study were collected. 72 University of Ghana http://ugspace.ug.edu.gh 3.3.2 Sample Size, Inclusion and Exclusion Criteria A sample size typically refers to the number of chosen units from which data is gathered (Bowling, 2014). The IAEA has suggested that NDRLs must be established, based on extensive scale surveys of the doses corresponding to typical clinical practice, for a patient group at a range of representative medical facilities (IAEA, 2013). A study (Vock and Frija, 2016) has also indicated that a national survey facility, from which NDRLs could be set, should cover about 30- 50% of the facilities in small countries, while LDRLs should cover much of local facilities in the population. In order to obtain broad data across the country, a total enumeration sample size determination method (Laerd, 2012) was used to include all the scanners in the study, in order to deepen the representativeness of the findings. However, four (4) of the technical heads responsible for the scanners did not sign the consent form (Appendix IV) to involve their equipment in this study, and were subsequently excluded (Phase 1). Therefore, 31 CT scanners and their respective technical heads participated in Phase 1 of the study. 3.3.3 Data Collection Tool According to studies (McElroy and Ladner, 2014; Choudhurg, 2017), a questionnaire is a reliable, very robust and very convenient means of collecting large amounts of technical information and useful comparable survey data from a large number of individuals in a study. The data collection tools utilised in Phase 1 of the study therefore consisted of two self-designed, semi- structured questionnaires, A and B (A = Appendix V; B = Appendix VI). The questionnaires in Phase 1 of the study were used to determine the status of CT infrastructure in Ghana within a short period of time and provided a decision roadmap for the DRL development. Questionnaire A was developed to generally gather technical data on CT scanners and common indications from the 73 University of Ghana http://ugspace.ug.edu.gh assigned technical heads of the authorised CT scanners. Variables considered on the questionnaire included respondents’ demographics (such as professional grade and gender), CT types, features and characteristics, equipment functionality, quality management systems and driving policies, attending professionals, nature of procedures and common indications for which CT scans were requested. Further to this, respondents were asked to specify indications (among the most common) that use similar scanning protocols by using same identification (alphabet) letters to denote them. This was done to ensure that DRLs were set for indications with different scan protocols and dose descriptors, while those with similar protocols were grouped together. Other sub-sectioned variables/items accounted for a total of 67 questions and statements for which answers were required. Many studies (Adejoh et al., 2017; Ngoya et al., 2016; Maskell et al., 2015; Korir et al., 2013; IAEA, 2013; Ofori et al., 2013; Kruskal et al., 2011; Wambani et al., 2010) provided the theoretical background and framework for the questionnaire and based on the outcome of the first questionnaire, the second was also developed. The second questionnaire was aimed at defining local diagnostic imaging requirements for each indication using mainly radiologists, since they determine the diagnostic requirements of each indication. Dedicated documented scanning protocols for various indications were not readily available in most facilities. Knowledge of diagnostic imaging requirements for each indication was necessary for the selection of only appropriate images (excluding the unnecessary ones) and their associated dose descriptors for the proposed DRLs. This was in line with the ICRP recommendation (ICRP, 2017). The items on Questionnaire B required data on scan coverage, scan series number including contrast usage, image quality acceptability level, required scan thickness, acceptable scan mode, preferred scan techniques, AEC usage acceptability and 74 University of Ghana http://ugspace.ug.edu.gh preferred phantom type. The existing literature (ICRP, 2017; IAEA, 2012; European Commission, 1999; ICRP, 2007) on scan protocols served as a guideline to identify the number of items that should be included in the questionnaire for the basic technical parameters. 3.3.4 Questionnaire Validity and Reliability Validity and reliability tests are crucial for self-designed questionnaires to ensure research accuracy and robustness. Validity is referred to as the degree to which a concept is correctly measured in a quantitative study while reliability entails the consistency of a measure or the ability of a tool to consistently produce the same result under the same conditions over time (Heale and Twycross, 2015). To guarantee the validity of the questionnaires, a content validity technique (the use of a review expert panel to evaluate the questionnaire), was employed to ensure that questionnaires adequately covered all the required contents with respect to the variables under investigation in Phase 1 of the study (Lobiondo-Wood and Haber, 2013). Two clinically qualified medical physicists at the University of Ghana School of Nuclear and Allied Sciences assessed and validated the suitability of the questionnaires. First, the developed tools (based on broad literature reviews) were sent to each of them with a categorical rater scale (0 not important to 1 important) to rate the importance of the items on the questionnaires. It was decided that any item/question which scored 0 from both raters would be rejected or removed. However, any question which scored a total of 1 or 2 from both raters was maintained. The raters were also asked to make suggestions on the questionnaires where applicable. Finally, none of the items on the questionnaire were rejected. However, sentences were eventually refined, and grammatical errors corrected. 75 University of Ghana http://ugspace.ug.edu.gh Cohen’s Kappa Statistic was then applied to test the agreement between scores obtained from the first and second rater. Cohen’s Kappa Statistic tool is a very reliable tool for inter-rater agreement testing (McHugh, 2012). The kappa scale ranges from 0 - 1 and graded as: 0 – 0.20: no agreement; 0.21– 0.39: minimal agreement; 0.40–0.59: weak agreement; 0.60–0.79 moderate agreement; 0.80–0.90: strong agreement; and ≥ 0.90: almost perfect agreement (McHugh, 2012). The 0.71 and 0.80 kappa values recorded on Questionnaire A and Questionnaire B, respectively, suggested moderate to strong agreement in responses between the two raters for both research instruments. Subsequently, a pilot study (using two technical heads responsible for CT facilities) and a test- retest reliability analysis (with one-week interval) were used to further assess the reliability of the questionnaires. For questionnaire A, the observed test-retest reliability coefficient was 0.81 (on a reliability coefficient scale ranging from 0-1, where 1: perfect reliability; ≥ 0.9: excellent reliability; ≥ 0.8 < 0.9: good reliability; ≥ 0.7 < 0.8: acceptable reliability; ≥ 0.6 < 0.7: questionable reliability; ≥ 0.5 < 0.6: poor reliability; < 0.5: unacceptable reliability; and 0: no reliability), while questionnaire B recorded a coefficient of 0.76 to suggest an acceptable reliability (Statisticshowto, 2019). 3.3.5 Data Collection Procedure According to ICRP, the establishment of NDRL values should be coordinated with support from national regulatory authorities (ICRP, 2017). The NRA of Ghana supported the data collection process by issuing an introductory letter (Appendix II) to facilitate the researcher’s access to the facilities. The NRA letter and the study’s information sheet and consent form were first sent by the researcher to all the facilities, while email addresses and telephone contacts of 76 University of Ghana http://ugspace.ug.edu.gh those who consented to participate were also retrieved for the study purposes. Questionnaire A was then administered to them by an online survey platform (www.esurvey.org) (eSurvey, 2017). Their email addresses constituted the medium for sending uncompleted and receiving completed questionnaires. This online survey platform was chosen because it was very reliable, free and fast. Moreover, survey responses collected could be viewed in real time. After follow-up calls, the technical heads responsible for 31 scanners responded and completed their respective questionnaires. This took place from December 2017 to March 2018. Questionnaire B was also administered through the same methods to the reporting radiologists and responsible radiographers (officers in charge, where there were no resident radiologists) for data acquisition on the basic diagnostic imaging requirements. 3.3.6 Data Analysis After the survey responses were received by email, the raw data were visually inspected and where necessary, respondents were called for verification. For example, typographical errors, missing of key data, misplaced data, and inconsistencies were checked and addressed. Descriptive analyses of the data in Phase 1 of the study were done using Microsoft Excel version 2013, and basic descriptive statistics such as frequencies and percentages were generated. Graphs, pie charts and tables were then used to present the results. Details of the results are presented in section 4.2. of Chapter Four. 77 University of Ghana http://ugspace.ug.edu.gh 3.4 Phase 2 Study: Scanners’ Performance Characteristics (QC tests) 3.4.1 Outline According to ICRP (2017), the quantity used for DRL establishment should be acquired in a practical method. Currently, MDCT dose descriptors for DRL establishments are CTDIvol and DLP, which are mandatorily displayed on the console of CT scanners (ICRP, 2017; International Electrotechnical Commission, 2011; Institute of Physics and Engineering in Medicine, 2005). Many studies (ICRP, 2017; Public Health England, 2016; IAEA, 2013; Smith-Bindman and Miglioretti, 2011; ICRP, 2007; ICRP, 1996) have indicated that CT-console-displayed quantities could be used to establish DRLs, if validation checks are performed as part of local QC measures. This dictates that scanner performance characteristics shown on CT consoles are comparable to the results of a well calibrated CT dose quantity measuring device (such as an ionisation chamber) scanned under same conditions. Since the study needed to use console-displayed CT data, dose delivery validation (QC testing) of the displayed quantities were undertaken to ensure that scanners accurately displayed measured quantities. This was also necessary to identify poorly functioning scanners to enable application of appropriate corrective factors in the DRL development, where necessary (Treier et al., 2010). The QC tests undertaken were CT dose delivery accuracy, reproducibility and geometric efficiency. Other assessments included tube voltage accuracy, half value layer (HVL), CT number (water), homogeneity and image noise testing. In all, twenty-five (25) scanners were involved in Phase 2 of the study, following the exclusion of non-functional, specially dedicated and technically challenged (inability to display dose descriptors) CT scanners that were determined in the Phase 1 study. Figure 3.3 presents a flow chart of how the scanners were selected for the Phase 2 of the study. 78 University of Ghana http://ugspace.ug.edu.gh Figure 3.3: Flowchart of scanner selection. NRA: Nuclear Regulatory Authority, CT: computed tomography, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 79 University of Ghana http://ugspace.ug.edu.gh 3.4.2 Materials The materials used in the QC assessment included: 1. CT dose profiler probe (RTI Electronics, Mölndal, Sweden). The calibration was undertaken by Swedac Ackreditering of RTI Electronics (ISO 17025) on 11th February, 2017. (Serial number; DP2-11110079, Certificate number; 1111F7445). A diagram of an RTI CT dose profiler is shown in Figure 3.4. Figure 3.4: RTI CT dose profiler (RTI Group, 2016). 2. Barracuda set with Ocean Software Interface (RTI Electronics, Mölndal, Sweden). The set was made up of two components, the cabinet and multipurpose detector (MPD). The calibration was undertaken by Swedac Ackreditering of RTI Electronics (ISO 17025) on 01 January, 2017. The serial number and certificate number for each device were, cabinet (EBW- 11100142; 1111G2964) and MPD (MPD-12010026; 121AB74B, respectively). A diagram of Barracuda set is shown in Figure 3.5. Figure 3.5: Barracuda set (A= Cabinet, and B=Multipurpose Detector, MPD) (RTI Group, 2016). 80 University of Ghana http://ugspace.ug.edu.gh 3. Standard CT head and body polymethyl methacrylate (PMMA) phantom (PTW, Freighburg, Germany). The head phantom is 16 cm in diameter while the body phantom is 32 cm in diameter. Both were 15 cm long. For the placement of dose profiler or ionisation chamber, five probe holes were provided (one in the centre and four around the perimeter) in the PMMA combined phantom of head and body. Figure 3.6 shows the PMMA phantoms used in the study. Figure 3.6: Standard PMMA Phantoms (A; head, A and B fused together to form body phantom) (RTI Group, 2016). 4. A laptop computer (Toshiba Satellite S55-C5274, 2016 model) 5. A uniform water phantom (PTW, Freighburg, Germany). 6. Microsoft excel spreadsheet installed on a computer (Microsoft version 2013). 7. ImageQC software v.1.43 (EllenWasbo, 2018; Stavanger, Norway). 81 University of Ghana http://ugspace.ug.edu.gh 3.4.3 Data Collection Procedure Dose assessments for CTDI were performed using a standard CT head and body PMMA phantoms (PTW, Freighburg, Germany) together with an RTI CT dose profiler probe (RTI Electronics, Mölndal, Sweden) with Barracuda set connected to Ocean Software Interface (RTI Electronics, Mölndal, Sweden). The advantage of the RTI CT dose profiler probe over the traditional pencil-shaped 10-cm ionisation chamber is that it limits inaccurate dose measurements due to underestimation of the dose profile for wide beams (RTI Group, 2016). The RTI CT dose profiler probe provides an opportunity for dose measurements in every point of the x-ray beam and the total dose profile is attained irrespective of beam width (RTI Group, 2016). Moreover, the CT dose profiler is grounded on solid-state technology, robust and fits into existing standard phantoms used for CTDI measurements (RTI Group, 2016). Another reason for using it in this study was that, it was based on an accurate algorithm that allows the placement of a CT dose profiler probe in the centre of a phantom and could generate a completed picture of the dose profile such as CTDI (100), CTDI(w), CTDIvol and DLP in one exposure (RTI Group, 2016). The allowance of only one helical scan, instead of the normal five axial scans (because of the automatic compensations in the program) supports the comparison of measured and console-displayed values for all the selected CT scanners in a timely manner. 3.4.3.1 CT Dose Delivery Accuracy CT dose delivery accuracy is a QC test used to examine the accuracy level of displayed dose quantities. To measure this, both head and body PMMA phantoms were used with the CT dose profiler in situ. The dose delivery accuracy QC was conducted separately for each phantom type. The CT dose profiler probe (detector) was connected to a Barracuda set with a cable which 82 University of Ghana http://ugspace.ug.edu.gh was also connected to a laptop computer with Ocean Software Interface (RTI Electronics, Mölndal, Sweden). The detector was placed in the central hole of the phantom (centred in the gantry), and the other holes were blocked with PMMA rods and secured with a tape to prevent movement during the experiment. The phantom was adjusted and secured to align with the lasers to ensure that the CT dose profiler was in the isocentre. Routine standard protocols for head and body (depending on the phantom in place) were selected. Scanograms of the phantoms under examination was undertaken and the helical scan was planned. The planning phase of the scan utilised the scan-measuring parameter template proposed by the CT dose profile software manual. The parameters included tube voltage, CT phantom type (either head or body), collimation (thickness of total detector area used), pitch (in helical mode), scan length, and tube rotation time. Scan speed and measuring time were also calculated automictically, after inputting the aforementioned exposure parameters. Scan types were selected to acquire the correct k-factor that allowed the CT scanner to extract all the parameters in one exposure. The total filtration was also selected. To begin each experiment, it was ensured that both the scanners and ocean software of the CT profiler had same exposure parameters and features. The measurement processing on the computer and the CT scanner were then initiated at the same time. Once the CT scan was complete, the displayed CT dose profile on the laptop and the CT console were compared. In each accuracy test, two experiments were conducted for both head and body phantoms using tube voltage values commonly used at each CT centre. The accuracy level of dose delivery values (CTDI) was estimated with equation 3.1: 83 University of Ghana http://ugspace.ug.edu.gh (D − D ) % error = 0 1 x100 (3.1) D1 where % error represents percentage of dose delivery accuracy error; D0 represents the Ocean software measured value of the dose profiler, and D1 is the console-displayed CT dose descriptor/parameter. Scanners with a deviation of less than ±20% of the measured values were considered to deliver acceptable CT dose reports on the CT console (IAEA, 2012). The experimental set up, and a CT dose profiler dose output for dose delivery accuracy are presented in Figures 3.7. and 3.8, respectively. Figure 3.7: Experimental set up for dose delivery accuracy tests. CT: computed tomography 84 University of Ghana http://ugspace.ug.edu.gh Figure 3.8: CT Dose profiler dose output for dose delivery accuracy. 3.4.3.2 CT Dose Delivery Reproducibility For dose delivery reproductivity testing, the method outlined for dose accuracy was used with the exception of comparing the results with the console dose reports. Additionally, the scans were repeated three (3) times with the same scan parameters and settings for each phantom and the consistency level of the dose values were determined with the Ocean software. The coefficient of variation was used to determine the level of reproducibility of the doses for each experiment in the three measurements via equation 3.2: SD CoV = x100 (3.2) x 85 University of Ghana http://ugspace.ug.edu.gh where CoV is the coefficient of variation of measurements, SD is the standard deviation of all measurements, and x̅ is the average of all measurements. An acceptable limit of ≤ ± 5% based on the recommendation of AAPM Report No. 35 (AAPM, 1992) was used to decide the suitability of the scanners’ dose delivery reproducibility. 3.4.3.3 Geometric Efficiency The geometric efficiency is defined as the ratio of the number of incident radiation quanta on the detector in a given interval to the number emitted by the radiation source in the same interval, expressed in percentage (Shefer et al., 2013). With respect to the geometric efficiency, free in air profiler measurements were conducted across the facilities. The CT dose profiler was adjusted so that the sensitive area was out of the phantom and positioned in the beam while the body phantom served as a holder. With the help of the lasers, the detector area of the profiler was positioned in the isocentre of the CT scanner. A scout scan was conducted for each scanner over the CT dose profiler detector. The measuring process of the Ocean software was engaged simultaneously with the CT scan. It was ensured that only the detector area was scanned free in air by avoiding the phantom holding areas in the scan. In each accuracy test, two experiments were conducted for both head and body phantoms using tube voltage values that were commonly used at each CT centre. The measuring parameter template proposed by the CT dose profile software manual and which included tube voltage, collimation, pitch, and tube rotation time, were used. Scan speed and time were also calculated automatically after inputting the aforementioned exposure parameters. Data saved in Ocean reports were downloaded from the software to determine the geometric efficiency. Scanners with a geometric efficiency >70% were considered to perform optimally good (Shefer et al., 2013). 86 University of Ghana http://ugspace.ug.edu.gh 3.4.3.4 Tube Voltage Accuracy and Half Value Layer Tube voltage accuracy test is used to test the correctness of the displayed tube voltage on the console of a CT machine (Public Health England, 2016). This is achieved by experimentally measuring and comparing the true values with the displayed values using the same equipment. The RTI MPD (RTI Electronics, Mölndal, Sweden) connected with a laptop computer with Ocean Software Interface (RTI Electronics, Mölndal, Sweden) was used for kVp accuracy testing. At each CT scanner, the RTI MPD was placed at the bottom rim of the inner portion of the CT gantry directly opposite the x-ray tube. This was achieved by ensuring that the vertical lasers coincided at the middle of the RTI MPD detector mark. A position check was then conducted to ensure that the detector was positioned in the middle of the beam and the lasers were correctly positioned. A topogram scan with the tube positioned at 12 o’clock was undertaken and the right position was ascertained with the Ocean software prior to the commencement of the scans. Subsequently, topogram scans were obtained for each scanner using the two commonly used tube voltage factors for head and body procedures. The results of the tube voltage accuracy measurements were generated by Ocean software for analysis. For each testing, the RTI MPD detector, together with the Ocean software provided additional data of the HVL (in total filtration of aluminium) value in a single exposure. For a specified voltage, a measurement of the HVL provides data on the total filtration (specified quality) in the x-ray beam. Low filtration provides unnecessary radiation dose to the patient while higher filtration leads to beam hardening (AL-Jasim et al., 2017). The RTI Group indicates that HVL results generated by Ocean software with a single exposure is very comparable to the standard test where 5-mm thickness of aluminium filters are used (RTI Group, 2016). The mathematical equation used in the kVp accuracy/error calculation is given by equation 3.3: 87 University of Ghana http://ugspace.ug.edu.gh V  % kV error=  o −Vs 100 (3.3) p   V  s  where V0 and Vs represent the measured and set voltages, respectively. The acceptable limit of ≤ ±5% of nominal values was used to assess the CT scanners, while total filtration ≥ 2.5 mm Al for tube voltages ˃100 kV was used as the acceptable HVL limit (IAEA, 2012). 3.4.3.5 CT Number (Water), Image Noise Testing and Homogeneity A water filled phantom (PTW, Freighburg, Germany), (Figure 3.9), was utilised for the CT number (water), homogeneity and image noise testing. It was made up of a thin uniform plastic container filled with distilled water. Figure 3.9: Water-filled phantom for CT number, homogeneity and image noise testing. During the testing procedure, the phantom was centred in the tomographic plane of each scanner to ensure that the middle of the phantom was in the isocentre. A routine body scan parameter was used to scan the phantom in each facility. In the absence of any image artefacts, a centrally located image slice of each scan was selected for analysis. The QC image analysis tool - ImageQC v.1.43 (EllenWasbo, 2018; Stavanger, Norway), was utilised in the analysis (ImageQC, 88 University of Ghana http://ugspace.ug.edu.gh 2017). In the software, CT number and image noise were measured in a centrally placed circular ROI (region of interest) of recommended diameter of 40% of the phantom image (ImageQC, 2017). Practically, the CT number was determined at the measured ROI mean value and the noise was determined at the ROI standard deviation as suggested by IAEA (IAEA, 2012). Mathematically, the CT number (Hs) of a sample of material (s) is defined by the equation 3.4:  s (E)− w (E) H   (3.4) s =  K    w (E)  where: µ𝑠 (E) and µ𝑤(E) are the linear attenuation coefficients at the energy (E) of the x-ray beam for the scanned sample and water, respectively and K is a constant (K=1000) if the CT value scale is in HU as in the case of all modern CT units (IAEA, 2012). From equation 3.4, the CT number for water is zero and, since the attenuation is negligible for air, the CT number for air is –1000. Therefore, scanners that recorded CT water number of ≤ ± 4 HU and noise derivation ≤ ± 10% of baseline values were considered to operate within the acceptable limits (IAEA, 2012). Figure 3.10 shows how the CT water number and image noise were analysed in the ImageQC software version 1.43. Arrows A and B in Figure 3.10 show how the results of CT number (water) and image noise are displayed in the software. 89 University of Ghana http://ugspace.ug.edu.gh Figure 3.10 : CT number (arrow A), image noise (arrow B) analysis using ImageQC v.1.43. In the case of the homogeneity (uniformity) testing, five ROIs of 10 mm each were measured at 12 o'clock, 15 o'clock, 18 o’clock, 21 o'clock and centre of the selected image slice. The uniformity was measured as the absolute difference of CT numbers between a centrally placed ROI with each of four ROIs placed on the edge. The ImageQC software v.1.43 was used in the homogeneity analyses. Each of these four values and their deviations from the central value were compared with the given tolerance of ±10 HU for head and body (IAEA, 2012). Figure 3.11 shows how the CT uniformity/homogeneity analysis was performed using the ImageQC v.1.43. 90 University of Ghana http://ugspace.ug.edu.gh Figure 3.11: Uniformity/homogeneity analysis using ImageQC v.1.43. 91 University of Ghana http://ugspace.ug.edu.gh 3.5 Phase 3 Study: CT Dose Data and Image Quality Assessment 3.5.1 Outline The Phase 3 of the study collected dose descriptor data and image quality information on patients’ examinations. A prospective data collection process was used following the QC assessments to ensure that only CT quantity data generated after the QC tests were included. This ensured that the data (e.g. CTDIvol and DLP) displayed by the CT console for collection were valid. Details of the procedure for Phase 3 of the study are presented in Sections 3.5.2 to 3.5.7. 3.5.2 Study Population ICRP Publication 135 recommends setting DRL values based on wide scale surveys of the suitable DRL quantities for examinations of a representative group of patients in an agreed weight range (ICRP, 2017). According to the Public Health England (2016), to ensure representative results for NDRLs, a successful national survey needs a timely gathering of important scan data from a robust sample of key procedures covering common CT practice. In line with the scope of this study, and as justified in Chapter One, the study population consisted of adults (18 years and above). Adults were used for this DRLs development because they accounted for the highest number of CT examinations in Ghana (Inkoom et al., 2014). Since the attenuation of the x-ray beam depends on the tissue penetration depth, standardisation of patient size (weight restriction) was needed in DRL development (ICRP, 2017). A study (Shirazu et al., 2017b) suggested that the commonest weight range of average Ghanaians was 48.6 - 90.0 kg. Therefore, a population of weight range of 50–90 kg was chosen as the reference weight for the DRLs. This was in line with the ICRP’s recommendations (ICRP, 2017). 92 University of Ghana http://ugspace.ug.edu.gh 3.5.3 Type of Data Set Collected The ICRP has suggested that data collected for CT dose quantity surveys should, when possible, involve the equipment manufacturer/model, procedure name, patient weight and DRL quantities (e.g. CTDIvol, DLP) (ICRP, 2017). The data must also include relevant scan parameters, if appropriate and available, for the examination types being surveyed (ICRP, 2017). In this study, the data collected in Phase 3 included equipment manufacturer/model, procedure name, patient weight, gender, age, CTDIvol and DLP. Where multiple scan sequences were undertaken, total DLP per indication were recorded. Other scan parameters collected included tube voltages (kVp) and tube loading (mAs), pitch, rotation time, scan thickness, number of series, number of slices per scan series, scan length, and the type of CT phantom used in estimating the dose quantities. Scanning mode, AEC application and contrast usage were also collected. A study (Vock and Frija, 2016) has also argued that data collection for indication-based DRLs should take into consideration image quality, since the greatest priority for any diagnostic imaging procedure is producing image quality satisfactory for the clinical purpose. Consequently, for each patient image, one selected slice (at the centre of a set of images) was collected alongside the aforementioned parameters for objective image quality analysis (Section 3.5.6). Subjective image quality information was also obtained prior to the data collection, by requesting the scanning radiographers’ comments on whether or not the acquired images in a particular CT folder/unit were diagnostically acceptable, not repeated, and had been accepted and reported by a radiologist. Practically, images accepted and reported by radiologists were considered as being of sufficient diagnostic quality to their need for the clinical purpose. Therefore, those images that the radiologists had reported without any image quality issues were considered “good data” sets for the study. This approach was used as the data sets were very large. 93 University of Ghana http://ugspace.ug.edu.gh 3.5.4 Patients’ Data Size It has been recommended in the literature (ICRP, 2017; Vock and Frija, 2016; Public Health England, 2016; IAEA, 2013) that it is very appropriate to collect data of 20 patients per indication in every selected CT centre for the development of NDRL. Accordingly, at every CT unit, 20 patients who had undertaken CT procedures for the same indication were selected at random. For indications such as CVA/stroke, head injury/trauma and suspicion of brain tumour/SOL, 20 data sets were collected from all the 25 CT units that passed the QC tests. However, for indications such as suspicion of lung cancer/tumour and CT lung lesion with CKD, 23 of the CT scanners undertook such examinations, and hence such data were not taken from the other two CT units. With regards to suspicion of abdomino-pelvic lesion and kidney stones, data were acquired from 24 CT units as such procedures were not undertaken at one CT unit. Nineteen and 10 CT units undertook examinations for suspicion of urothelial malignancy (CT-IVU) and PE, respectively. Therefore, 20 sets of data were selected from each unit for each indication, in order to provide better suggestions of typical practice at each CT centre. In total, 3,960 data sets were collected for the development of the indication-based DRLs. 3.5.5 Data Acquisition Tool and Process This section describes in detail how the data sets mentioned in section 3.5.3 and 3.5.4 were collected. In Ghana, patient dose information recording and automatic dose tracking systems were not mandatory and readily available in most facilities at the time of the study. Data collection forms 1-9 (Appendix VII) were designed for the CT quantity data collection process using a similar validation process as described in Section 3.3.4. The scanning parameters and quantity data that were considered “good data” sets were received from the CT Picture Archiving and Communication System (PACS), workstations and Radiation Dose Structured Reports (RDSRs). 94 University of Ghana http://ugspace.ug.edu.gh A 2-TB (Terabyte) hard-drive was used to store the soft copies of images, one each of selected examination for subjective analysis. The data collection process used in this Phase of the study was first tested and refined in a pilot survey conducted at two hospitals in Accra. This process effectively managed the study, including protocol assortment to suitability, ease of usage of the data collection forms, and data recording. Following the pilot study, the researcher visited all the 25 facilities for the data. The initial data were entered into Microsoft Excel spreadsheet version 2013 and errors checked. The data was further analysed for statistical control (detailed in section 3.5.7) using a Microsoft Excel version 2013 and Statistical Package for the Social Sciences (SPSS) version 23.0 (SPSS Inc., Chicago, IL USA) prior to the DRL estimation. The data collection process in this Phase of the study took a year to complete. 3.5.6 Image Quality Assessment of the Collected Image Data The quality of all the various images (of which CT descriptors were used in DRLs) was assessed prior to using the data for the DRLs. The image quality was used as a QC measure of a scanner’s performance and to optimise practice. For effective development of DRLs, it was needful to ensure that all the images, whose quantity data or parameters were used, were within the acceptable diagnostic limit. Images of poor quality could lead to diagnostic challenges. As indicated earlier (Section 3.5.3), due to the large data sets involved, radiographers at the CT units of the study sites provided subjective image quality remarks on all the collected imaging folders prior to data collection. Objectively, the images were again assessed for their signal to noise ratio (SNR) properties. Though the contrast to noise ratio (CNR) and SNR are currently the mainly used objective image quality parameters in CT imaging, there is incoherent limitation concerning the use of CNR (Lu 95 University of Ghana http://ugspace.ug.edu.gh and Nishikawa, 2012). This is because at low-contrast signals, the CNR does not account for background noise correlations across different types of reconstruction algorithms (Lu and Nishikawa, 2012). The SNR, on the other hand, associates well with objective analysis (Lu and Nishikawa, 2012). Mathematically, the SNR for the images at specific region of interests (ROI) were estimated using equation 3.5: Mean signal value witnin ROI SNR = (3.5) SD within ROI where the symbols bear their defined meanings and SD is standard deviation. An ImageJ software programme version 1.52 manufactured by the National Institute of Health, USA, which is reliable for objective image quality assessment (Ferreira and Rasband, 2012) was used to mathematically predict the mean signal and SD values within a given ROI. The mean signal and SD values were inputted in equation 3.5 to determine the SNR. Figure 3.12 shows how the ROI was structured in each of the anatomical region. A B C Figure 3.12: Positioning of ROI in chest (A), abdomen and pelvis (B) and head (C) examinations. 96 University of Ghana http://ugspace.ug.edu.gh The Rose criterion (model), which states that an SNR of at least 5 (≥ 5) is required to distinguish image features at 100% certainty (Bushberg et al., 2011), was used to evaluate the suitability of the SNR in the images. 3.5.7 Statistical Control of Collected Dose Descriptors In the data collection process, an unexpected outcome could happen as a result of errors (Neuburger et al., 2017). Particularly, during the collection and entering of large data sets such as the CT dose descriptors used in this study. Quality assessments in relation to both raw data and their analyses, represented an important feature of data management for the national CT survey to reinforce confidence in reported results (Public Health England, 2016). In addition to the CT QC tests performed in the Phase 2 study to assess the performances of the CT scanners, statistical control analyses were also used to assess the internal QC of the raw data. Internal data QC involves a continuous, critical evaluation of the data collection process or acquired data and working routines using a statistical method to mitigate errors (Neuburger et al., 2017). To ensure reliability in the data entering consistency, the first 20 data sets were re-evaluated, and data re-entered and compared with the original entries using SPSS version 23.0 (SPSS Inc., Chicago, IL USA). A successful data entering consistency was assumed when discrepancies between the initial and follow up entries were < 2% (Pedersen et al., 2018). Subsequently, statistical control charts (Shewharts) (Neuburger et al., 2017) were used to analyse performance through data collection time and also ensured no problems with the data entering process. To generate the control chart, the mean, the standard deviation, as well as the upper and lower control limit values of the critical dose descriptors such as CTDIvol and DLP were calculated. 97 University of Ghana http://ugspace.ug.edu.gh The upper control limit (UCL) was derived from the equation 3.6: 𝑈𝐶𝐿 = 𝑀 + (3 x SD) (3.6) while the lower control limit (LCL) was calculated from equation 3.7: 𝐿𝐶𝐿 = 𝑀 − (3 x SD) (3.7) where: M and SD are the mean and standard deviation, respectively. The mean UCL and LCL were plotted using Microsoft Excel to generate the X-charts for quantity data sets. Data points above the UCL and below LCL were double-checked to ensure that the true data sets were entered. All wrong entries were then corrected. The charts are shown in Appendix VIII and indicate a good final control X-charts for all indications, which made the data suitable for DRL estimation. Once the statistical control process was over and image qualities were acceptable, the CT quantity data values for all the various indications were transferred to SPSS version 23.0 (SPSS Inc., Chicago, IL USA) for DRLs estimation. 98 University of Ghana http://ugspace.ug.edu.gh 3.6 Phase 4 Study: DRL Values Estimation and Dose Monitoring Tool 3.6.1 Outline The procedures used in estimating the DRL values and developing a tool for dose monitoring are discussed in Sections 3.6.2 and 3.6.3. 3.6.2 DRL Values Estimation The ICRP (2017) has recommended that NDRL values ought to be established at the 75th percentile of median values obtained in a sample of representative facilities. The median value has been recommended over the mean values in the ICRP Publication 135 for DRL estimation, as it is recognised to be more robust in DRLs. Accordingly, SPSS version 23.0 (SPSS Inc., Chicago, IL USA) was used to analyse the data. For each indication, the median, minimum, maximum, and standard deviations were estimated. Moreover, quartiles (25th, 50th and 75th percentiles) were estimated. The 50th percentile was referred to as the achievable dose (AD). Microsoft Excel Version 2013 was used to produce bar charts for visual presentation of frequency distribution for various key dose descriptors in the survey. The developed DRLs were compared with previously established NDRLs reported in literature. The results are presented in Section 4.4 of Chapter Four. 3.6.3 Tool for Dose Monitoring For the proposed indication-based DRLs to be well established in Ghana, a simple Microsoft Excel-based tool (also called BOTB in this study) was developed for its application in periodic inspection and monitoring of DRLs compliance purposes. It was based on the assumption that there would be a need for compliance assessment across the facilities once the DRLs are 99 University of Ghana http://ugspace.ug.edu.gh established by the appropriate authority. However, without an easy-to-use tool for assessment, the compliance assessment exercise could be limited. On a Microsoft Excel page, the proposed DRLs (CTDIvol and DLP) values were used as benchmarks and formatting rules were set. Columns and rows (entering space) were provided for demographics and dose descriptors/quantity data. Conditions in terms of weight and examination sequences were also set based on the study’s results. Excel formulas were developed using equation functions, and conditional and formatting tools. Also, colours were used to communicate results output. The Excel interphase (Figure 4.24; for each indication) included a results box that shows the DRLs, and a results interpretation/action section that gives interpretation to the results. When a sample of patients’ data is entered into the entering space, the functions calculate the DRLs and equate them to that of the national values automatically. When the calculated values are above the NDRLs, the colour appears red; when it turns black it indicates an acceptable level; however when it turns blue, it indicates that the facility’s doses are below the 25th percentile and hence, image quality tests are required to ensure that examinations of acceptable diagnostic quality are produced. 100 University of Ghana http://ugspace.ug.edu.gh 3.7 Phase 5 Study: Dose Impact on Patients and Cancer Risk 3.7.1 Outline The procedures used in studying the dose impact and possible cancer risk of CT imaging of the various indications on patients are presented in Sections 3.7.2 and 3.7.3. 3.7.2 Effective Dose To understand the radiation dose impact associated with each of the indications and propose appropriate optimisation methods, effective doses were first estimated for each indication using the equation 2.8 expressed in Chapter Two. The k-factors suggested by the ICRP Publication 103 (ICRP, 2007) were used in the estimation of the effective dose for each indication with respect to the anatomical region. The k-factors (conversion factors for changing DLP to effective dose) for the respective anatomical regions are presented in Table 3.1. Table 3.1: ICRP Publication 103 recommended region-specific DLP-ED conversion factors Anatomical region DLP to ED coefficients (mSv /mGy cm) Head 0.0023 Chest 0.0170 Pelvis 0.0190 Abdomen 0.0153 Abdomen-Pelvis 0.0150 DLP; dose length product, ED; effective dose; ICRP; International Commission on Radiological Protection To better understand the magnitude of the doses, the average effective doses for the indications were then compared to the global average natural background radiation of approximately 2.4 mSv (Canadian Nuclear Safety Commission, 2019). The results are presented in Table 4.51. 101 University of Ghana http://ugspace.ug.edu.gh 3.7.3 Cancer Risk To understand the cancer risk associated with the observed radiation doses across the CT facilities, the BEIR VII model was applied to predict the Lifetime Attributable Risk (LAR) of cancer based on the magnitude of a single radiation exposure, patient’s age and gender (National Research Council, 2006). The LAR is defined as additional cancer risk above and beyond baseline cancer risk (Lee et al., 2012). The model was developed based on the extensive studies on the survivors of Hiroshima and Nagasaki atomic bombs and medical, occupational and environmental radiation studies (National Research Council, 2006). Theoretically, the model is grounded on the linear no-threshold model (LNT) concept which is centred on the assumption that the smallest dose has the potential to cause a small increase in radiation risk to humans (National Research Council, 2006). LAR data as presented in Tables 2.1 and 2.2, respectively, were utilised to estimate the cancer risks. Using the age of exposure and gender parameters, the LAR of cancer incidence (LARi) and cancer mortality (LARm) from organ doses were subsequently extrapolated for patients (age 20, 40 and 60 years) by BEIR VII models as presented in proportions via equations (3.8) and (3.9) below. Dorg LAR𝑖 = [( Gy) LAR𝑖𝑓] in 100,000 patients (3.8) 0.1 Dorg LAR𝑚 = [( Gy) LAR𝑚𝑓] in 100,000 patients (3.9) 0.1 102 University of Ghana http://ugspace.ug.edu.gh where LARi and LARm represent the lifetime attributable risk of cancer incidence and mortality, respectively. The LARif and LARmf represent the BEIR VII organ-specific cancer incidence and mortality coefficients, normalised to age and gender, and Dorg is the organ dose in gray (Gy). The organ doses associated with each of the indication-based CT imaging examinations were estimated and used in the above cancer risk equations (3.8 and 3.9). To predict organ doses in the human body subjected to CT x-ray photons, two different methods are plausible. These include experimental measurement and computer simulation (Lee et al., 2012). In the case of the experimental dose measurements, physical phantoms (mostly) with capabilities to receive organ site-specific types of dosimeters are scanned with CT (Lee et al., 2011). However, this procedure is complex and time-consuming, and physical phantoms provide inadequate sampling of the disparities of age, gender, and body morphometries. Moreover, the procedure cannot, in some cases, accurately represent the average organ dose when high dose gradients exist, and also when the distributed body tissues (e.g. active bone marrow) are of concern (Lee et al., 2012; Lee et al., 2011). Contrarily, computer simulation software programmes are good option when estimating organ doses across many facilities (Lee et al., 2012) as was the case in Ghana for each of the indications. The National Cancer Institute dosimetry system for CT (NCICT) (software version 2.1) was used in the organ dose estimations (Lee et al., 2015). The software is based on a comprehensive library of computational human phantoms (surrogate anatomy for patients) combined with Monte Carlo radiation simulation of reference CT scanners (Lee et al., 2015). It has been reported to be the most current, sophisticated and reliable way to obtain accurate values of organ dose under CT imaging (Lee et al., 2015; Lee et al., 2011). Although there are many available computer simulation software systems such as CT Dosimetry, CT-Expo, CTDOSE, 103 University of Ghana http://ugspace.ug.edu.gh RadimetricsTM for estimating organ doses, the computational phantoms within some of these existing software tools are limited to mathematical stylised phantoms which practically, may be different from the reference anatomy recommended by the ICRP (Lee et al., 2015). Moreover, most of these computer simulation systems are relatively old and do not have all the current CT scanner models in the software for dose estimations. Some studies (Lee et al., 2012; Liu et al., 2010) have reported some significant discrepancies in radiation doses between the actual and the estimated values when the correct scanner model and information are not selected in the software for estimation. However, the NCICT is known to provide a more realistic anatomy based on the ICRP reference phantoms and the most up-to-date bone marrow dosimetry with several convenient features compared to previous tools (Lee et al., 2015), hence, its usage in this study. In this study, the CTDIvol values were displayed by all the scanners. The organ doses were normalised through the CTDIvol and patient- and scan-specific parameters by an algorithm. Mathematically, the algorithm is expressed in equation 3.10: D(organ,age, gender,spectrum)= Z=SEZ=SS DC(organ,age, gender,spectrum,Z )xCTDI vol (3.10) where D (organ, age, gender, spectrum) is the absorbed dose (mGy) for the organ of interest, the age and gender of a given scanned patient with an x-ray spectrum; SS and SE are the respective Scan Start and Scan End distances (cm) from the top of the head (cranially) in patients; DC (organ, age, gender, spectrum, Z) is the 5D matrix of organ dose coefficient (DC) (mGy/mGy) per 1-cm axial slice: organ dose (mGy) normalised to the CTDIvol (mGy) of the reference CT scanner used for the Monte Carlo calculations. Organ is the organ of 104 University of Ghana http://ugspace.ug.edu.gh interest, age and gender are derived from a given patient, and spectrum is one of the six combinations of four tube potentials (80, 100, 120, and 140 kVp) and two filtrations (head and body) of a particular CT scan, and Z is the slice number ranging from the top of the head to the bottom of the patient’s feet; CTDIvol is from the particular scan for which organ doses are calculated (Lee et al., 2015). Through the graphical user interface (GUI), the relevant parameters were entered into the software. Figure 3.13 shows a NCICT software GUI which allows organ doses to be obtained based on the entered CTDIvol and patient- and scan-specific parameters. Figure 3.13: Graphical user interface (GUI) of the NCICT program showing an example of entered patient- and scan-specific parameters and estimated organ dose calculation for adult male head (brain tumour/SOL) scan. 105 University of Ghana http://ugspace.ug.edu.gh Using the equations (3.8 and 3.9) and the estimated organ doses, the LARi and LARm for each organ were also estimated. The results of the organ dose and the LARi and LARm values are presented in Tables 4.52-4.56. The organs, whose doses were estimated, have been presented in Table 3.2. Table 3.2: Indication-related organ whose doses were estimated Anatomical region Indication Organs CVA, brain tumour, head injury/ trauma Brain Pituitary gland Lens Eyeballs Head Salivary glands Oral cavity Spinal cord Thyroid Lung tumour/cancer, Chest lesion with Thyroid CKD, PE Oesophagus Trachea Chest Thymus Lungs Breast Heart wall Abdomino-pelvic lesion, kidney stones, Stomach wall urothelial malignancy (CT-IVU) Liver Gall bladder Adrenals Spleen Pancreas Kidney Abdomino-pelvic Small intestine Colon Rectosigmoid Urinary bladder Prostate Uterus Testes Ovaries Skin Muscle Active marrow Shallow marrow Key: CVA= cerebrovascular accident, CKD =chronic kidney disease, PE= pulmonary embolism, CT-IVU =computed tomography intravenous urography. 106 University of Ghana http://ugspace.ug.edu.gh 3.8 Phase 6 Study: Optimisation The procedures used in developing CT optimisation methods are outlined in Sections 3.8.1- 3.8.2.5. 3.8.1 Optimisation Method 1: Management of Scan Length Optimisation, through the management of scan lengths utilised in the facilities for CT imaging, was explored to improve the radiation protection of patients. The additional scan distances (along the z-axis) covered above (DAUT) and below (DBLT) the organ or anatomical region of interest for each indication was evaluated for optimisation options. The concept of evaluating the DAUT and DBLT is that, CT scanning exactly to the ends of the structure/organ of interest using a planner scanogram may lead to a cutting off of some essential part of the organ of interest. To prevent this and the associated risk of repeats and reporting uncertainties, there is a need to adequately cover the area/organ under examination by scanning slightly above or below the interested organ/region. The criteria for characterising image coverage in CT have been defined and agreed upon in Europe for some specific adult and paediatric examinations (European Commission, 1999). However, these criteria are now over 23 years and do not reflect all the common indications in CT. Moreover, these criteria are mostly anatomy based. The European Commission (1999), in particular, states that the volume of investigation for head procedures should cover from the foramen magnum to the skull vertex. However, it does not provide recommendations for all specific indications of the head as they vary. In Phase 1, most of the CT facilities in Ghana lacked documented scan protocols for all the indications and the observed imaging requirements were numerically vague on the acceptable DAUT and DBLT. The lack of standardisation could lead to a great variation in practice and associated radiation doses. 107 University of Ghana http://ugspace.ug.edu.gh In evaluating extra scan coverages, all patients’ folders containing image information used for developing the DRLs were selected with their identification numbers and the images reconstructed. CT calibrated callipers at the CT workstations were used to undertake the DAUT and DBLT measurements, after reconstructing the acquired image slices into grid lines on scan planners. A template sheet used to collect the DAUT and DBLT data is presented in Appendix IX. To ensure accuracy in the measurements, the scanned PMMA CT head phantom length (of known length of 150 mm) used in the QC assessment was measured on the CT workstations. The measured length on the CT workstations was compared with original length for accuracy prior to taking the main measurement data. Each measurement was conducted twice, and the average values were used. Diagrams showing how the measurements of DAUT and DBLT were undertaking for each indication are shown in Figure 3.14. The DAUT measurements for head indications were measured from the skull vertex to the last slice line above the head (cranially) (Line D) and for the DBLT it was measured from base of skull (at foramen magnum) down the cervical to the last slice line caudally (Line F). For all chest indications, the DAUT was measured from the lung apex (first thoracic vertebra) to the last slice line cranially (Line H) and for the DBLT, it was measured from the lowest costophrenic angle to the last slice line caudally (Line I). Regarding abdomino-pelvic lesions/tumours, the DAUT was measured from the dome of diaphragm to the last slice line cranially (Line A). For the DBLT, it was measured from the symphysis pubis to the last slice line caudally (Line B). In the case of kidney stone and CT-IVU for urothelial malignancy, the DAUT was measured from 1 cm above the most cranial pole of the kidneys to where the last slice line ended upwards (Line C) and for the DBLT, it was measured from the symphysis pubis to the last slice line caudally (Line B). 108 University of Ghana http://ugspace.ug.edu.gh Figure 3.14: The measurements of DAUT and DBLT for each indication 109 University of Ghana http://ugspace.ug.edu.gh A Microsoft excel spreadsheet version 2013 was used to generate mean and standard deviation values for DAUT and DBUT that were used across the country for all the indications and the results presented in Table 4.57 of Chapter Four. To recommend an appropriate standard protocol to optimise and harmonise the extra scan length values and reduce unnecessary radiation for clinical application in Ghana, an experimental study using routine CVA/stroke indication as a case study was then undertaken. This CVA/stroke indication was selected because it was the most observed indication in this study. 3.8.1.1 Phantom-Based Optimisation Study An anthropomorphic Alderson RANDO phantom was used to investigate the desirable scan coverage for CT CVA/stroke indication. The details are presented in Sections 3.8.1.1.1-3.8.1.1.4. 3.8.1.1.1 Materials A Siemens CT Somatom Emotion scanner and an anthropomorphic Alderson RANDO phantom, which consists of a human skeleton embedded in plastic, were used in the experiment. The phantom has equivalent radiological properties to human tissues. Its physical characteristics were similar to an average male patient of weight, 73.5 kg, and height, 175 cm tall and without arms or legs. The phantom consists of 2.5 cm-thick plastic slices with pre- drilled holes into which TLDs can be inserted, and segmented into sections numbered from 0 at the cranial end to 35 at the caudal end. An ImageJ software version 1.52 and two radiologists were used to assess the image quality while Microsoft Excel version 2013 was used for data entering and processing for analysis. A diagram showing the anthropomorphic Alderson RANDO phantom and how it was positioned in a Siemens CT Somatom Emotion scanner for the optimisation study is presented in Figure 3.15. 110 University of Ghana http://ugspace.ug.edu.gh Figure 3.15: An anthropomorphic Alderson RANDO phantom as shown in Plate A, while Plate B shows the position of the phantom in a Siemens CT Somatom Emotion scanner. 3.8.1.1.2 Procedure First, the standard protocol for routine CVA procedure was used to scan the phantom using a Siemens Somatom Emotion positioned for CVA procedure. A total of six different scan coverages (scan lengths) were used to acquire different sets of images. The imaging parameters used are presented in Table 3.3. Table 3.3: Parameters used for performing the experimental study for scan coverage for routine CVA Protocol/ kVp Eff. Pitch Trot(s) ST Coll. SL DAUT DBLT image ID mAs (mm) (mm) (mm) (mm) A 110 120 0.55 1.5 3 16 x 0.6 165 ≈8 ≈20 B 110 120 0.55 1.5 3 16 x 0.6 137 0 0 C 110 120 0.55 1.5 3 16 x 0.6 132 10* 0 D 110 120 0.55 1.5 3 16 x 0.6 157 10 10 E 110 120 0.55 1.5 3 16 x 0.6 142 0 5 F 110 120 0.55 1.5 3 16 x 0.6 147 5 5 Key: * = below the vertex, kVp: peak kilovoltage, Eff. mAs: effective milliampere-second, Trot(s): rotation time, ST: scan thickness, Coll.: collimation, SL: Scan length, DAUT: distance above upper target, DBLT: distance below lower target. CVA: cerebrovascular accident. Note: upper and lower targets for CVA imaging were the skull vertex and foramen magnum, respectively. 111 University of Ghana http://ugspace.ug.edu.gh For study A, the scan coverage employed the mean DAUT and DBLT values used across the facilities to scan the phantom after a scanogram. With all factors remaining the same, the DAUT and DBLT values were adjusted to study the dose and image quality response. The second measurement (B) was undertaken by adjusting the scan length from exactly the vertex of the skull down to exactly the level of the foramen magnum at the base of the skull. The third measurement (C) was also undertaken 10 mm below the skull vertex and 0 mm below the base of the skull at the foramen magnum. The other measurements were taken; 10 mm above vertex and 10 mm below the base of the skull (D); no extra scan lengths above vertex and 5 mm extra scan lengths below the base of the skull (E), and 5 mm extra scan length above vertex and below the base of the skull (F). Six different scan coverages (scan lengths) were used to acquire different sets of images and, subsequently, the CTDIvol, DLP values were obtained from the CT scanner’s console. The scan coverages used in scanning the Alderson RANDO phantom are shown in Figure 3.16. Figure 3.16: Scan coverages used in scanning the Alderson RANDO phantom. 112 University of Ghana http://ugspace.ug.edu.gh 3.8.1.1.3 Image Quality Assessment Image quality in terms of SNR was calculated using equation (3.5) and the procedures described in Section 3.3.3.6 using ImageJ software version 1.52. Moreover, the images of each set of procedures were given to two senior radiologists to estimate the suitability and subjective image quality of all the suggested protocols for their intended tasks in clinical practice in Ghana. The two radiologists had 9-years and 5-years working experience in image reporting, respectively. A 5-point scale (1= poor to 5= very good) was used to rate the images on: coverage of the region of interest, image contrast resolution, acceptable spatial resolution, diagnostic image acceptability, visualization of anatomical structures of interest and image noise. 3.8.1.1.4 Data Analysis An SPSS-based Intraclass Correlation Coefficients (ICC) was run to determine statistical agreement between the two raters in image quality scoring. A p-value of ≤ 0.05 was considered statistically significant in the ICC tests. The dose descriptor data for all the six protocols were descriptively compared using Microsoft Excel version 2013 and percentages of DLP dose reduction over the current practice (originally used across the facilities) were estimated. The results are presented in Table 4.58 of Chapter Four. 113 University of Ghana http://ugspace.ug.edu.gh 3.8.1.2 Patient-Based Optimisation Study As the phantom could not absolutely mimic a real CVA condition, a clinical study using patients’ data was conducted to evaluate the application of the phantom results in clinical practice. The average scan coverage used in the facilities was referred to as “Full range” while the best scan coverage protocol (protocol C) observed in the phantom study was referred to as “Reduced range”. Details are presented in Sections 3.8.1.2.1 to 3.8.1.2.7. 3.8.1.2.1 Materials Five CT scanners, including Toshiba Aquilion ONE, Philips Brilliance Extended, GE Optima 660, Siemens, Somatom Emotion and Hitachi Spuria together with their workstations with details presented in Table 3.4, were used in the patient-based study. These scanners were used because there was the need to have a representation of all the scanner models across the country in this study. Table 3.4: CT equipment and scanning parameters Scan parameters CT scanners Pitch Tube voltage Tube loading Beam collimation (kVp) (mAs) (mm) Toshiba Aquilion ONE 0.75 120 225 32. x 0.5 Philips Brilliance Extended 0.75 120 400 64 x 0.5 GE Optima 660 0.60 120 180 16 x 1.0 Siemens Somatom Emotion 0.55 130 250 6 x 1.0 Hitachi Spuria 0.75 120 263 16 x 1.25 kVp: peak kilovoltage, mAs: milliampere-second, CT: computed tomography 114 University of Ghana http://ugspace.ug.edu.gh Other materials and people engaged in this study included: I. PACS connected to each of the aforementioned scanners were also used to retrieve patients’ clinical images. II. NCICT software version 2.1 (A Monte Carlo-based software developed by the National Cancer Institute, USA). III. ImageJ software version 1.52 (National Institute of Health, USA). IV. Two radiologists V. A Microsoft Excel version 2013. 3.8.1.2.2 Sample Size Determination A sample size calculator was first used to determine the required sample size capable of producing statistically reliable responses when the “Full range” (used in the facilities) and “Reduced range” CT intervention values are compared. Charan and Biswas (2013) recommended equation (3.11) as a suitable means of estimating the required sample size for interventional studies involving two groups as the case for Full range CT and the “Reduced range CT” scans. 2D2ev ( 2 Z / 2 + Z ) S = 2 d (3.11) where S is the sample size, Dev is standard deviation from previous studies of pilot study, 𝑍 𝛼 is standard normal variate for level of significance (p < 0.05). 2 𝑍ᵦ is standard normal variate for statistical power 𝑑2 is the effect size. 115 University of Ghana http://ugspace.ug.edu.gh At 5% type 1 error (p < 0.05), the 𝑍𝛼 was chosen as 1.96 since the majority of studies of this 2 kind consider p-values significant below 0.05. From the DAUT and DBLT, the collective SD for CVA was 21.3 and this was used in the formula. The precision/absolute error of 5% was tolerable and the 𝑍ᵦ was chosen at 80% (0.84) as it is very reliable (Charan and Biswas, 2013). From the phantom study, effect size of 10% was considered an appreciable dose reduction. Hence, the sample size was calculated as: 2 2(23.1)2 (1.96 + 0.84) S = = 83.67 102 From the above, the least sample size that could be used for each group (“Full range” and “Reduced range” CT) was approximately 84. However, to account for a better statistical outcome, a sample size of 100 was used for each study group. 3.8.1.2.3 Inclusion and Exclusion Criteria With the exception of non-CVA indicated CT procedures and patients under 18 years of age, all patients’ images which had been undertaken for the evaluation of CVA within the study period (2017- 2019) in the facilities indicated in Table 3.4 were considered eligible for their inclusion in this study. 3.8.1.2.4 Procedure Two groups of data, consisting of “Full range” and “Reduced range” were first created from the PACS and the CT workstations of the CT scanners. The data set in each group comprised the DAUT and DBLT. In the “Full range” group, the DAUT and DBLT of 100 original images were collected from all the five scanner facilities. The mean DAUT was 8.3±7.1 mm above the vertex and the mean DBLT was 20.2 ±11.3 mm. Subsequently, “Reduced range” data variables covering from 10 mm below the vertex and 0 mm the below 116 University of Ghana http://ugspace.ug.edu.gh base of the skull at foramen magnum were reconstructed or reformatted from the original images using CT reconstruction tools. Each image set had both full range and reduced range values. The mean DAUT and DBLT parameters and pictorial representation of the scan coverages for full and reduced ranges are presented in Table 3.5, and Figures 3.17 to 3.19. Table 3.5: The DAUT and DBLT used in the patient-based study Protocol Average coverage of skull for CVA Average DAUT (mm) DBLT (mm) CTDIvol(mGy) Full 8.3±7.1 mm above the 20.2±11.3 mm below the base 54.6 range CT vertex of skull (foramen magnum) Reduced 10* mm or 1 cm below or 0 mm below the base of the 54.6 range CT inferior to the vertex skull at (foramen magnum) DAUT = distance above upper target, DBLT = distance below lower target. * = below vertex, CVA = cerebrovascular accident, CT = computed tomography; CTDIvol = volume weighted CT Dose index Figure 3.17: A typical scan coverage (grid areas) for a “Full range” CVA CT. Plate A shows the last two slices from the base of the skull (caudally) and Plate B shows the last two slices cranially which contain no information for CVA diagnosis. The red arrow shows a typical mean DAUT and the black arrow shows a typical mean DBLT used across the facilities. 117 University of Ghana http://ugspace.ug.edu.gh Figure 3.18: A typical scan coverage (grid areas) for a “Reduced range” CT for CVA. Plate I shows the last two slices from the skull base (caudally) and Plate II shows the last two slices cranially. The red arrow shows a 1 cm allowable distance from the vertex of the skull that can be used to optimise radiation in CVA examinations without compromising on image quality. 118 University of Ghana http://ugspace.ug.edu.gh Figure 3.19: Unnecessary areas (grid lines) identified with Full range scan along z-axis and their corresponding unnecessary images for disgnosing CVA conditions are also displayed. Red arrow shows Reduced range CT area capable for CVA eaxaminations. 119 University of Ghana http://ugspace.ug.edu.gh 3.8.1.2.5 Radiation Dose Assessment Following the reconstruction of the acquired images, the radiation dose descriptors (DLP, ED and organ doses) for the Full and Reduced scan protocols were calculated using the NCICT software version 2.1 (National Cancer Institute, USA). The software has an interactive tool which allows for manual adjustment of the scan range to determine the aforementioned dose indices based on the scan parameters and the anthropometric information of the patients. With fixed scan parameters, the scan lengths for the Full and Reduced protocols of each patient data were reformatted and organ doses for brain, pituitary gland, eye lens, eyeballs, salivary glands, oral cavity, spinal cord and thyroid were assessed. 3.8.1.2.6 Image Quality Assessment The image quality assessment procedures outlined in the phantom-based study were replicated in the patient-based study (Section 3.8.1.1.3) and the results are presented in Table 4.60. Moreover, radiologists reported on the pathologies (CVA-based and incidental findings) on the images of both Full and Reduced range protocols. The results on the conditions that were diagnosed on both the reduced and full range images are presented in Table 4.62. 3.8.1.2.7 Data Analysis The data analysis procedure used in the phantom-based study (Section 3.8.1.1.4) was also employed here. In addition, the organ doses, DLP and ED for both protocols were then compared. A Shapiro-Wilk test was used to test for normal distribution of the data. Based on the normality of the distribution of data (Appendix X), a parametric T-test was used to compare the Full-range and Reduced-range data variables. A p-value of ≤ 0.05 was considered statistically significant in all statistical analyses. 120 University of Ghana http://ugspace.ug.edu.gh The MINITAB statistical tool version 19 was also used to model relationship between some organ doses (brain, salivary gland, eye lens, eyeballs, oral cavity, spinal cord, thyroid and pituitary glands) and some exposure parameters (such as CTDIvol, slice thickness (T) and scan length). The modelling process was centred on linear regression. To satisfy the assumptions of using linear regression, a large sample size > 20, continuous data variables and Shapiro-Wilk test of normality (as explained above for normality test) were used in the study and the data variables satisfied the normality rule as shown in Appendix X. Grubbs' Test, a statistical tool for testing for outlies in MINITAB version 19 was used to ensure that there was a lack of strongly influential outliers in the data variables, as shown in Appendix XI. A Grubbs' Test p-value < 0.05 would suggest the presence of strongly influential outliers. However, all the p-values were > 0.05 at 95% significance level (Appendix XI). Studies (Stoltzfus, 2011; Tabachnick and Fidell, 2007) have suggested that there should be an absence of multicollinearity in the independent variables for linear regression models as correlations or multiple correlations of enough magnitude have the ability to adversely affect regression estimate. This was achieved by using Variance Inflator Factors (VIF) to indicate the degree to which the standard error was inflated due to the levels of collinearity (Stoltzfus, 2011). Initial assessment showed that there were high VIF greater than 3 among some of the independent variables which showed moderately correlated independent variables (Appendix XII). Subsequently, those variables were eliminated to reduce the multicollinearity among the independent variables, prior to using the regression model. 121 University of Ghana http://ugspace.ug.edu.gh 3.8.2 Optimisation Method 2: The Role of AEC Utilisation 3.8.2.1 Overview Follow up queries across the facilities scanning without AEC, as identified in this study revealed that these scanners had default setting limitation where the AECs were not activated for certain procedures. Also, some practitioners were not aware the AECs were off and due to limited knowledge about AEC operations they were not often checked. To effectively evaluate the impact of the AEC and make needed optimisation recommendations, an experimental study was undertaken in four facilities in Ghana scanning without AEC using an adult whole-body CT anthropomorphic phantom PBU-60. 3.8.2.2 Materials The phantom and the CT scanner characteristics/specifications used in this experiment are shown in Tables 3.6 and 3.7, respectively. Table 3.6: Characteristics of CT phantom PBU-60 Parameters Characteristics Phantom estimated weight 50 kilogram (kg) Phantom estimated height 165 cm Phantom estimated age 30 years (adult) Manufacturer KYOTO KAGAKU Co., LTD Phantom composition: Soft tissue/organs: Urethane based resin Synthetic bones: Epoxy resin Joint attachments: Epoxy, urethane with carbon fibre Screws: Polycarbonate 122 University of Ghana http://ugspace.ug.edu.gh Table 3.7: Characteristics/specifications of CT scanners CT ID Manufacturer Model Scanning parameters Fixed mAs AEC CT2 Siemens Somatom AEC: off AEC: CareDose4D Emotion Scan length: 46 Scan length: 46 cm cm kV: 100; kV: 100; Pitch: 1 Pitch: 1 Rotation time: 0.5, Rotation time: 0.5 Ref. mAs 250 Fixed mAs: 240 Mean effective mAs 166 CT8 GE Lightspeed AEC: off AEC: AutomA 3D Pro 16 Scan length: 46 Scan length: 46 cm cm kV: 120; kV: 120; Pitch: 1.38; Pitch: 1.38; Rotation time: 0.8 Rotation time: 0.8 Mean mAs: 105 Fixed mAs: 170 CT14 Toshiba Aquilion 16 AEC: off AEC: SureExposure 3D TSX Scan length: 46 Scan length: 46 cm cm kV: 120; kV: 120; Pitch: 0.94; Pitch: 0.94; Rotation time: 0.75, Rotation time: Mean mAs: 137.7 0.75, Fixed mAs: 221 CT19 Philips Brilliance 64 AEC: off AEC: DoseRight Z-DOM, Scan length: 46 ACS cm Scan length: 46 cm kV: 120; kV: 120; Pitch: 1.17; Pitch: 1.17; Rotation time: 1, Rotation time: 1, Fixed mAs: 276 Mean mAs 191.8 AEC: automatic exposure control; kV: kilovoltage, mAs: milliampere-second, CT: computed tomography, ACS: Automatic Current Selection 123 University of Ghana http://ugspace.ug.edu.gh 3.8.2.3 Methods and Procedure For each facility, the phantom was placed on the CT table and positioned for a routine AP CT examination. It was placed supine, and with the help of the laser beams, an isocentre position was achieved. The midsagittal plane coincided with the vertical lasers while the true coronal plane coincided with the horizontal lasers as shown in Figure 3.20. A scanogram was acquired from the top of higher hemidiaphragm to below the symphysis pubis. Subsequently, actual imaging, using the routine protocol of the AP region (as the case of non-contrast AP lesion) was undertaken with and without the activation of the AEC. The position of the lasers on the phantom and scan positions/coverage (46 cm) were kept constant for the scans. Figure 3.20: Positioning of PBU-60 phantom for AEC studies The dose output descriptors for experiments with and without AEC were documented in a Microsoft Excel version 2013. The dose reduction was calculated from the DLP values using equation 3.12: 𝐷𝐿𝑃𝐴𝐸𝐶 𝑜ff −𝐷𝐿𝑃𝐷𝑅 = 𝐴𝐸𝐶 × 100% (3.12) 𝐷𝐿𝑃𝐴𝐸𝐶 off 124 University of Ghana http://ugspace.ug.edu.gh where DR is dose reduction DLPAEC off is measured dose length product without AEC DLPAEC is measured dose length product with AEC 3.8.2.4 Image Quality and Data Analysis Image quality in terms of noise levels (using the standard deviation, SD, of CT number) and SNR were measured at proximal, middle and distal part of the phantom (at 577, 722 and 902 z-positions) using ImageJ version 1.52 software. First, ROIs of area 10 mm2 were drawn at the specific places on image slices and the above-mentioned software programmes were used to determine the noise (SD) and the SNR. As already shown in Figure 3.12, the same measurement approach was used. 3.8.2.5 Data analysis Data were then analysed using the Microsoft Excel version 2013 and SPSS version 23.0 (SPSS Inc., Chicago, IL USA). The dose descriptive data for all the protocols with fixed mAs and those with AEC were descriptively compared using Microsoft Excel version 2013 and percentages of DLP dose reduction over the current practice (originally used across the facilities) were estimated. In addition, based on the normal distribution of the data, a comparative analyses of the dose output (DLP) accruing from AEC and fixed mAs examinations were also compared with a T-test using SPSS. A p-value of ≤ 0.05 was considered statistically significant. The results are presented in Table 4.63. 125 University of Ghana http://ugspace.ug.edu.gh 3.9 Ethical Consideration and Data Handling This study was conducted in adherence to the international research ethical guidelines and in compliance with the Helsinki declaration (World Medical Association Declaration of Helsinki, 2001). Prior to the study, ethical clearances (Appendices XIII- XV) were sought and approved by the Ethics Committee for Basic and Applied Sciences (ECBAS), University of Ghana (REF. No: ECBAS 041/17-18), the Ghana Health Service Ethics Review Committee (REF NO: GHS-ERC002/04/18), Korle Bu Teaching Hospital’s Scientific and Technical Committee (KBTH-STC) and the Institutional Review Board (KBTH-IRB) (REF NO: KBTH- IRB/00092/2017). The Nuclear Regulatory Authority (NRA) of Ghana also granted permission and supported the study. Subsequently, permissions were sought and granted by all the Heads and Managers of institutions responsible for the CT scanners that were used in this study. Some of the permission letters are shown in Appendix XVI. All the CT facilities which finally took part in the study were coded and identified with alphabets to prevent public linkage of information to any of them. Information sheets were used to educate all participants who provided questionnaire data prior to recruiting them. They were made aware that the research study carries very negligible risk and the outcome had the potential to help to establish indication-based DRL values and dose optimisation methods for CT examinations in Ghana for policy decisions and directions in the country. They were made aware that their participation in this research was voluntary, and they were at liberty to decline or withdraw from the research without providing an explanation at any time until the work is completed. In addition, they were informed that their personal details were not going to be used in the study and consequently, their identities were coded to ensure their confidentiality. All those who participated in the Phase 1 stage of the study finally consented to participate in the study before their involvement. 126 University of Ghana http://ugspace.ug.edu.gh With respect to the patients’ image folders retrieved from the CT Archives and workstations during the study period, the identities of patients were safeguarded by blocking the names during the data collection. The CT scanners display software programmes had “identity hide” tools for blocking patients’ names and identities. The activation of the tools concealed patients’ identities during the data collection and analysis process. The collected data sets were also stored on a computer and encrypted with a password to further ensure that confidentiality of information and details peculiar to the patients were also protected. All generated data sets were used for research purposes only and nothing else and would be kept for at least five years for future review if necessary. All the aforementioned were done to ensure patient’s anonymity and confidentiality were considered during the study. 127 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR RESULTS AND DISCUSSION 4.1 Overview This Chapter presents the results and the discussion of the thesis and gives account of the six phases of the study. The results and discussions of Phase 1 covered the technical data on the CT infrastructure in the various facilities, common indications and definition of the basic diagnostic imaging requirements. Phase 2 addressed the CT scanner performance characteristics data. The outcomes of Phases 3 and 4 covered CT dose descriptors data, image quality assessment and DRLs estimations. Results and discussions on estimation of cancer risks associated with the doses and optimisation methods are presented under Phases 5 and 6, respectively. 4.2 PHASE 1: CT Technical Data, Common Indications and Imaging Requirements As part of preliminary steps to achieve the objectives of the entire study, CT technical /infrastructure data, common indications and imaging requirements were first investigated (Phase 1 of the study). The results of Phase 1 are presented in two sections: the results on technical data on CT scanners and common indications are presented in Section 4.2.1, while those on basic diagnostic imaging requirements are presented in Section 4.2.2. 4.2.1 CT Technical Data and Common Indications CT is a key component of modern healthcare in all countries. According to Statista, as of the year 2017, the distribution of CT scanners per one million inhabitants were 64 (Australia), 44 (United States of America), 35 (Germany) and 15.28 (Canada), respectively (Statista, 2017). However, 35 authorised CT scanners existed in Ghana at the time of the study. This was approximately one (1) CT scanner per 850,000 inhabitants or 1.18 scanners per one 128 University of Ghana http://ugspace.ug.edu.gh million inhabitants (with an estimated population of 30 million inhabitants as indicated by the Ghana Statistical Services, 2019). Comparatively, this is 5.78% of the EU average of 20.4 CT scanners per 1 million population (Vanckavicienea et al., 2016). Out of the 35 scanners in the country, responses on CT technical data, and common indications were received from the technical heads responsible for 31 CT scanners, representing 88.6% of the total number. The demographics of the technical heads are presented in Table 4.1. Table 4.1: Demographics of study respondents (Technical Heads) Demographics Number Percentage (%) Male 23 74.2 Gender Female 8 25.8 Chief Radiographer 3 9.7 Rank (Grade) Deputy Chief Radiographer 3 9.7 Principal Radiographer 16 51.6 Senior Radiographer 7 22.6 Principal Technical Officers 2 6.5 Most (n=16, 51.6%) of the technical heads were principal radiographers in their respective hospitals and majority (n=23, 74.2%) were males. This suggests a male dominated technical leadership at CT facilities in Ghana. A study (Steffens et al., 2019) has also noted that gender balance has been particularly scarce at leadership level with men mostly being in the majority. The technical data on CT models, manufacturers, year of manufacture (YoM), year of installation (YoI) and number of detector rows/slices identified across the CT facilities are presented in Table 4.2. 129 University of Ghana http://ugspace.ug.edu.gh Table 4.2: CT models, manufacturers, years of manufacture, installation and number of detector rows/slices (N=31) Hospital CT Manufacturer Model YoM YoI Detector ID row/slice A CT 1 Toshiba Aquilion One TSX-301A 2012 2012 320/640 B CT 2 Siemens Somatom Emotion 6 2006 2011 6 C CT 3 GE Brightspeed Elite 2011 2011 16 D CT 4 Philips Brilliance ICT 2015 2016 128 E CT 5 Siemens Somatom Perspective 2016 2016 16 F CT 6 GE VCT Lightspeed 2008 2009 64 G CT 7 Siemens Somatom Perspective 2016 2017 64 H CT 8 GE Lightspeed Pro 16 2011 2011 16 I CT 9 GE Brivo CT 385 series 2015 2016 16 CT 10 Siemens Somatom Emotion 2007 2008 6 J CT 11 Toshiba Aquilion GX 2012 2012 128 K CT 12 Toshiba Aquilion TSX-101A 2016 2016 16 L CT 13 Siemens Somatom Emotion 2010 2010 16 M CT 14 Toshiba Aquilion TSX-101A 2013 2013 16 CT 15 Toshiba Aquilion CXL TSX-101A 2015 2015 32 N CT 16 Philips MX 16 2015 2016 16 O CT 17 Toshiba Aquilion CXL TSX101A 2012 2015 32 P CT 18 GE Revolution Evo 2017 2017 64 Q CT 19 Philips Brilliance 2009 2010 4 R CT 20 Hitachi Supria 2015 2015 16 S CT 21 Philips Brilliance extended 2007 2010 64 T CT 22 Picker PICKER PQ5000 1998 2017 1 International U CT 23 Siemens Somatom Emotion 2009 2010 16 V CT 24 GE Brightspeed Edge# 1998# 2009# 8# W CT 25 Siemens Somatom Emotion 2010 2011 16 X CT 26 Siemens Somatom Definition AS 2015 2016 64 Y CT 27 Philips Brilliance 16 2010 2016 16 Z CT 28 Toshiba Asteion 2009 2016 4 AA CT 29 Toshiba Aquilion TSX-101A 2015 2015 16 CT 30 Toshiba Aquilion TSX-101A 2012 2013 16 AB CT 31 GE Optima 660 2016 2016 64 Key: YoM: year of manufacture, YoI: year of installation. CT: computed tomography. The sign “#” means that that machine was replaced with a 64 slice GE Revolution 5492001. It was manufactured and installed in 2018. Majority of the CT scanners as indicated in Table 4.2 were Toshiba (n=9, 29.0%) and Siemens (n=8, 26.0%) models. All recently installed public CT scanners by the Government of Ghana were Toshiba models, which contributed to the increased tally. The detector row and equipment slices ranged from 1 to 320 and 1 to 640, respectively, of which majority (n=14, 130 University of Ghana http://ugspace.ug.edu.gh 45.2%) were 16 slice scanners. The results indicated that 30 out of 31 scanners, representing 96.8%, were MDCT systems. Advanced MDCT scanners with higher detector configuration offer faster image acquisition speed, more patient body coverage, high spatial resolution, optimal image quality and optimised radiation dose levels (if properly utilised) (Gao et al., 2017). The year of manufacture ranged from 1998 to 2017 while installations were undertaken from 2008 to 2017. Only three facilities had two installed scanners, while the others had a single scanner. The mean age of the CT equipment was 7.3 ± 4.4 years old and 25.6% of them were 10 or more years old. According to a report (European Society of Radiology, 2014), scanners older than 10 years are no longer considered state-of-the art equipment and may not have the current technology for full patient benefits. Moreover, newer CT technologies offer 10–30% lower radiation exposure levels compared to systems installed 5 years ago (European Society of Radiology, 2014). There is, therefore, a need to upgrade older scanners (25.6%) in Ghana. Interestingly, one of the very old scanners (GE Brightspeed Edge manufactured in 1998 and installed in 2009) was being replaced with a 64 slice GE Revolution (manufactured in 2018) at the time of the study. This is a commendable action. The scanners’ ability to display dose descriptors (such as CTDIvol and DLP parameters), scan mode systems incorporated in the CT scanners, and the availability of AEC systems on the CT scanners are presented in Figure 4.1. 131 University of Ghana http://ugspace.ug.edu.gh Ability to display SCANNERS SCAN MODE Availability of AEC CTDIvol & DLP 96.8 70.96 parameters % % 100.0% 3.23 % 80.0% 60.0% 25.81 96.77 % 40.0% % 20.0% 3.2% 3.23% 0.0% Yes No Yes No Axial Helical Both Scan mode A B C Response Figure 4.1: Equipment ability to display CTDIvol and DLP parameters (a); the scan mode systems incorporated in the CT scanners (b); availability of AEC systems on the CT scanners (c). From Figure 4.1(a), 30 (96.77%) of the CT scanners could display CTDIvol and DLP parameters on the console, while 1 (3.23%) could not. Currently, it is an international requirement that all operational CT scanners should display dose descriptors such as CTDIvol and DLP on the CT console, post examination (ICRP, 2017; International Electrotechnical Commission, 2011; Institute of Physics and Engineering in Medicine, 2005). It was observed that one of the scanners lacked a self-display dose output configuration. Consequently, there is the need for the operator of this scanner to ensure compliance with the international mandatory requirement of console-displayed CT dose outputs. In Figure 4.1(b), although majority (n=22, 70.97%) of the scanners had the ability to scan in both helical and sequential modes, 8 (25.81%) and 1 (3.23%) of them were designed to scan in only helical and sequential modes, respectively. Unfortunately, 1 scanner (3.2%) was not incorporated with as AEC system. The AEC system’s advantage of yielding significant reductions in patient doses while maintaining appropriate image quality could thus not be realised in this CT machine and needs to be replaced or upgraded (Higaki et al., 2019; Fillon et al., 2018). 132 Frequency, % Frequency, % University of Ghana http://ugspace.ug.edu.gh The geographical distribution, ownership and functionability of the CT scanners across the country are presented in Table 4.3. Table 4.3: Geographical location, ownership status and functionality of the CT scanners (N=31) Location/ Ownership status Functionability region Public Private Quasi Total Yes No GA 4(12.9%) 11(35.5%) 3(9.7%) 18(58.1%) 17(54.9%) 1(3.2%) Ashanti 2(6.5%) 4(12.9%) - 6(19.4%) 5(16.1%) 1(3.2%) Northern 2(6.5%) 1(3.2%) - 3(9.7%) 2 (6.5%) 1(3.3%) Western 1(3.2%) - - 1(3.2%) 1(3.2%) - Eastern 1(3.2%) - - 1(3.2%) 1(3.2%) - BA 1(3.2%) - - 1(3.2%) 1(3.2%) - Central 1(3.2%) - - 1(3.2%) 1(3.2%) - Total 12(38.7%) 16(51.6%) 3(9.7%) 31(100%) 28(90.3%) 3(9.7%) GA= Greater Accra; BA= Brong Ahafo; quasi: public-private partnership. Out of the 31 CT scanners, the majority (n=16, 51.61%) were operated by the private sector, while 12 (38.7%) were operated in public health facilities, and 3 (9.68%) by quasi- governmental establishments (establishments controlled both by the government and private sectors). The CT scanners were not evenly distributed across the 10 regions of the country at the time of the study. In particular, most (18/31) CT scanners were operated in the Greater Accra Region inhabited by 16.3% of the national population (Ghana Statistical Services, 2019), and none installed in the Upper West, Upper East and the Volta Regions. The lack of CT scanners in these locations means that patients requiring CT services (even in emergency cases) would have to travel to other regions to access the service, which could compromise effective patient management. At the time of the study, 28 (90.32%) of the scanners were functional, while 3 (9.7%) were out of operation and could not be used in the DRLs study. 133 University of Ghana http://ugspace.ug.edu.gh The categories of health professionals associated with the operation, quality and safety maintenance, and image reporting activities of the CT scanners are shown in Table 4.4. Table 4.4: Number of CT scanning radiographers, reporting radiologists and attending medical physicists Professional Number Ratio to population Radiographers 107 1: 280,374 Radiologists 60 1: 500,000 Medical Physicists 10 1: 3,000,000 There were 107 radiographers operating the CT scanners, 60 reporting radiologists and 10 medical physicists in the diagnostic radiology responsible for CT scanners. Per the estimated population of Ghana as of 2019 (Ghana Statistical Services, 2019), the findings further suggest that the staff to populace ratios were 1: 280,374 for radiographers, 1: 500,000 for radiologists and 1: 3,000,000 for medical physicists. Compared to the situation in developed countries (Germany, USA, France) (Silvestrin, 2016), these numbers are very low, necessitating the need to address the staffing-gap situation. In particular, the professional distribution with respect to medical physicists is worse because of the lack of their recruitment by the public sector into the diagnostic radiology, although they are available for employment. Due to this limitation, many facilities mainly rely on the QC tests undertaken by the NRA for purposes of inspections, verifications of compliance with regulatory requirements, and issuance of authorisation to practice. This authorisation is renewed every 3 years. The inspection conducted every three years means that there is a lot of scope for something going wrong and not being fixed for a long time. This has the potential to compromise effective healthcare delivery. There is thus, an urgent need to employ the relevant health professionals in CT facilities to promote patient care, protection and safety during health delivery. 134 University of Ghana http://ugspace.ug.edu.gh QA is an interdisciplinary management tool including policies that offer a means for guaranteeing that all activities, including CT imaging, are effectively planned, correctly performed and assessed to meet specifications. The availability of QA infrastructure at the various CT facilities is presented in Table 4.5. Table 4.5: QA infrastructure availability at the CT facilities (N=31) in Ghana Quality Assurance structures Responses Yes No n % n % Availability of QA or QC Committee 17 54.8 14 45.2 Availability of a documented protocol for CT scanning 15 48.4 16 51.6 Post-major repair QC assessment records 21 67.7 10 32.3 Do you have records of regular QC checks 20 64.5 11 35.5 Authorisation by NRA 31 100.0 0 0.0 Availability of established acceptance testing procedure 19 61.3 12 38.7 Availability of effective planned maintenance schedules 12 38.7 19 61.3 Equipment performance record keeping 29 93.5 2 6.5 Availability of radiation protection devices 31 100.0 0 0.0 Availability of systems for justifying CT exposures 31 100.0 0 0.0 Availability of adopted or established DRLs 0 0.0 31 100.0 Availability of scheduled optimisation programmes 20 64.5 11 35.5 Keeping of patient dose records 20 64.5 11 35.5 Availability of an established patient dose and image quality 5 16.1 26 83.9 audit programmes Planned schedules for frequent cleaning of equipment 10 32.3 21 67.7 Documented training program and records 5 16.1 26 83.9 Key; QA: quality assurance, QC: quality control; NRA: Nuclear Regulatory Authority. DRLs: diagnostic reference levels, CT: computed tomography. QMS and quality programmes are the physical and non-physical structures required by radiology personnel to ensure patient satisfaction, protection and safety in radiology departments (Kruskal et al., 2009). The results indicated some CT facilities in Ghana lacked 135 University of Ghana http://ugspace.ug.edu.gh the needed QA infrastructure. In particular, 14 (45.2%) of the facilities lacked essential infrastructure such as QA/QC Committees, while a documented protocol for CT scanning was absent in 16 (51.6%) of them. A study (Ofori et al., 2013) has equally observed this situation in the general x-ray facilities in Ghana. According to the IAEA (IAEA, 2012), the QA/QC Committees have the primary function to keep lines of communication amongst all groups with QA, QC and QI responsibilities. These include the maintenance of acceptable standards of quality and periodical review or audit of practices for effectiveness and compliance with quality standards. Therefore, the absence of operational tools like documented scanning protocols and appropriate committees to ensure functional QA, QC and QI could have severe consequences on the radiation protection culture (Ofori et al., 2013). A study (Schandorf and Tetteh, 1998) reported that established acceptance testing procedures, commissioning, regular QC tests and post-major repair records are important international indices for assessing equipment performance after an installation and over a period of time. However, in this study, regular and post-major repair QC assessment records were missing in 10 (32.3%) and 11 (35.5%) of the facilities, respectively, while established acceptance testing procedures were lacking in 12 (38.7%) facilities. A similar situation was reported in Tanzania, where only 6% of the facilities had records of all QC (Ngoye et al., 2015). Although it is commendable that all facilities had received authorisation from the NRA and had radiation protection devices and systems for justifying CT exposures, it was also worrying that some facilities lacked effective planned maintenance schedules (n=19, 61.3%), equipment performance records (n=2, 6.5%), and scheduled dose optimisation programmes (n=1, 35.5%), respectively. Patients’ dose record keeping, and establishment of patient dose and image quality programmes are important optimisation processes. However, these programmes were missing in majority (n=26, 83.9%) of the facilities, respectively. Furthermore, planned schedules for frequent cleaning of equipment and documented training 136 University of Ghana http://ugspace.ug.edu.gh programme and records were available in only 10 (32.3%) and 5 (16.1%) facilities. These need to be improved. Figure 4.3 presents the breakdown of the regularity of the routine QC among the facilities (n=20) that indicated their involvement in regular QC testing. Public 25.00 Private 20.00 Quasi 15.00 10.00 5.00 0.00 Daily Weekly Quarterly Every 6months Frequency of routine QC Figure 4.2: Regularity of routine Quality control (QC) (N=20) Daily and weekly QC assessments were undertaken in two public sector facilities, while quarterly assessments were done in 3 (60.0%) others. In the private sector category, QC assessments were conducted daily (n=5, 41.7%), weekly (n=5, 41.7%), and quarterly (n=3, 16.7%), while same were done daily, weekly and quarterly in each (33.33%) of the 3 quasi- government facilities. The lack of regular QC tests across all the CT facilities could be attributed to the inadequate numbers of medical physicists in the facilities. This lack of systems to ensure effective QC tests and record keeping across the CT facilities does not promote quality service delivery (Korir et al., 2013). The availability of basic QI structures in the CT facilities in Ghana is presented in Table 4.6. 137 Number (%) University of Ghana http://ugspace.ug.edu.gh Table 4.6: Basic quality improvement (QI) structures in the CT facilities (N=31) in Ghana QI structures Availability Yes No n % N % Do you have institutional leadership and support toward QI? 31 100 - - Do you have regular meetings involving all the stakeholders to 5 16.1 26 83.9 communicate QC results? Has your facility establish a culture of quality in your practice? 3 9.7 28 90.3 Has your facility established an improvement team? 5 16.1 26 83.9 Do you have a QI protocol or manual 3 9.7 28 90.3 Do you have a system that engages all the professional groups in 6 19.4 25 80.6 the department on QI? Do you have a system that regularly receives and analyses 10 32.3 21 67.7 feedback from customers and stakeholders? Do you have surveillances system for monitoring QI indicators? 3 9.7 28 90.3 Does your facility have a system to reward hard work associated 18 58.1 13 41.9 with QI? Do you have an educational programme on QI? 5 16.1 26 83.9 QI: quality improvement, QC: quality control Apart from available institutional leadership and support toward QI (100%), and QI reward systems (58.1%), over 67% of facilities lacked all the pertinent structures that constitute effective QI systems. Over 90% of the facilities lacked the protocol or manual and surveillance systems for monitoring QI indicators. This lack of effective QI infrastructures is a concern for the imaging services in the country which may have dire consequences on improved care. This is because the goal of QI systems, in particular, is to do the needful in a timely fashion for every patient every time and to ensure that a particular level of quality performance is improved (Kruskal et al., 2011). The availability of Quality Management Teams, established culture of quality and regular meetings involving all the stakeholders for purposes of communicating QC results for instance, were only available in 5 (16.1%), 3 (9.7%) and 5 (16.1%) of the facilities, respectively. However, it is known that a well-functioning, cohesive team, stakeholder relations 138 University of Ghana http://ugspace.ug.edu.gh and a culture that recognises quality are crucial for the effective implementation and operation of a quality radiation protection and safety program in any facility (Ngoye et al, 2015; UNSCEAR; 2012). The data on CT policy driven infrastructure and their availability are shown in Table 4.7. Table 4.7: Policy infrastructure and their availability in Ghana Policy infrastructure Availability Yes No n % N % Policy on CT authorisation for use in Ghana 31 100 - - Policy driving CT infrastructure and distribution - - 31 100 Policy on operation and maintenance - - 31 100 Policy and availability of quality management systems - - 31 100 Policy on purchasing, construction and installations - - 31 100 Policy on decommission of a CT facility 31 100 Policy on education and training 31 100 - - Policy and availability of diagnostic reference levels - - 31 100 Policy on CT referrer guidelines - - 31 100 Policy on recommended frequency for QC tests - - 31 100 Standardised policy on AEC application - - 31 100 Policy on patient dose optimisation, image quality and - - 31 100 audit programmes Policy on acceptance testing and record keeping - - 31 100 Policy on human resources in CT facilities - - 31 100 Policy on CT maintenance systems - - 31 100 QC: quality control, AEC: automatic exposure control CT infrastructure is generally driven by policies in respect of directing purchasing, installation, commissioning and operation of CT facilities to ensure a sustainable service to the population (Kruskal et al., 2011). However, Table 4.7 showed the lack of many structured policies, with the exception of CT authorisation, and education and training of staff. A study 139 University of Ghana http://ugspace.ug.edu.gh (Qureshi, 2019) has cautioned that the lack of practice-driven structured policies could limit economic growth and development as well as improved care. Table 4.8 shows the duration (down time) taken to repair broken down CT equipment operated in the various categories of sectors. Table 4.8: Duration to repair broken down equipment (down time) Organisation Sector Duration Public 1 day -2 years Private 1 hour -1 month Quasi 1 day - 3 months A fault is a key index for determining the down time of CT imaging equipment. Depending on the type, broken-down CT equipment in the public hospitals were repaired between a day to 2 years. On the contrary, shorter down times of an hour to one month, and a day to 3 months were required in the private and quasi-government hospitals, respectively. Many health institutions have applied planning and scheduling to their maintenance functions to enhance the use of their equipment, decrease unplanned downtime and more efficiently manage their maintenance budgets. However, effective planned maintenance schedules, which were missing in 61.3% of the facilities, could be the cause for long down times in the public facilities in particular. A study (Korir et al., 2013) had also found in Kenya that only 22% of the health centres had reports of semi-annual preventive maintenance. As a result, equipment faults are likely not to be detected on time, and when a breakdown occurs, it could take a long time to have the equipment fixed. The lack of parts supply, after sales service and training of local human resource to perform maintenance has long been noted as part of the major problem (ICRP, 2007). The absence of formal organisational structures for effective planned maintenance systems for CT facilities in Ghana, QMS and management culture have clinical, 140 University of Ghana http://ugspace.ug.edu.gh radiation protection and financial implications due to the long downtimes. The Information on the equipment down time was crucial in determining when a broken-down equipment could be waited upon to be fixed and included in this DRL study. Schematic data on whether or not CT facilities benchmark their dose information with internationally established indication-based DRLs, and agreement levels of respondents on the need for the establishment indication-based DRLs in Ghana are presented in Figure. 4.3. 120 100 100 80 80 60 60 40 40 20 20 0 0 None Other  Yes No Response β Response Figure 4.3: Responses on (A) whether or not CT facilities benchmark their dose information with internationally established indication-based DRLs and (B) respondents agreement levels on the need for the establishment indication-based DRLs in Ghana. A DRL is a QA/QC tool used as a trigger to find imaging centres using remarkably high radiation doses in a particular examination, for which optimisation measures are needed (ICRP, 2017; IAEA, 2013). It needs to be established or adopted so that dose outputs from facilities could be benchmarked (ICRP, 2007) for effective management practices. Figure 4.3 shows that CT facilities in Ghana do not compare their dose parameters with any indication-based DRLs. However, all the respondents agreed on the need for the establishment and consequent practice of indication-based DRLs in Ghana. 141 Frequency (%) Frequency (%) University of Ghana http://ugspace.ug.edu.gh The various examination indices (number and percentage) performed each year across the facilities are presented in Table 4.9 and Figure 4.4. Table 4.9: CT examinations undertaken in a year Location Examinations (Region) Head Neck Chest Abdomen Pelvic AP L/S Ext Special Total GA 66462 1873 11204 7737 3174 21713 1313 636 8398 122510 Ashanti 25387 1133 3073 1009 872 10364 773 297 2242 45150 Northern 6434 621 1460 201 310 1565 342 556 192 11680 Western 3391 278 705 239 71 819 210 38 88 5840 Eastern 3559 262 652 278 147 1321 196 13 117 6545 BA 3899 260 169 141 141 821 157 26 78 5690 Central 4385 385 611 398 277 864 185 75 164 7345 Total 113517 4812 17874 10003 4990 37467 3177 1640 11280 204,760 GA= Greater Accra; BA= Brong Ahafo; Ext= Extremities; L/S= Lumbar spine, AP= abdominopelvic. A B 59.2 C % 60.0% 0.8% 3.2% 2.8% 3.6% 5.5% 2.9% 1.6% 50.0% 35.8 5.7% 40.0% % 18.3% 22.0 2.4%59.8 4.9% 55.4% 30.0% % % 8.7% 20.0% 2.4% 10.0% 5.0% Greater Accra Ashanti Head Neck Northern Western Chest Abdomen 0.0% Pelvic Abdomino-pelvic Public Private Quasi Eastern Brong Ahafo L/spine Extremities Central Special cases Facilities Figure 4.4: Percentage distribution of CT examinations per region (Plate A), anatomical part (Plate B) and hospital type (C). 142 Frequency, % University of Ghana http://ugspace.ug.edu.gh Table 4.9 and Figure 4.4 show that 204,760 CT examinations were performed in a year, of which head (n=113,517, 55.4%), AP (n=37,467, 18.4%) and chest (n=17874, 8.7%) were the most prevalent examinations. These three types of examinations accounted for 82.4% of the CT procedures conducted in Ghana in a year. The frequency of these procedures is generally consistent with the worldwide trend (Jaffe et al., 2010). The extremities (n=1640, 0.8%) were the least prevalent CT procedures. The Greater Accra and Ashanti Regions contributed 59.8% and 22.0% of the CT examinations, respectively, because of the larger number of installed CT scanners in the two Regions. Majority of CT examinations were undertaken in the public (n=121,279, 59.2%) and private (n =73,366, 35.8%) imaging facilities. While only 5.0% were reportedly performed in quasi hospital organisations. The total number of CT examinations performed in a particular year translates into a lower national annual average of 6.8 CT procedures per 1000 people compared to 226.6, 94.4 and 85 CT procedures per 1000 people in the USA, Canada and UK, respectively (OECD, 2019; Statista, 2017). However, Kenya has a national annual number of 3 CT procedures per 1000 inhabitants, which is about twice less than the Ghanaian annual average (Korir et al., 2016). The results of the analysis of the various common indications for adult CT examinations as well as their scanning protocol similarity index are presented in Table 4.10. 143 University of Ghana http://ugspace.ug.edu.gh Table 4.10: Identified common CT indications for adult head, chest and abdomino-pelvic regions Indication per anatomical Number Percentage Scanning Prioritised region (n) (%) protocol indications for similarity index DRLs Head region CVA/Stroke 37234 32.8 A ✓ Head injury/trauma 25314 22.3 B ✓ Headaches with suspicion of 17482 15.4 C ✓ SOL/tumour Sinusitis 2724 2.4 D Dizziness 1135 1.0 C Loss of consciousness 1249 1.1 C Psychiatric disorder 3406 3.0 C Others 24974 22.0 - Chest region Cough with suspicion of lung 5720 32 E ✓ cancer/tumour Suspicion of chest lesion with 1752 9.8 F ✓ chronic kidney disease (CKD) Tuberculosis (TB) 1662 9.3 E Chronic obstructive pulmonary 429 2.4 E disease (COPD) Pulmonary embolism (PE) 2681 15 G ✓ Pneumonia 518 2.9 E Pleural effusion 500 2.8 E Lung abscess 894 5 E Interstitial lung disease 894 5 E Haemoptysis 375 2.1 E Others 2449 13.7 - Abdomino-pelvic regions (include abdomen, pelvis, both) Suspicion of AP lesion/abscess 14584 27.8 H ✓ Kidney stones (plain) 9705 18.5 I ✓ Liver metastasis 5403 10.3 J Urothelial malignancy 7974 15.2 K ✓ Bowel obstruction 1626 3.1 H AP distension 2990 5.7 H Prostate cancer 2361 4.5 L Bladder mass 1312 2.5 L Others 6505 12.4 - Key: AP: Abdomino-pelvic. CVA: cerebrovascular accident. Note: Indications with similar scanning protocol index (eg. C, E, H, L alphabets) were scanned with very common protocol such as collimation, number of sequences, contrast usage or not, scan length etc. Indication with exclusive alphabet (eg. A, B, D, F, G, I, J and K) used a unique scan scanning protocol. DRLs: diagnostic reference levels. 144 University of Ghana http://ugspace.ug.edu.gh Many publications (ICRP, 2017; European Society of Radiology, 2017; Vocks and Frija, 2016; Lajunen, 2015) have recommended baseline DRLs for common diagnostic procedures and indications which contribute significantly to the population dose. CVA/stroke, a brain cell death condition, resulting from poor blood flow to the brain, and the second leading cause of death worldwide (WHO, 2019), was identified as the most common (n=37234, 32.8%) indication for head CT imaging. Sanuade et al. (2019), indicated that it was 2.6% prevalent in Ghana. Other common indications included head injury (n=25314, 22.3%) and headaches with suspicion of SOL or tumour (n=17482, 15.4%). Indications such as sinusitis, dizziness and loss of consciousness were less common. Scanning protocols for the former three indications as well as sinusitis differed from the rest. However, scanning protocols for dizziness and psychiatric disorder were similar to those of headaches with suspicion of SOL/tumour. Based on the frequency and protocol dynamics, CVA/Stroke, head injury/trauma, and headaches with suspicion of SOL/tumour were prioritised for DRLs in the head region. In the chest category, as indicated in Table 4.10, cough with suspicions of lung cancer/tumour, chest lesion with CKD, TB, COPD, PE, pneumonia, pleural effusion, lung abscess and interstitial lung diseases were identified as the common indications for adult patient CT examination of the chest region. However, cough with suspicion of chest lung cancer/tumour (n=5720, 32%), suspicion of chest lesion with CKD (n=1752, 9.8%) and PE (n=2681, 15%) were the prioritised indications for DRLs in the chest region because of their prevalence and differences in the scanning protocol. For the abdomen and pelvis region, the most common indications identified for adult patient CT examinations were suspicion of AP lesion/abscess (n=14584, 27.8%), kidney stones (n=9705, 18.5%) and urothelial malignancy (n=7974, 15.2%). Based on their frequency, they were prioritised for DRLs development. The prioritised indications across the body regions were similar to those suggested in literature (Vocks and Frija, 2016). 145 University of Ghana http://ugspace.ug.edu.gh 4.2.2 Definition of Basic Imaging Requirements for Indications From the initial results (Table 4.5), it was found that most of the facilities did not have written down working protocols for imaging. Hence, in order to select the appropriate technical dose descriptors for the various indications, the basic diagnostic imaging requirements were investigated, and the concept defined. In particular, this was needful for selecting the appropriate images (without including the unnecessary ones) and their associated dose descriptors for the DRL development. The findings of the basic diagnostic imaging requirements for each of the prioritised indications are summarised and presented in the Tables 4.11-4.13. The demographic characteristics of the respondent are presented in Figure 4.5. Chief Radiogarpher Male Female Speacialist Radiologist Consultant Radiologist Principal Radiographer 4% 8% 20% 12% 8% 80% 68% Figure 4.5: Grade of the respondents Figure 4.5 shows that the majority of the respondents were male (n=20, 80%), and specialist radiologists (n=17, 68%) who are qualified in image interpretations. Therefore, their involvement provided crucial determinants for basic clinical diagnostic imaging requirements in this study. The basic diagnostic imaging requirements for CVA/stroke, head injury and brain tumour/SOL procedures in Ghana are presented in Table 4.11. 146 University of Ghana http://ugspace.ug.edu.gh Table 4.11: Basic diagnostic imaging requirements for CVA/stroke, head injury and brain tumour/SOL procedures in Ghana (Number of respondents =25). CVA/stroke Head injury Brain tumour/SOL Diagnostic imaging requirements Response Response Response Yes No Yes No Yes No n (%) n (%) n (%) n (%) n (%) n (%) Scan coverage: Scan should cover from just below 25 (100%) - - 12 (48%) 25 (100%) - the base of the skull to the vertex Scan should cover from C3 to the vertex if the base of the skull injury - - 13 (52%) - - - is suspected Scan series: None contrast 25 (100%) - 25 (100%) - - - Once (only IV contrast phase) - - - - 2 (8%) 23 (92%) Twice (both non-contrast and - - - - 23 (92%) 2 (8%) contrast phases) Image quality should be acceptable to the reporting radiologists and should meet national/international 25 (100%) - 25 (100%) - 25(100%) - standards Slice thickness- 5 mm-10 mm 16 (64%) 9 (36%) 16 (64%) 9 (36%) 17 (68%) 8 (36%) <5 mm 9 (36%) 16 (64%) 9 (36%) 16 (64%) 8 (32%) 17 (62%) Scan mode: Helical only - - - - - - Axial only - - - - - - Both 25 (25%) - 25 (25%) - 25 (25%) - Scan Technique: Low dose - - - Optimised dose 25(100%) - 25 (25%) 25(25%) High dose - - - AEC usage: Yes 25 (25%) - 25 (25%) 25(25%) No - - - Key: C3= Third Cervical vertebrae; CVA=cerebrovascular accident; SOL =space occupying lesion; AEC= automatic exposure control system, IV =intravenous, n= frequency. 147 University of Ghana http://ugspace.ug.edu.gh From Table 4.11, the basic diagnostic imaging requirements for CVA/stroke examinations in Ghana are that, the scan should cover regions just below the base of skull to the vertex, no contrast media is needed, and the scan should be performed once, while the quality of the images should be acceptable to the reporting radiologists and meet national and international requirements. Many (n=16, 64%) of the facilities preferred a scan thickness of 5-10 mm for CVA/stroke examinations while few (n=9; 36%) preferred scan thickness of less than 5 mm. In addition, the scan could be either helical or sequential (axial) and the technique should be optimised dose with AEC activated. For a head injury/trauma, about half (n=12, 48%) of the facilities required a scan coverage of regions just below the base of the skull to the vertex while the majority (n=13, 52) indicated that the scan should cover from the third cervical vertebra (C3) to the vertex. Moreover, just one optimised plain scan is required, which could be in either helical or axial modes. In addition, scan thicknesses of 5-10 mm were preferred by many (n=16, 64%) facilities while few preferred scan thicknesses of less than 5 mm. It was also necessary that images were generated with AEC and the image quality acceptable to the reporting radiologists and meet national/international standards. With CT brain tumour/SOL examinations, all the facilities preferred the scans to cover from just below the base of skull to the vertex and the majority (n=23, 92%) required a dual scan (both non-contrast and contrast phases) while few, (n=2, 8%), required one-phase scan (only IV contrast). All the respondents required acceptable resultant image qualities by the reporting radiologists that met national/international standards. Scan thicknesses of 5-10 mm were required by many (n=17, 68%), and few (n=8, 32%), required scans thicknesses of less than 5 mm. Either helical or axial scanning was acceptable to all facilities, except that the scanning technique had to be optimised and utilising AEC systems. The basic diagnostic imaging requirements for CT lung tumour, chest lesion with CKD and PE procedures in Ghana are presented in Table 4.12. 148 University of Ghana http://ugspace.ug.edu.gh Table 4.12: Basic diagnostic imaging requirements for lung tumour, chest lesion with CKD and PE CT procedures in Ghana (Number of respondents: lung tumour = 23; Chest lesion with CKD = 23, PE = 10). Lung tumour CL CKD PE Diagnostic imaging requirements Yes No Yes No Yes No n (%) n (%) n (%) n (%) n (%) n (%) Scan coverage: Scan should cover from just above lung apices to 23(100%) - 23(100%) - 10(100%) - below lung bases Scan series: Once (only non- contrast phase) - - 23(100%) - - - Once (only IV routine contrast phase) 2(9%) - - - - - Once (only IV contrast “angiogram” phase) - - - - 3(30%) 7(70%) Twice (both non-contrast and contrast - - - - 7(70%) 3(30%) “angiogram” phases Twice (both non-contrast and contrast phases) 20(87%) - - - - - Twice (IV contrast scan of the lung filed and 1(4%) - - - - - scan of the liver) Image quality should be acceptable to the reporting 23(100%) - 23(100%) - 23(100%) - radiologists and should meet national/international standards Slice thickness < 5 mm 3(13%) - 3(13%) - 10(100%) - 5 mm-10 mm 20(87%) - 20(87%) - - Scan mode: Helical only 23(100%) - 23(100%) - 10(100%) - Axial only - - - - - - Both - - - - - - Scan Low dose - - - - - - technique: Optimised dose 23(100%) - 23(100%) - 10(100%) - High dose - - - - - - AEC usage: Yes 23(100%) - 23(100%) 10(100%) - No - - - - - - Key: Chest lesion with CKD= CL CKD, AEC= automatic exposure control system, IV= intravenous, f=frequency. 149 University of Ghana http://ugspace.ug.edu.gh Table 4.12 shows that all respondents (100%) specified that lung tumour CT scans should cover regions just above lung apices to below lung bases. Few centres preferred only IV contrast- phase, while 1 (4%) required IV contrast scan of the lung filed and scan of the liver (2 sequences). Majority (n=20, 87%), of the facilities required both non-contrast and contrast phases (2 sequences). Scan thicknesses of 5 to 10 mm were required by the majority (n=20, 87%), and few (n=3, 13%) facilities preferred scan thicknesses of less than 5 mm. All the imaging facilities wanted acceptable image qualities that met national/international requirements. Additionally, all facilities also needed lung tumour CT images to be acquired in helical mode utilising AEC systems. It was evident that a CT scan for a ‘chest lesion with CKD’ examination had to cover from just above lung apices to below lung bases, with only a single non-contrast phase and scanned in helical mode. The basic requirement for most (n=20, 87%) facilities was scan thickness of 5 -10 mm, and less than 5 mm for a few (n=3, 13%). The elementary requirements for all the CT imaging facilities included dose optimisation, utilisation of AEC systems, and acceptable image qualities that meet national/international standards. It was also evident that the diagnostic imaging requirements for CT PE examinations included scanning of lung apices to below lung bases (n=10, 100%). Seven (70%) of the facilities undertaking such examinations required a two-sequence scan (both non-contrast and contrast angiogram phases), and only the angiogram phase by a few (n=3, 30%). Scan thicknesses of ≤ 5 mm with helical mode were considered necessary by all. In addition, AEC application, as well as dose optimisation, was required. All the facilities also required image qualities to be acceptable to the reporting radiologists and meet national/international standards. The basic diagnostic imaging requirements for abdomino-pelvic lesion, kidney stone and urothelial malignancy indication (CT-IVU) examinations in Ghana are presented in Table 4.12. 150 University of Ghana http://ugspace.ug.edu.gh Table 4.13: Basic diagnostic imaging requirements for abdomino-pelvic lesion, kidney stone and urothelial malignancy indication (CT- IVU) examinations in Ghana (Number of respondents: abdomino-pelvic lesion =24, kidney stone =24, urothelial malignancy =19). AP lesion Kidney stone UM (IVU) Diagnostic imaging requirements Yes No Yes Yes No Yes n (%) n (%) n (%) n (%) n (%) n (%) Scan coverage: Scan should cover from just top of higher 24(100%) - 24(100%) - 19(100%) - hemidiaphragm to below the ischium or symphysis pubis Scan series: Once (Non-contrast phase) - - 24(100%) - - - Once (only contrast phase; “oral with IV”) 2(8%) 22(92%) Twice (both non-contrast and contrast phases) 22(92%) 2(8%) - - - - *3-4 phases (pre-contrast and 2-3 other post IV - - - - 17(89%) 2(11%) contrast phases including nephrographic, corticomedullary* and excretory phases) 2 phases (slip bolus technique involving - - - 2(11%) 17(89%) nephrographic and excretory phases) Image quality should be acceptable to the reporting radiologists and 24(100%) - 24(100%) - 19(100%) - should meet national/international standards Slice thickness: ≤ 5 mm 12(50%) 12(50%) 13(54%) 11(46%) 19(100%) - 7-10 mm 12(50%) 12(50%) 11(46%) 13(53%) - 19(100%) Scan mode: Helical only 24(100%) - 24(100%) - 19(100%) - Axial only - - - - - - Both - - - - - - Scan technique: Low dose - - - - - - Optimised dose 24(100%) - 24(100%) - 19(100%) - High dose - - - - - - AEC usage: Yes 24(100%) - 24(100%) - 19(100%) - No - - - - - - Key: IV represents intravenous; UM=Urothelial malignancy; CT-IVU= computed tomography intravenous urography, AP = Abdomino-pelvic lesion; AEC= automatic exposure control system, n =frequency. 151 University of Ghana http://ugspace.ug.edu.gh From Table 4.13, one of the basic diagnostic imaging requirements for AP lesion is that the scan should cover from just top of higher hemidiaphragm to below the symphysis pubis. The majority (n=22, 92%) of the centres preferred both contrast and non-contrast phases while only a few (n=2, 8%) needed one phase scan (contrast phase involving oral and intravenous, IV). Helical scanning and AEC applications are key requirements. Also, the protocol involved the use of optimised dose and image quality acceptable to the reporting radiologists and meeting other national and international requirements. Half of the facilities required 7-10 mm scan thickness while the rest preferred ≤ 5 mm cuts. For a CT of kidney stone indication, a non-contrast scan covering from the top of the higher hemidiaphragm to below the ischium or symphysis pubis was a basic diagnostic imaging requirement by all the facilities. Scan thickness ≤ 5 mm was a basic requirement for 13 (54%) centres, while scan thickness of 7-10 mm was required by 11 (46%) centres. Other requirements included helical mode scanning and utilisation of AEC systems. As observed for CT imaging of other body regions, all the facilities required dose optimisation and acceptable image qualities that meet national/international standards. Urothelial malignancy indication examination that falls under CT-IVU required scanning regions from the top of higher hemidiaphragm to below the symphysis pubis. The majority (n =17, 89%) of the centres preferred 3 to 4 scan phases (pre-contrast and 2-3 other post IV contrast phases including nephrographic, corticomedullary and excretory phases). However, as a basic requirement, few (n = 2, 11%) centres required only two phases (split bolus technique involving nephrographic and excretory phases). Also, all the centres preferred scan thickness of ≤ 5 mm, helical scanning, optimised dose and AEC applications, and image quality acceptable to the reporting radiologists and meeting national and international requirements. The above basic diagnostic requirements were relied upon to select the appropriate data for the DRLs. 152 University of Ghana http://ugspace.ug.edu.gh 4.3 PHASE 2: Performance Characteristics Data on CT Scanners No major scanner deficiencies were found from the various QC and dose delivery validation tests undertaken. The various deviations were within the acceptable limits as suggested by many international bodies (Shefer et al., 2013; IAEA, 2012; AAPM, 2008; Institute of Physics and Engineering in Medicine, 2005; European Commission, 1999; AAPM, 1992). Details of the results are presented in Tables 4.14-4.21. Table 4.14 shows the comparison of measured and console displayed CTDIvol values for head phantom (16 mm diameter). Table 4.14: Comparison of measured and console displayed CTDIvol values for head phantom 1st Measurement 2nd Measurement kVp Measured Displayed % Deviation kVp Measured Displayed % Deviation CT(ID) CTDIvol CTDIvol (CTDIvol) CTDIvol CTDIvol (CTDIvol) (mGy) (mGy) (mGy) (mGy) CT 01 100 48.1 48.5 -0.8 120 55.4 56.0 -1.1 CT 02 110 35.7 35.9 -0.6 130 61.1 60.3 1.3 CT 03 100 37.6 36.1 4.0 120 69.4 68.2 1.7 CT 04 120 36.9 34.8 5.7 140 54.0 51.3 5.0 CT 05 100 32.3 34.1 -5.6 120 49.3 50.8 -3.0 CT 06 100 33.2 34.5 -3.9 120 52.6 51.9 1.3 CT 07 110 26.1 23.53 9.9 130 37.4 35.5 5.1 CT 08 100 41.7 40.3 3.4 120 65.2 64.3 1.4 CT 09 100 25.7 25.4 1.2 120 39.3 38.1 3.1 CT 10 110 40.1 39.4 1.7 130 61.3 62.1 -1.3 CT 12 100 45.0 43.9 2.4 120 68.2 66.8 2.1 CT 14 100 37.9 37.4 1.3 120 67.6 69.4 -2.7 CT 15 100 45.6 45.9 -0.7 120 67.7 69.3 -2.4 CT 17 100 31.1 30.1 3.2 120 47.8 48.3 -1.0 CT 18 120 14.1 13.1 7.1 140 23.3 21.4 8.2 CT 19 120 67.8 67.0 1.2 140 48.6 47.2 2.9 CT 20 100 42.3 41.4 2.1 120 63.3 65.0 -2.7 CT 21 100 28.7 29.2 -1.7 120 40.2 38.1 5.2 CT 23 110 33.6 32.3 3.9 130 55.6 54.8 1.4 CT 24 100 14.3 13.2 7.7 120 25.9 26.3 -1.5 CT 26 110 30.6 31.5 -2.9 130 47.4 46.7 1.5 CT 28 100 26.6 26.0 2.3 120 39.9 40.1 -0.5 CT 29 100 33.29 34.1 -2.4 120 51.4 52.6 -2.3 CT 30 120 18.59 20.17 -8.5 135 125.9 127.3 -1.1 CT 31 120 84.72 83.1 1.9 140 87.6 86.1 1.7 CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, kVp=peak kilovoltage, CT (ID): computed tomography identity. 153 University of Ghana http://ugspace.ug.edu.gh The CTDIvol delivery accuracy tests involving the head phantom showed that the discrepancies between measured and console displayed CTDIvol values were all within the acceptable deviation limit of 20% of the displayed manufacturer’s specifications (IAEA, 2012, AAPM, 2008). The first measurements recorded a deviation range from -8.5% to 9.9%, while the second measurements ranged from -2.7% to 8.2%. These results suggest that the scanners were delivering acceptable CTDIvol readings on head projections on the console. The inter-comparison results of the measured and console displayed CTDIvol values for the body phantom (32 mm diameter) are presented in Table 4.15. Table 4.15: Comparison of measured and displayed CTDIvol values for body phantom 1st Measurement 2nd Measurement kVp Measured Displayed % kVp Measured Displayed % Deviation CT(ID) CTDIvol CTDIvol Deviation CTDIvol CTDIvol (CTDIvol) (mGy) (mGy) (CTDIvol) (mGy) (mGy) CT 01 110 17.8 17.2 3.4 130 27.5 26.7 2.9 CT 02 110 7.9 8.0 -1.3 130 10.9 11.2 -2.8 CT 03 100 15.7 16.1 -2.5 120 18.9 19.0 -0.5 CT 04 120 19.5 17.8 8.7 140 22.1 21.3 3.6 CT 05 100 12.6 12.7 -0.8 130 18.8 18.2 3.2 CT 06 100 15.3 14.3 6.5 120 19.4 20.2 -4.1 CT 07 110 11.9 11.3 5.0 130 17.2 15.9 7.6 CT 08 100 26.2 26.9 -2.7 120 28.7 28.1 2.1 CT 09 100 9.0 9.4 -4.4 120 15.4 16.0 -3.9 CT 10 110 10.6 11.1 -4.7 130 16.4 15.5 5.5 CT 12 100 19.1 18.9 1.0 120 23.5 23.4 0.4 CT 14 100 25.6 25.0 2.3 120 27.4 27.6 -0.7 CT 15 100 14.1 14.3 -1.4 120 17.9 18.1 -1.1 CT 17 100 16.9 17.5 -3.6 120 20.1 21.4 -6.5 CT 18 120 11.9 11.2 5.9 140 17.7 17.9 -1.1 CT 19 120 16.5 16 3.0 140 24.3 23.9 1.6 CT 20 100 11.3 11.5 -1.8 120 17.4 18.7 -7.5 CT 21 100 7.9 8.3 -5.1 120 11.3 10.7 5.3 CT 23 110 10.9 10.6 2.8 130 16.6 15.3 7.8 CT 24 100 7.9 7.2 8.9 120 14.5 14.8 -2.1 CT 26 110 9.8 9.0 8.2 130 16.0 15.7 1.9 CT 28 100 26.6 26.2 1.5 120 23.4 24.2 -3.4 CT 29 100 17.3 18.2 -5.2 120 26.3 27.0 -2.7 CT 30 120 31.4 34.1 -8.5 135 32.0 33.7 -5.3 CT 31 120 15.9 14.9 6.3 140 23.6 24.0 -1.7 CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, kVp=peak kilovoltage, CT (ID): computed tomography identity. 154 University of Ghana http://ugspace.ug.edu.gh From the CTDIvol dose delivery accuracy test using body phantom (32 mm diameter), the measured and the console displayed values were within the acceptable deviation limit (≤ ± 20% of the displayed manufacturer’s specifications) as recommended by IAEA (2012) and AAPM (2008). The minimum observed deviations for the first measurements ranged from - 8.5% to 8.9%, while the second measurements ranged from -7.5% to 7.8%. The findings indicated a satisfactory dose delivery accuracy for body projections which was important in the utilisation of dose descriptors in the DRL study. The results of the CTDIvol dose delivery reproducibility tests are also presented in Table 4.16. Table 4.16: CTDIvol dose delivery reproducibility for both head and body phantoms Head phantom Body phantom CT(ID) kVp Measured CTDIvol (mGy) kVp Measured CTDIvol (mGy) 1 2 3 COV (%) 1 2 3 COV (%) CT 01 100 48.1 48.1 48.2 0.12 110 17.8 17.6 17.7 0.56 CT 02 110 35.7 35.7 35.7 0.00 110 7.9 7.9 7.9 0.00 CT 03 100 37.6 37.5 37.6 0.15 100 15.7 15.6 15.7 0.37 CT 04 120 36.9 36.8 36.8 0.16 120 19.5 19.6 19.4 0.51 CT 05 100 32.3 32.4 32.3 0.18 100 12.6 12.4 12.5 0.80 CT 06 100 33.2 33.2 33.1 0.17 100 15.3 15.3 15.3 0.00 CT 07 110 26.1 26.1 26.1 0.00 110 11.9 11.9 11.9 0.00 CT 08 100 41.7 41.6 41.6 0.14 100 26.2 26.3 26.3 0.22 CT 09 100 25.7 25.7 25.6 0.22 100 9.0 9.1 9.1 0.64 CT 10 110 40.1 40.2 40.1 0.14 110 10.6 10.5 10.6 0.55 CT 12 100 45.0 45.1 45.0 0.13 100 19.1 19.1 19.2 0.30 CT 14 100 37.9 37.9 37.9 0.00 100 25.6 25.5 25.5 0.23 CT 15 100 45.6 45.5 45.6 0.13 100 14.1 14.2 14.1 0.41 CT 17 100 31.1 31.1 31.2 0.19 100 16.9 16.9 16.9 0.00 CT 18 120 14.1 14.2 14.2 0.41 120 11.9 11.9 11.9 0.00 CT 19 120 67.8 67.7 67.8 0.09 120 16.5 16.6 16.5 0.35 CT 20 100 42.3 42.4 42.3 0.14 100 11.3 11.4 11.3 0.51 CT 21 100 28.7 28.7 28.6 0.20 100 7.9 7.9 7.9 0.00 CT 23 110 33.6 33.6 33.7 0.17 110 10.9 10.9 10.9 0.00 CT 24 100 14.3 14.4 14.3 0.40 100 7.9 7.9 7.9 0.00 CT 26 110 30.6 30.6 30.6 0.00 110 9.8 9.7 9.7 0.59 CT 28 100 26.6 26.6 26.6 0.00 100 26.6 26.8 26.7 0.37 CT 29 100 33.29 33.29 33.29 0.00 100 17.3 17.3 17.3 0.00 CT 30 120 18.59 18.59 18.59 0.00 120 31.4 31.5 31.4 0.18 CT 31 120 84.72 84.73 84.72 0.01 120 15.9 15.9 15.9 0.00 CTDIvol: volume weighted computed tomography dose index, DLP: dose length product; CoV: coefficient of variation, kVp: peak kilovoltage, CT (ID): computed tomography identity. 155 University of Ghana http://ugspace.ug.edu.gh In general, dose delivery reproducibility provides information on the ability of a scanner to consistently generate similar dose outputs over time under the same conditions (AAPM, 2008). The variations in dose output observed (Table 4.16) over three different exposures under same conditions were within the acceptable limit of ≤ ± 5% or within a coefficient of variation (COV) of ≤ 0.05 (AAPM,1992). Hence, all the scanners passed the dose delivery reproducibility tests and were reliable for generating dose outputs for the DRL study. Table 4.17 shows results of the geometric efficiency (GE) assessments undertaken at the study sites. Table 4.17: Measured geometric efficiency values for CT scanners 1st Measurement 2nd Measurement CT(ID) kVp Measured Geometric kVp Measured Geometric Efficiency Efficiency (%) (%) CT 01 100 87.9 120 86.3 CT 02 110 88.6 130 87.8 CT 03 100 83.2 120 85.2 CT 04 120 100 140 100 CT 05 100 76.3 120 79.2 CT 06 100 80.4 120 87.9 CT 07 120 82.0 140 75.2 CT 08 100 70.8 120 73.1 CT 09 100 100.0 120 76.0 CT 10 110 79.5 135 80.4 CT 12 100 76.7 120 78.2 CT 14 100 75.2 120 77.3 CT 15 100 87.3 120 88.8 CT 17 100 71.7 120 81.8 CT 18 120 87.3 140 95.6 CT 19 120 71.4 140 73.4 CT 20 100 88.7 120 88.6 CT 21 100 87.2 120 89.1 CT 23 110 78.3 130 84.1 CT 24 100 80.5 120 96.0 CT 26 110 90.4 130 97.5 CT 28 100 80.1 120 80.3 CT 29 100 71.0 120 83.3 CT 30 120 74.0 135 71.8 CT 31 120 72.4 140 100 kVp=peak kilovoltage, CT (ID): computed tomography identity. 156 University of Ghana http://ugspace.ug.edu.gh The geometric efficiency (GE) of a detector system is the percentage of the radiation quanta incident on the detector in a given interval. The recommended GE acceptable limit is >70% (Shefer et al., 2013, IAEA, 2012; AAPM, 2008). In this study, the minimum and maximum recorded GEs were 70.8% and 100%, respectively, for two measurements across the facilities. Therefore, all the scanners were performing satisfactorily in this regard. The outcomes of the tube voltage accuracy testing at the various CT facilities are shown in Table 4.18. Table 4.18: Measured kVp accuracy values for CT scanners 1st Measurement 2nd Measurement CT Centre I kVp M kVp % I kVp M kVp % (ID) Deviation Deviation CT 01 100 97.9 2.1 120 122.2 -1.8 CT 02 110 111.9 -1.7 130 132.1 -1.6 CT 03 100 101.3 -1.3 120 116.5 3.0 CT 04 120 117.6 2.0 140 136.3 2.7 CT 05 100 101.0 -1.0 120 122.0 -1.6 CT 06 100 101.7 -1.7 120 121.1 -0.9 CT 07 120 117.5 2.1 140 138.1 1.4 CT 08 100 101.6 -1.6 120 117.2 2.4 CT 09 100 99.9 0.1 120 119.8 0.2 CT 10 110 107.5 2.3 135 134.1 0.7 CT 12 100 99.1 0.9 120 118.3 1.4 CT 14 100 102.8 -2.7 120 119.6 0.3 CT 15 100 102.5 -2.4 120 122.1 -1.7 CT 17 100 102.0 -2.0 120 117.4 2.2 CT 18 120 118.6 1.2 140 137.1 2.1 CT 19 120 119.9 0.1 140 139.7 0.2 CT 20 100 102.8 -2.7 120 118.7 1.1 CT 21 100 97.8 2.2 120 123.3 -2.7 CT 23 110 111.8 -1.6 130 126.3 2.9 CT 24 100 102.6 -2.5 120 116.8 2.7 CT 26 110 111.7 -1.5 130 132.5 -1.9 CT 28 100 101.5 -1.5 120 118.5 1.3 CT 29 100 102.4 -2.3 120 121.7 -1.4 CT 30 120 121.3 -1.1 135 137.8 -2.0 CT 31 120 121.6 -1.3 140 138.8 0.9 kVp=peak kilovoltage. I kVp = Indicated kVp; M kVp = Measured kVp; % = percentage; CT (ID): computed tomography identity. 157 University of Ghana http://ugspace.ug.edu.gh Tube voltage accuracy testing was undertaken to ensure optimal tube potential for each x-ray exposure and that the peak energy of the output beam did not differ significantly to ensure an acceptable image is produced. All the scanners (Table 4.18) recorded acceptable tube voltage accuracy deviations consistent with the IAEA and AAPM recommended limit of ± 5% nominal value (IAEA, 2012; AAPM, 2008). The minimum and maximum obtained deviations for the first tube voltage measurements were -2.7% and 2.3%, while the minimum and maximum deviations for the second measurements were 2.7% and 3.0%, respectively. The results further indicated that all the scanners optimally displayed values which were integral to a successful DRL development. Table 4.19 shows the measured half value layer (HVL) values for the various CT scanners. Table 4.19: Measured HVL values for CT scanners CT (ID) kVp Measured HVL (mmAl) Acceptable deviation value CT 01 130 7.85 ≥ 2.5 mm Al for kV ˃100 CT 02 130 7.65 ≥ 2.5 mm Al for kV ˃100 CT 03 120 7.51 ≥ 2.5 mm Al for kV ˃100 CT 04 140 8.81 ≥ 2.5 mm Al for kV ˃100 CT 05 120 7.34 ≥ 2.5 mm Al for kV ˃100 CT 06 120 7.32 ≥ 2.5 mm Al for kV ˃100 CT 07 140 8.79 ≥ 2.5 mm Al for kV ˃100 CT 08 120 7.51 ≥ 2.5 mm Al for kV ˃100 CT 09 120 7.57 ≥ 2.5 mm Al for kV ˃100 CT 10 135 7.83 ≥ 2.5 mm Al for kV ˃100 CT 12 120 7.45 ≥ 2.5 mm Al for kV ˃100 CT 14 120 7.61 ≥ 2.5 mm Al for kV ˃100 CT 15 120 7.62 ≥ 2.5 mm Al for kV ˃100 CT 17 120 7.53 ≥ 2.5 mm Al for kV ˃100 CT 18 140 8.87 ≥ 2.5 mm Al for kV ˃100 CT 19 140 8.84 ≥ 2.5 mm Al for kV ˃100 CT 20 120 7.54 ≥ 2.5 mm Al for kV ˃100 CT 21 120 7.51 ≥ 2.5 mm Al for kV ˃100 CT 23 130 7.98 ≥ 2.5 mm Al for kV ˃100 CT 24 120 7.61 ≥ 2.5 mm Al for kV ˃100 CT 26 130 8.83 ≥ 2.5 mm Al for kV ˃100 CT 28 120 7.64 ≥ 2.5 mm Al for kV ˃100 CT 29 120 7.54 ≥ 2.5 mm Al for kV ˃100 CT 30 135 8.78 ≥ 2.5 mm Al for kV ˃100 CT 31 140 8.85 ≥ 2.5 mm Al for kV ˃100 kVp: peak kilovoltage, HVL: half value layer; CT (ID): computed tomography identity. 158 University of Ghana http://ugspace.ug.edu.gh Table 4.19 shows that the measured HVLs in aluminium thickness (mm Al) equivalent were all acceptable according to the United States Food and Drug Administration (2017) recommendations, which requires ≥ 2.5 mm Al for kVp ˃100. The minimum and maximum HVLs were 7.32 mm Al at 120 kV and 8.79 mm Al at 140 kV, respectively. The acceptable outcome is important because low filtration gives unnecessary radiation to patients while higher filtration leads to beam hardening minimising skin dose (AL-Jasim et al., 2017). The measured CT number for water in Hounsfield Unit (HU) and image noise in the various CT scanners are presented in Table 4.20. Table 4.20: Measured CT number for water and image noise in scanners CT number and acceptable deviation Image noise and acceptable deviation CT Measured Acceptable Noise Acceptable limit Remarks (ID) water CT deviation Remarks level number (HU) limit (HU) CT 01 1.65 ≤ ±4 Pass 5.920 ≤ ±10% baseline value Pass CT 02 1.00 ≤ ±4 Pass 6.956 ≤ ±10% baseline value Pass CT 03 0.00 ≤ ±4 Pass 4.945 ≤ ±10% baseline value Pass CT 04 0.00 ≤ ±4 Pass 5.130 ≤ ±10% baseline value Pass CT 05 2.00 ≤ ±4 Pass 6.001 ≤ ±10% baseline value Pass CT 06 2.50 ≤ ±4 Pass 6.105 ≤ ±10% baseline value Pass CT 07 0.00 ≤ ±4 Pass 3.231 ≤ ±10% baseline value Pass CT 08 0.00 ≤ ±4 Pass 5.988 ≤ ±10% baseline value Pass CT 09 0.70 ≤ ±4 Pass 5.911 ≤ ±10% baseline value Pass CT 10 0.00 ≤ ±4 Pass 6.337 ≤ ±10% baseline value Pass CT 12 1.45 ≤ ±4 Pass 5.989 ≤ ±10% baseline value Pass CT 14 4.00 ≤ ±4 Pass 8.710 ≤ ±10% baseline value Pass CT 15 4.00 ≤ ±4 Pass 9.830 ≤ ±10% baseline value Pass CT 17 0.50 ≤ ±4 Pass 8.960 ≤ ±10% baseline value Pass CT 18 0.80 ≤ ±4 Pass 8.800 ≤ ±10% baseline value Pass CT 19 1.03 ≤ ±4 Pass 9.660 ≤ ±10% baseline value Pass CT 20 0.00 ≤ ±4 Pass 9.800 ≤ ±10% baseline value Pass CT 21 0.97 ≤ ±4 Pass 6.880 ≤ ±10% baseline value Pass CT 23 4.00 ≤ ±4 Pass 7.830 ≤ ±10% baseline value Pass CT 24 0.50 ≤ ±4 Pass 4.990 ≤ ±10% baseline value Pass CT 26 4.00 ≤ ±4 Pass 6.780 ≤ ±10% baseline value Pass CT 28 2.00 ≤ ±4 Pass 8.560 ≤ ±10% baseline value Pass CT 29 2.00 ≤ ±4 Pass 9.230 ≤ ±10% baseline value Pass CT 30 1.00 ≤ ±4 Pass 7.400 ≤ ±10% baseline value Pass CT 31 0.00 ≤ ±4 Pass 5.330 ≤ ±10% baseline value Pass HU: Hounsfield Unit; CT (ID): computed tomography identity. 159 University of Ghana http://ugspace.ug.edu.gh The attenuation coefficient of radiation within a tissue is used during CT reconstruction to produce a grayscale image (DenOtter and Schubert, 2019). Based on a linear transformation of the original linear attenuation coefficient, the HU scale is calculated as gray tones to project image details (DenOtter and Schubert, 2019). The HU values are associated as CT number scale and are checked regularly during normal QC tests of the CT system. To test for the HU display of the scanners, the CT number in water was evaluated for each of them which according to the literature is 0 HU (Pan, 2009). However, there is an acceptable limit of ± 4 HU from which optimally functional scanners should be able to display x-ray attenuation information (AAPM, 2008). The minimum (0.0 HU) and maximum (4.0 HU) measured values from Table 4.20 suggest that all scanners passed the CT number (water) tests. Image noise in CT imaging is “the degree of uncertainty in measuring attenuation of an x-ray beam passing through a patient” (Zarb et al., 2010, p. 3). Excessive image noise can degrade image quality; hence, it has to be corrected whenever detected (Zarb et al., 2010). The image noise levels ranging from 3.2 to 9.8 observed across the scanners were all within the tolerable limit of ≤ ±10% baseline value as recommended by the IAEA (IAEA, 2012), and thus suggest optimal CT noise levels in images produced by the scanners. The results of the image uniformity (homogeneity) analyses performed at the various CT facilities are presented in Table 4.21. 160 University of Ghana http://ugspace.ug.edu.gh Table 4.21: Image uniformity (homogeneity) findings at different locations CT ID 12 15 18 21 Centre Diff Diff Diff Diff o'clock o'clock o'clock o'clock HU C 12 C15 C18 C 21 CT 01 2.189 2.321 1.053 1.679 1.35 0.839 0.971 -0.297 0.329 CT 02 2.047 1.587 0.909 1.872 1.407 0.84 0.181 -0.498 0.465 CT 03 2.159 1.909 1.443 1.719 1.197 0.962 0.713 0.247 0.522 CT 04 2.473 1.874 1.537 2.295 1.138 1.335 0.736 0.399 1.157 CT 05 2.143 2.209 1.459 2.318 0.683 1.460 1.526 0.776 1.635 CT 06 3.364 3.130 3.108 2.604 1.120 2.244 2.010 1.989 1.484 CT 07 1.785 1.377 0.820 1.405 0.871 0.914 0.507 -0.050 0.535 CT 08 2.129 1.965 0.869 1.515 1.466 0.663 0.499 -0.597 0.050 CT 09 2.907 2.967 2.690 2.840 1.200 1.707 1.767 1.49 1.639 CT 10 2.392 1.953 1.587 2.221 1.078 1.314 0.875 0.509 1.143 CT 12 1.788 1.864 1.344 2.021 1.255 0.534 0.609 0.09 0.766 CT 14 1.977 1.58 1.385 1.624 1.301 0.676 0.278 0.084 0.323 CT 15 2.641 2.223 1.619 1.770 1.270 1.371 0.955 0.348 0.500 CT 17 2.461 2.008 0.995 2.045 1.114 1.347 0.894 -0.12 0.93 CT 18 2.541 1.483 1.61 1.587 1.319 1.222 0.165 0.291 0.269 CT 19 2.452 1.726 1.289 2.05 1.448 1.004 0.278 -0.159 0.601 CT 20 2.309 1.851 1.025 1.639 1.097 1.212 0.753 -0.072 0.541 CT 21 2.430 2.109 1.275 1.574 1.255 1.174 0.854 0.02 0.318 CT 23 2.391 2.144 1.404 1.953 1.181 1.211 0.963 0.223 0.772 CT 24 1.869 1.963 1.601 1.566 1.267 0.602 0.696 0.334 0.298 CT 26 2.159 2.164 1.144 1.778 0.942 1.217 1.222 0.202 0.836 CT 28 2.687 2.815 2.064 2.642 0.822 1.866 1.994 1.242 1.820 CT 29 2.433 2.143 1.624 1.983 1.277 1.156 0.866 0.347 0.706 CT 30 2.403 2.488 1.350 2.501 1.246 1.157 1.243 0.105 1.255 CT 31 1.737 1.927 1.174 2.008 1.136 0.802 0.791 0.038 0.872 HU: Hounsfield Unit; CT (ID): computed tomography identity. The image homogeneity test results (Table 4.21), describe how uniform a homogenous material appears, and indicated that the scanners operated within the IAEA tolerable recommended limits of ±10 HU for a head and body phantom (IAEA, 2012). In particular, the deviations of uniformities at all the five ROIs (12 o'clock, 15 o'clock, 18 o'clock, 21 o'clock and the centre) of the respective images were all within the ±10 HU of the central ROI (HU). 161 University of Ghana http://ugspace.ug.edu.gh 4.4 PHASES 3 & 4: CT Dose and Image Quality Assessment and Estimation of DRLs This section presents the results of the sample demographics, scan parameters, and DRL values for various indications as well as the developed tool for dose monitoring, following a successful data validation. The age, gender and weight of the patients are shown in Table 4.22. Table 4.22: Demographic characteristics of patients’ data Indication Age (years) Weight (kg) Gender Mean (Range) Mean (Range) Male Female CVA or stroke 60.7±15.0(18-96) 71.4±8.9(50-90) 239(42.8%) 261(52.2%) Head trauma/Injury 41.7±18.2(18-97) 72.6±9.9(50-90) 332(66.4%) 168(33.6%) Brain tumour/ SOL 44.4±17.5(18-92) 73.3±9.7(50-90) 227(45.4%) 273(54.6%) Lung tumour/cancer 54.0±16.7(18-93) 74.4±8.7(50-90) 236(51.3%) 224(48.7%) Chest lesion with CKD 52.1±16.2(18-87) 74.1±8.9(50-90) 226(49.1%) 234(50.9%) Abdominopelvic lesion 51.6±16.1(18-96) 76.6±9.5(50-90) 228(47.5%) 252(52.5%) Kidney stones 49.9±16.2(18-94) 74.8±9.4(50-90) 249(51.9%) 231(48.1%) Urothelial malignancy 45.9±15.7(18-89) 73.4±8.4(50-90) 226(59.5%) 154(40.5%) Pulmonary embolism 58.8±14.1(23-88) 74.4±9.4(55-90) 86(43.0%) 114(57.0%) Overall 50.5±17.4(18-97) 73.8±9.3(50-90) 2049(51.7%) 1911(48.3%) CVA: cerebrovascular accident, SOL: space occupying lesion, CKD: chronic kidney disease; kg: kilogram The CT dose descriptor and image quality data were collected from 25 scanners following the exclusion of non-functional, specially dedicated and technically challenged (inability to display dose descriptors) CT scanners (Figure 3.3). This number represented 71.4% of the CT scanners in Ghana. The ICRP has recommended NDRL surveys to include at least 30-50% of facilities, where all the facilities could not be used (ICRP, 2017). The large number of CTs used in this study gives a representative sample. In total, 3,960 patient data sets were used to develop the indication-based DRLs. Majority (2,049, 51.7%) of the study samples were males. The overall mean age and weight of participants were 50.5 ± 17.4 years and 73.8 ± 9.3 kg, respectively, which was similar to a reference mean weight of an adult population (ICRP, 2017). The scan parameters and factors used for generating the dose output data for the respective indications are presented in Table 4.23. 162 University of Ghana http://ugspace.ug.edu.gh Table 4.23: Descriptive statistics of scanning parameters used to acquire images Indication Scan parameters (mean ± SD) Tube voltage Tube loading Pitch Trot Slice thickness No. of slices Scan length No. of Phantom kVp mAs (s) (mm) per series (mm) series* Diameter (cm) CVA/stoke 121.8 ± 7.4 238.0 ± 80.6 0.75 ± 0.2 0.97 ± 0.3 4.30 ± 1.1 51.4 ± 61.9 161.9 ± 24.0 1 16 HT/I 120.3 ± 6.4 229.4 ± 73.5 0.72 ± 0.2 0.96 ± 0.4 4.40 ± 1.1 61.2 ± 71.1 197.1 ± 40.0 1 16 BT/S 121.0 ± 7.8 238.2 ± 83.1 0.74 ± 0.2 1.01 ±1.5 4.40 ± 1.1 53.1 ± 66.4 164.5 ± 21.0 2 16 LT/C 119.0 ± 7.4 114.0 ± 76.8 1.16 ± 0.2 0.71 ±0.3 5.50 ± 2.5 93.2 ± 125.0 313.4 ± 43.0 1-2 32 CLCKD 119.0 ± 7.3 114.0 ± 74.7 1.15 ± 0.2 0.70 ± 0.3 5.50 ± 2.5 91.7 ± 123.0 310.2 ± 43.0 1 32 APL 118.7 ± 7.5 137.0 ± 91.4 1.11 ± 0.3 0.76 ± 0.3 5.40 ± 2.6 162.0 ± 208.0 459.9 ± 43.0 1-2 32 KS 118.0 ± 8.3 138.4 ± 98.4 1.12 ± 0.3 0.80 ± 0.3 5.30 ± 2.5 152.0 ± 187.0 455.1 ± 41.0 1 32 UM 118.0 ± 8.5 114.0 ± 82.0 1.09 ± 0.3 0.71 ± 0.4 5.10 ± 2.4 165.4 ± 200.0 461.2 ± 56.0 2-4 32 PE 117.8 ± 4.0 167.5 ± 92.9 1.20 ± 0.3 0.64 ± 0.2 2.20 ± 1.7 294.6 ± 216.0 303.7 ± 41.0 1-2 32 HT/I=Head trauma/Injury; BT/S=Brain tumour/SOL; LT/C= Lung tumour/cancer; CLCKD=Chest lesion with CKD; APL=Abdomino-pelvic lesion, KS=Kidney stones; UM=Urothelial malignancy, PE=Pulmonary angiogram, SD=standard deviation, kVp=peak kilovoltage, mAs=milliampere-second, Trot=rotation time. *For a number of series, the scout or scanogram sequences and monitoring phases (e.g. in the cases of PE) were not included as it is the situation for all established DRLs. 163 University of Ghana http://ugspace.ug.edu.gh For the CVA/stroke indications, the mean tube voltage was 121.8 ± 7.4 kVp while the tube loading, pitch and gantry rotation time (Trot) were 238.0 ± 80.6 mAs, 0.75 ± 0.2, and 0.97 ± 0.3 s, respectively. The average scan slice thickness used was 4.3 ± 1.1 mm and an average of 51.4 ± 61.9 images were acquired over an average scan range of 161.9 ± 24 mm. The mean scan parameters used in the case of head trauma or injury were 120.3 ± 6.4 (kVp), 229.4 ± 73.5 (mAs), 0.72 ± 0.20 (pitch), 4.40 ± 1.10 mm (slice thickness) and 197.1 ± 40.0 mm (scan length). Both CVA/stroke and head injury procedures were scanned using a single sequence while brain tumour/SOL procedures were scanned twice. The scan parameters of brain tumour/SOL were closely similar to CVA/stroke indications except the number of scan sequences. A 16 cm phantom was utilised for generating dose outputs for all head related indications such as head injury, CVA/stroke and brain tumour/SOL while a larger mean scan length was utilised in the former indication compared to the rest. Hence, a correspondingly larger DLP was envisaged in the dose output. The mean scan parameters for lung tumour/cancer imaging were 119.0 ± 7.4 kVp, 114.0 ± 76.8 mAs, 1.16 ± 0.2 (pitch), 5.5 ± 2.5 mm (slice thickness) and 313.4 ± 43.0 mm (scan length). Lung tumour/cancer and chest lesion with CKD parameters were similar except that scan sequences for the latter was 1, and the former was 1-2. Moreover, the mean scan parameters for AP lesion, kidney stones and urothelial malignancy (CT-IVU) were closely related except that kidney stone procedures were scanned in a single sequence mode while AP lesion and CT-IVU procedures were scanned using 1-2 and 2-4 sequences, respectively. PE scans were also generated with parameters which included voltage (117.8 ± 4.0 kVp), tube loading (167.5 ± 92.9 mAs), pitch (1.20 ± 0.3), slice thickness (2.2 ± 1.7 mm), scan length (303.7 ± 41.0 mm) and scan sequence of 1-2. All the body procedures were undertaken with a 32-cm diameter phantom in conformity with the international norm (AAPM, 2008). 164 University of Ghana http://ugspace.ug.edu.gh The descriptive statistics of scanning mode, contrast and AEC utilisation are described in Table 4.24. Table 4.24: Descriptive statistics of scanning mode and contrast and AEC utilisation Indication Contrast Scanning mode AEC Used Unused Helical Axial Used Unused CVA/stroke - 500 (100%) 344 (68.8%) 156 (32.2%) 241(48.2%) 259 (51.8%) HT/I - 500 (100%) 389 (77.8%) 111 (22.2%) 219 (43.8%) 281(56.2%) BT/S 500 (100%) - 340 (68.0%) 160 (32.0%) 190 (38.9%) 310 (62.0%) LT/C 500 (100%) - 460 (100%) - 440 (95.7%) 20 (4.3%) CLCKD - 500 (100%) 640 (100%) - 440 (95.7%) 20 (4.3%) APL 480 (100%) - 480 (100%) - 400 (83.3%) 80 (16.7%) KS 480 (100%) - 480 (100%) - 400 (83.3%) 80 (16.7%) UM 380 (100%) - 380 (100%) - 380 (100%) - PE 200 (100%) - 200 (100%) - 200 (100%) - HT/I=Head trauma/Injury; BT/S=Brain tumour/SOL; LT/C=Lung tumour/cancer; CLCKD=Chest lesion with CKD; APL= Abdomino-pelvic lesion, KS=Kidney stones UM= Urothelial malignancy, PE= Pulmonary angiogram. Table 4.24 indicates that contrast agents were not used in routine CVA/stroke, head injury, chest lesion with CKD and kidney stone examinations. They were, however, used for the rest of the indications. Moreover, for CVA/stroke, head injury and brain tumour/SOL indications, it was found that 156 (32.2%), 111 (22.2%), 160 (32.0%), were scanned in axial mode. However, the rest (representing the majority) were all scanned in helical mode. AEC systems were not utilised in some indication-based procedures during scanning, albeit few in the body examinations. According to Higaki et al., (2019) and Merzan et al., (2017), such examinations may not benefit from the radiation dose reduction advantages of AEC configurations and hence, would require dose optimisation in those CT facilities. 165 University of Ghana http://ugspace.ug.edu.gh Following a successful data validation, where all dose descriptors were in statistical control (within the upper and lower control limits) (Appendix VIII), the DRL values for each indication were projected. The descriptive statistics and projected DRL values for CT CVA indications with respect to CTDIvol are presented in Table 4.25 and Figure. 4.6, respectively, while those of DLP are shown in Table 4.26 and Figure 4.7. A comparison of developed DRL values (CTDIvol and DLP) for CVA with international values is shown in Table 4.27. Table 4.25: Descriptive statistics of representative CTDIvol values for CVA CT Total AD Min. Median Max. Median SD scanners CTDIvol (mGy) CTDIvol (mGy) CTDIvol (mGy) 25 50.9 28.1 91.9 20.1 CT14 CT8 CT 1 CT12 CT17 CT29 CT30 CT28 75th percentile=77.2 mGy CT10 25th percentile=39.3 mGy CT19 CT23 CT15 CT21 CT6 CT18 CT3 CT20 CT4 CT2 CT26 CT 9 CT5 CT31 CT24 CT7 0 20 40 60 80 100 CTDIvol (mGy) AD= Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.6: Distribution of representative CTDIvol values for CVA CT. 166 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.26: Descriptive statistics of representative DLP values for CVA CT Total AD DLP (mGy.cm) Min. median DLP Max. median DLP SD scanners (mGy.cm) (mGy.cm) 25 845.0 513.4 1540.2 350.2 CT17 CT 1 CT14 CT29 CT30 CT12 CT8 CT21 75th percentile CT28 =1312.9 mGy.cm CT10 25th percentile CT19 =635.2 mGy.cm CT23 CT15 CT4 CT3 CT2 CT20 CT6 CT26 CT5 CT31 CT18 CT 9 CT24 CT7 0 200 400 600 800 1000 1200 1400 1600 1800 DLP (mGy.cm) The DLP values are from 1 sequence (Non-contrast examination). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.7: Distribution of representative DLP values for CVA CT Table 4.27: A comparison of CVA indication-based DRL values with international values. Country 75th percentile Image quality CTDIvol (mGy) DLP (mGy.cm) Mean SNR Current study (Ghana) 77 1313* 8.7± 2.1 Public Health England, (2016), UK 80 970* - Treier et al., (2010), CH 65 1000* - *=Single sequence procedure; CH represents Switzerland and UK represents the United Kingdom. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 167 CT Scanners University of Ghana http://ugspace.ug.edu.gh In accordance with the Rose criterion (model) (Bushberg et al., 2011), the image quality test results (SNR = 8.7± 2.1) indicated that images from which dose descriptors were used for CVA DRLs, were within the clinical requirements needed to distinguish image features at 100% certainty. The calculated 75th and 25th percentiles for CVA CT indications have been displayed in blue and green dashed lines, respectively, in both Figures 4.6 and 4.7. The results indicated that the minimum and maximum median CTDIvol values across the facilities varied by a factor of 3.3. The proposed NDRL for CVA with respect to CTDIvol was approximately 77 mGy. The achievable CTDIvol which is the 50 th percentile of the calculated doses was approximately 51 mGy. The projected NDRL for CVA with respect to DLP was approximately 1313 mGy.cm and the achievable DLP dose was approximately 845 mGy.cm. Compared to internationally established values, it was observed that Ghana’s NDRL values for CVA (CTDIvol = 77 mGy; DLP = 1313 mGy.cm) were higher than those reported by Treier et al. (2010) in Switzerland, although SNR values were not reported in the latter results. Comparatively, the UK CTDIvol value (80 mGy) was higher but with an interestingly lower DLP (970 mGy.cm). Since DLP is a product of CTDIvol and scan length, it may be projected that longer scan lengths are used in Ghana compared to their UK counterparts, if all other factors remain the same. The descriptive statistics and projected DRL values for head trauma/injury CT with respect to CTDIvol are presented in Table 4.28 and Figure 4.8, respectively, while that of DLP are shown in Table 4.29 and Figure 4.9. A comparison of developed DRL values (CTDIvol and DLP) for head trauma/injury with international values is shown in Table 4.30. 168 University of Ghana http://ugspace.ug.edu.gh Table 4.28: Descriptive statistics of representative CTDIvol values for head trauma/injury CT Total AD CTDIvol Min. median CTDIvol Max. median CTDIvol SD scanners (mGy) (mGy) (mGy) 25 51.7 27.0 91.9 18.8 CT14 CT1 CT12 CT17 CT29 CT30 CT28 CT8 75th percentile=76.3 mGy CT10 25th percentile=40.3 mGy CT19 CT23 CT2 CT15 CT6 CT18 CT3 CT21 CT20 CT4 CT5 CT26 CT9 CT24 CT31 CT7 0 20 40 60 80 100 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.8: Distribution of representative CTDIvol values for head trauma or injury CT 169 CT scanners University of Ghana http://ugspace.ug.edu.gh Table 4.29: Descriptive statistics of representative DLP values for head trauma/injury CT Total AD DLP (mGy.cm) Min. median DLP Max. median DLP SD scanners (mGy.cm) (mGy.cm) 25 1206.5 485.2 2331.2 480.3 CT30 CT1 CT29 CT12 CT10 CT17 75th percentile CT14 =1595.7 mGy.cm CT23 25th percentile CT2 = 797.7 mGy.cm CT15 CT19 CT28 CT21 CT8 CT20 CT3 CT24 CT4 CT18 CT6 CT5 CT26 CT9 CT31 CT7 0 500 1000 1500 2000 2500 DLP (mGy.cm) The DLP values are from 1 sequence (Non-contrast examination). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4. 9: Distribution of representative DLP values for head trauma/injury CT Table 4.30: A comparison of indication-based DRL values for head trauma/injury CT, with international values. Country 75th percentile Image quality CTDIvol (mGy) DLP (mGy.cm) Mean SNR Current study (Ghana) 77 1596* 7.1 ± 1.3 Widmark, (2018), NO 60 950* - *=Single sequence procedure; NO represents Norway. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 170 CT Scanners University of Ghana http://ugspace.ug.edu.gh The CT dose levels used for head trauma/injury examinations varied across the facilities in Ghana. The minimum and maximum median CTDIvol values varied by 3.4-fold. The mean SNR of 7.1 ± 1.3 indicated that head trauma/injury images were of acceptable image quality. The calculated 75th and 25th (CTDIvol and DLP) percentiles for CT of head trauma/injury are displayed in red and green dashed lines, respectively, in Figures 4.8 and 4.9. The findings indicated that the projected national CTDIvol DRL for CT of head trauma/injury was approximately 76 mGy, while the achievable dose was approximately 52 mGy. The corresponding projected DLP DRL was approximately 1596 mGy.cm, while the achievable DLP DRL was also approximately 1207 mGy.cm. Comparatively, this study’s DRLs for head trauma were higher than the projected NDRLs (CTDIvol = 60 mGy; DLP= 950 mGy.cm) in Norway (Widmark, 2018). Table 4.31 and Figure 4.10 show the descriptive statistics and projected CTDIvol DRL values for CT of brain tumour/SOL indication, while the descriptive statistics and projected DLP DRL values for the same indication are also presented in Table 4.32 and Figure 4.11. A comparison of these developed DRLs with international values are shown in Table 4.33. 171 University of Ghana http://ugspace.ug.edu.gh Table 4.31: Descriptive statistics of representative CTDIvol values for brain tumour/SOL CT indication Total AD CTDIvol (mGy) Min. median CTDIvol Max. median CTDIvol SD scanners (mGy) (mGy) 25 51.7 27.7 91.9 20.6 CT14 CT8 CT1 CT19 CT30 CT17 CT29 CT12 CT10 CT28 75th percentile=77.2 mGy CT2 CT23 25 th percentile=39.3 mGy CT15 CT6 CT3 CT21 CT18 CT20 CT4 CT26 CT5 CT9 CT31 CT24 CT7 0 10 20 30 40 50 60 70 80 90 100 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.10: Distribution of representative CTDIvol values for brain tumour/SOL CT indication 172 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.32: Descriptive statistics of representative DLP values for brain tumour/SOL CT indication Total AD DLP Min. median DLP Max. median DLP SD scanners (mGy.cm) (mGy.cm) (mGy.cm) 25 1845.2 985.4 3119.0 726.2 CT17 CT1 CT30 CT12 CT29 CT14 CT19 CT8 75th percentile CT2 =2696.0 mGy.cm CT21 th CT10 25 percentile CT28 =1291.3 mGy.cm CT15 CT23 CT4 CT3 CT20 CT26 CT31 CT5 CT6 CT18 CT24 CT9 CT7 0 500 1000 1500 2000 2500 3000 3500 DLP (mGy.cm) The DLP values are from 1-2 sequences (Pre- and post-contrast examinations). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.11: Distribution of representative DLP values for brain tumour/SOL CT indication Table 4.33: A comparison of indication-based DRL values for brain tumour/SOL CT, with international values. Country 75th percentile Image quality CTDIvol (mGy) DLP (mGy.cm) Mean SNR Current study (Ghana) 77 2696+ 6.7 ± 1.5 Treier et al., (2010), CH 65 1000* - *=Single sequence procedure; + =1-2 sequences (pre- and post-contrast examinations). CH represents Switzerland. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 173 CT Scanners University of Ghana http://ugspace.ug.edu.gh As seen in Figures 4.10 and 4.11, the red and green dashed lines represent the calculated 75th and 25th (CTDIvol and DLP) percentiles for brain tumour/SOL indication CT, respectively. Image quality in terms of SNR for brain tumour/SOL CT indication was 6.7 ± 1.5, which was considered good for diagnostic purposes. The observed dose variation in CTDIvol values (minimum and the maximum) was 3.3-fold. The projected NDRL for brain tumour/SOL CT with respect to CTDIvol was therefore approximately 77 mGy, while the calculated achievable dictation CTDIvol dose was approximately 52 mGy. Moreover, the projected NDRL for brain tumour/SOL CT with respect to DLP and the calculated achievable DLP dose were approximately 2696 mGy.cm and 1845 mGy.cm, respectively. Comparatively, the projected brain tumour/SOL DRL values from this study were higher than the established values (CTDIvol = 65 mGy; DLP = 1000 mGy.cm) in Switzerland as reported by Treier et al., (2010). A critical factor effecting the large variation in DRL values across the two studies was the scan sequences. Particularly, it was observed that radiographers in Ghana used 1-2 sequences (pre- and post-contrast), while their Swiss counterparts used a single sequence (post-contrast) to diagnose brain tumour/SOL conditions. It may be important for medical professionals in Ghana to explore the possibility of using a single sequence in brain tumour/SOL conditions since it is practicable in Switzerland. Its implementation may further optimise patient’s radiation levels. The projected CTDIvol DRLs (CTDIvol) for lung tumour/cancer CT indication as well as the respective descriptive statistics are presented in Table 4.34 and Figure 4.12. The corresponding descriptive statistics and projected DLP DRLs are also presented in Table 4.35 and Figure 4.13. A comparison of these developed DRL for lung tumour/cancer CT indication with international values is shown in Table 4.36. 174 University of Ghana http://ugspace.ug.edu.gh Table 4.34: Descriptive statistics of representative CTDIvol values for lung tumour/cancer CT indication Total AD CTDIvol Min. median CTDIvol Max. median CTDIvol SD scanners (mGy) (mGy) (mGy) 23 7.7 3.4 21.3 4.4 CT14 CT8 CT20 CT4 CT18 CT31 CT21 75th percentile CT3 =12.4 mGy CT12 CT6 25th percentile CT28 = 5.7 mGy CT29 CT30 CT24 CT17 CT23 CT10 CT7 CT26 CT9 CT1 CT2 CT15 0 5 10 15 20 25 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4. 12: Distribution of representative CTDIvol values for lung tumour/cancer CT indication 175 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.35: Descriptive statistics of DLP values for lung tumour/cancer CT indication Total AD Min. median Max. Median SD Scanners DLP (mGy.cm) DLP (mGy.cm) DLP (mGy.cm) 23 463.7 160.4 1290.6 310.5 CT14 CT8 CT4 CT20 CT31 CT21 th CT6 75 percentile CT3 =827.9 mGy.cm CT29 CT12 25th percentile CT18 =386.7 mGy.cm CT1 CT24 CT30 CT17 CT9 CT28 CT23 CT10 CT7 CT15 CT2 CT26 0 200 400 600 800 1000 1200 1400 DLP (mGy.cm) The DLP values are from 1-2 sequences (Pre- and post-contrast examinations). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.13: Distribution of DLP values for lung tumour/cancer CT indication Table 4.36: A comparison of indication-based DRL values for lung tumour/cancer CT, with international values. Country 75th percentile Image quality CTDIvol (mGy) DLP (mGy.cm) Mean SNR Current study (Ghana) 12 828+ 11.3 ± 3 Widmark, (2018), NO 9 350* - Public Health England, (2016), UK 12 610? - Danish Health Authority, (2015), DK 16 620? - Radiation and Nuclear Safety - 11 430* Authority, (2013), (FI) *=Single sequence procedure; + =1-2 sequences (pre- and post-contrast),? =undeclared number of sequences. NO represents Norway, UK represents United Kingdom, DK represents Denmark, FI represents Finland. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product 176 CT scanners University of Ghana http://ugspace.ug.edu.gh A mean SNR value of 11.3 ± 3 was observed across the images whose dose descriptors were used in developing DRL values for lung tumour/cancer. The result indicated a good objective image quality following successful subjective comments by radiographers. The dose levels used for diagnosing lung tumour/cancer across the facilities were wide-ranging. In particular, the CTDIvol values between the minimum and maximum varied by a factor of 6.2. The red and blue dashed lines in Figures 4.12 and 4.13 show the respective 75th and 25th (CTDIvol and DLP) percentiles for lung tumour/cancer CT indications. The suggested NDRL for lung tumour/cancer CT indication with respect to CTDIvol was approximately 12 mGy, while the observed achievable CTDIvol dose across the facilities was approximately 8 mGy. The corresponding projected DLP was approximately 828 mGy.cm and the observed achievable DLP was approximately 464 mGy.cm. Compared with available literature, this study’s DRLs were higher than the projected Norwegian NDRL values (CTDIvol = 9 mGy; DLP = 350 mGy.cm) (Widmark, 2018) and and Finish DRLs (CTDIvol = 11 mGy; DLP = 430 mGy.cm) (Radiation and Nuclear Safety Authority, 2013), respectively. The CTDIvol DRL was similar to the established national value (12 mGy) in the UK, but less than the 16 mGy reported in Denmark (Danish Health Authority, 2015). However, the projected DLP value in this study was higher than all its counterparts. In Ghana, the diagnostic imaging requirement for tumour/cancer largely required a two-sequence (pre- and post-contrast) procedure with few facilities scanning with a single sequence (Table 4.12). On the contrary, the Norway and Finland DRLs were based on a single sequence values due to their practice norm. However, the number of scan sequence from which the DLP values were established for the UK and Denmark was unclear as available literature was silent on it. 177 University of Ghana http://ugspace.ug.edu.gh For lung lesion with CKD indications, the descriptive statistics and projected DRL values are presented in terms of CTDIvol (Table 4.37 and Figure 4.14) and DLP (Table 4.38 and Figure 4.15), respectively. Table 4.37: Descriptive statistics of representative CTDIvol values for CT of chest lesion with CKD indication Total AD CTDIvol Min. median CTDIvol Max. median CTDIvol SD scanners (mGy) (mGy) (mGy) 23 7.7 3.0 21.3 4.9 CT14 CT12 CT8 CT4 CT20 CT18 th CT21 75 percentile CT31 =13.4 mGy CT3 25thCT6 percentile CT28 =5.9 mGy CT29 CT24 CT23 CT30 CT17 CT7 CT10 CT2 CT26 CT9 CT1 CT15 0 5 10 15 20 25 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.14: Distribution of representative CTDIvol values for CT of chest lesion with CKD indication 178 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.38: Descriptive statistics of DLP (mGy.cm) values for CT of chest lesion with CKD Total AD DLP Min. median DLP Max. median DLP SD Scanners (mGy.cm) (mGy.cm) (mGy.cm) 23 230.8 131.4 644.4 150.7 CT14 CT4 CT18 CT8 CT21 CT20 75 th percentile CT31 =467.3 mGy.cm CT12 th CT3 25 percentile CT6 = 195.3 mGy.cm CT29 CT24 CT17 CT1 CT30 CT7 CT28 CT23 CT9 CT10 CT2 CT26 CT15 0 100 200 300 400 500 600 700 DLP(mGy.cm) The DLP values are from 1 sequence (Non-contrast examination). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.15: Distribution of DLP values for CT of chest lesion with CKD indication Available literature (Patschan et al., 2018) indicates that patients suffering from kidney diseases such as CKD do not often undergo CT with iodinated contrast agents because it could increase serum creatinine and potentially cause nephropathy. The observed median dose levels (CTDIvol) used for this indication using non-contrast imaging varied by a factor of 7. The proposed national CTDIvol DRL was approximately 13 mGy, while the observed CTDIvol achievable dose (50th percentile) across the facilities for CT of lung lesion with CKD indication was approximately 7.7 mGy. The projected associated DLP value was approximately 467 179 CT Scanners University of Ghana http://ugspace.ug.edu.gh mGy.cm, while the achievable DLP dose was also approximately 231 mGy.cm. The mean SNR associated with the indication-specific images was 11.8 ± 2.2 which implied that the DRLs were derived from dose descriptors whose image characteristics were within an acceptable level (Bushberg et al., 2011). Moreover, no internationally established DRL value was found for chest lesion with CKD. Hence, this was the first time it has been projected. The descriptive statistics and DRL values for AP lesion are presented in Table 4.39 and Figure 4.16. for CTDIvol descriptors and in Table 4.40 and Figure 4.17 for DLP descriptors. A comparison of developed DRL values with international values is shown in Table 4.41. Table 4.39: Descriptive statistics of CTDIvol (mGy) values for CT of AP lesion indication Total AD CTDIvol (mGy) Min. median CTDIvol Max. median CTDIvol SD scanners (mGy) (mGy) 24 10.9 4.8 27.2 6.1 CT14 CT29 CT20 CT19 CT21 CT30 75th percentile CT17 =16.7 mGy CT8 CT12 25th percentile CT6 = 7.3 mGy CT4 CT31 CT3 CT28 CT7 CT18 CT23 CT24 CT9 CT2 CT26 CT10 CT15 CT1 0 5 10 15 20 25 30 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.16: Distribution of representative CTDIvol values for CT of AP lesion indication 180 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.40: Descriptive statistics of representative DLP values for CT of AP lesion indication Total AD DLP (mGy.cm) Min. median DLP Max. median DLP SD Scanners (mGy.cm) (mGy.cm) 24 1026.4 432.5 2688.8 583.4 CT29 CT14 CT20 CT21 CT8 CT31 75thCT19 percentile CT6 =1298.6 mGy.cm CT4 25thCT17 percentile CT30 =674.5 mGy.cm CT3 CT12 CT18 CT7 CT24 CT2 CT23 CT9 CT15 CT28 CT26 CT1 CT10 0 500 1000 1500 2000 2500 3000 DLP (mGy.cm The DLP values are from 1-2 sequences (Pre- and post-contrast examinations). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.17: Distribution of DLP values for CT of AP lesion indication Table 4.41: A comparison of indication-based DRL values for CT of AP lesion, with international values. Country 75th percentile Image quality CTDIvol (mGy) DLP (mGy.cm) Mean SNR Current study (Ghana) 17 1299+ 11.8 ± 2.2 Widmark, (2018), NO 11 800+ - Public Health England, (2016), UK 15 745? - Treier et al., (2010), CH 15 650* - Wachabauer et al., (2017), AU - 650* - *=Single sequence procedure; + =1-2 sequences (pre- and post-contrast),? =undeclared number of sequences. NO represents Norway, UK represents United Kingdom, CH represents Switzerland and AU represents the Australia. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 181 CT Scanners University of Ghana http://ugspace.ug.edu.gh For AP lesion, the observed CT dose levels (CTDIvol) between the minimum and maximum median values across the facilities varied by 5.7-fold. The estimated 75th and 25th (CTDIvol and DLP) percentiles for CT of AP lesions are respectively displayed in red and blue dashed lines in Figures 4.16 and 4.17. The projected national CTDIvol DRL and calculated CTDIvol achievable dose values for CT of AP lesion were approximately 17 mGy and 11 mGy, respectively. The corresponding national DLP DRL and calculated DLP achievable doses for CT of AP lesion were approximately 1299 mGy.cm and 1026 mGy.cm, respectively. A good image quality (mean SNR = 11.8 ± 2.2) was observed across the images. The projected DRL values were higher than those reported in literature by Widmark, (2018), Public Health England, (2016), Treier et al., (2010) and Wachabauer et al., (2017). Particularly, the DLP values reported by Treier et al., (2010), and Wachabauer et al., (2017) were about half of values projected in this study. The scan diagnostic imaging requirement where some radiographers in Ghana used a pre- and post-contrast protocol, while radiographers in some other countries utilised a single sequence procedure for AP lesion CT imaging could be one of the factors leading to the wide variation between the projected and literature values. Table 4.42 and Figure 4.18 present the descriptive statistics and projected CTDIvol DRLs for kidney stone indication, respectively, while the same for the DLP DRLs are also presented in Table 4.43 and Figure 4.19. A comparison of developed DRLs for kidney stone with international values is shown in Table 4.44. 182 University of Ghana http://ugspace.ug.edu.gh Table 4.42: Descriptive statistics of CTDIvol values for CT of kidney stone indication Total AD CTDIvol (mGy) Min. median CTDIvol Max. median CTDIvol SD scanners (mGy) (mGy) 24 9.9 4.4 26.6 6.1 CT14 CT29 CT20 CT21 CT18 75thCT19 percentile CT8 =15.2 mGy CT12 th CT30 25 percentile CT3 = 6.9 mGy CT4 CT31 CT28 CT17 CT6 CT7 CT2 CT23 CT10 CT24 CT9 CT26 CT15 CT1 0 5 10 15 20 25 30 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.18: Distribution of CTDIvol values for CT of values for kidney stone indication 183 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.43: Descriptive statistics of representative DLP values for CT of kidney stone indication Total AD DLP (mGy.cm) Min. median DLP Max. median DLP SD Scanners (mGy.cm) (mGy.cm) 24 455.4 231.2 1251.2 296.4 CT29 CT14 CT21 CT18 CT20 75th percentile CT8 =731.1 mGy.cm CT31 CT19 25th percentile CT4 = 312.5 mGy.cm CT3 CT12 CT30 CT6 CT2 CT23 CT7 CT24 CT10 CT28 CT17 CT9 CT15 CT1 CT26 0 200 400 600 800 1000 1200 1400 DLP (mGy.cm) The DLP values are from 1 sequence (Non-contrast examination). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.19: Distribution of DLP values for CT of kidney stone indication Table 4.44: A comparison of indication-based DRL values for CT of kidney stone indication with international values. Country 75th percentile Image quality CTDIvol DLP Mean SNR (mGy) (mGy.cm) Current study (Ghana) 15 731* 6.7 ± 1.4 Public Health England, 2016 (UK) 10 460* - Lajunen (2015 (FI) 7 330* - Wachabauer et al., 2017 (AU) - 329* - *=Single sequence procedure; UK represents United Kingdom, FI represents Finland and AU represents the Australia. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 184 CT Scanners University of Ghana http://ugspace.ug.edu.gh The projected 75th and 25th (CTDIvol and DLP) percentiles for CT of kidney stone have been displayed in red and blue dashed lines, respectively, in Figures 4.18 and 4.19. The median dose values (CTDIvol) varied by a factor of 6, although they were all of acceptable image quality with a mean SNR of 6.7 ± 1.4. The suggested national CTDIvol DRL and the calculated CTDIvol achievable doses for CT of kidney stone are approximately 15 mGy and 10 mGy, respectively. Accordingly, the proposed national DLP DRL is approximately 731 mGy.cm. Additionally, the achievable DLP dose was approximately 455 mGy.cm. Values reported in the literature were generated from a single sequence as done in this study. However, the DRL values were higher than those reported by Public Health England (2016) (CTDIvol = 10 mGy; DLP = 460 mGy.cm), Lajunen (2015) (CTDIvol = 7 mGy; DLP = 330 mGy.cm), and Wachabauer et al. (2017) (DLP = 329 mGy.cm) respectively. The descriptive statistics and projected DRL values for urothelial malignancy (CT- IVU) indication with respect to CTDIvol are presented in Table 4.45 and Figure 4.20 respectively, while those of DLP are shown in Table 4.29 and Figure 4.9. A comparison of developed DRL values (CTDIvol and DLP) for urothelial malignancy (CT-IVU) indications with international values are shown in Table 4.47. 185 University of Ghana http://ugspace.ug.edu.gh Table 4.45: Descriptive statistics of representative CTDIvol values for CT of urothelial malignancy indication Total AD CTDIvol Min. median CTDIvol Max. median CTDIvol SD scanners (mGy) (mGy) (mGy) 19 9.1 4.3 20.4 4.2 CT21 CT18 CT4 CT30 CT17 75th percentile CT28 =11.1 mGy CT6 CT9 th CT31 25 percentile CT12 = 5.8 mGy CT3 CT7 CT24 CT2 CT23 CT15 CT10 CT26 CT1 0 5 10 15 20 25 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.20: Distribution of CTDIvol values for CT of urothelial malignancy (CT-IVU) indication 186 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.46: Descriptive statistics of representative DLP values for CT of urothelial malignancy indication Total AD DLP (mGy.cm) Min. median DLP Max. median DLP SD Scanners (mGy.cm) (mGy.cm) 19 1101.4 607.6 3024.0 641.8 CT21 CT18 CT4 CT6 CT17 CT31 75th percentile CT9 =1448.6 mGy.cm CT30 CT12 25th percentile CT28 = 813.8 mGy.cm CT7 CT3 CT1 CT24 CT15 CT2 CT26 CT10 CT23 0 500 1000 1500 2000 2500 3000 3500 DLP (mGy.cm) The DLP values are derived from between 2-4 sequences (Pre-contrast and 2-3 other post-contrast series). AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.21: Distribution of DLP values for CT of urothelial malignancy indication Table 4.47: A comparison of indication-based DRL values for CT of urothelial malignancy indication/ CT-IVU with international values. Country 75th percentile Image quality CTDIvol DLP Mean SNR (mGy) (mGy.cm) Current study (Ghana) 11 1449*+ 10. 8 ± 2.4 Widmark, (2018), NO 13 1300? - Public Health England, (2016), UK 13 1150? - Lajunen (2015), (FI). - - - Van der Molen et al., (2013), (NL) - 1371? - *+=2-4 sequences (pre-contrast and 2-3 other post-contrast series),? = undeclared number of sequences. NO represents Norway, UK represents United Kingdom, FI represents Finland and NL represents Netherlands. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 187 CT Scanners University of Ghana http://ugspace.ug.edu.gh Nineteen CT facilities undertook CT-IVU scanning mainly for the urothelial malignancy indication. A mean SNR of 10.8 ± 2.4 observed across the images suggested that good images were produced for such examinations in the facilities. The median dose observed across the facilities varied by a factor of 4.7. The estimated 75th and 25th (CTDIvol and DLP) percentiles for the CT-IVU (urothelial malignancy indication) procedure have been displayed in green and blue dashed lines, respectively, in Figures 4.20 and 4.21. Accordingly, the proposed national CTDIvol DRL for urothelial malignancy (CT-IVU) indication was approximately 11 mGy, while the achievable CTDIvol dose was approximately 9 mGy. The projected national DLP DRL for the same indication was approximately 1449 mGy.cm, whereas the achievable DPL dose was approximately 1101 mGy.cm. Comparatively, the DRL (CTDIvol) value CT for urothelial malignancy from this study was lower than the established value of 13 mGy in Norway (Widmark, 2018) and the UK (Public Health England, 2016). Interestingly, the corresponding DRL value in terms of DLP was slightly higher than values from their counterparts (Table 4.47). It is worth noting that facilities in Ghana use 2-4 sequences (pre-contrast and 2-3 other post-contrast series) to produce diagnostic images in CT-IVU examinations for the detection of urothelial malignancy and its related conditions. However, there was a lack of information about the number of sequences used by other countries outlined in Table 4.47. Therefore, conclusions have to be drawn with caution. 188 University of Ghana http://ugspace.ug.edu.gh Table 4.48 and Figure 4.22 present the descriptive statistics and projected DRL values with respect to CTDIvol for PE, respectively. The respective DRL values with respect to DLP are also presented in Table 4.49 and Figure 4.23. A comparison of developed DRL values (CTDIvol and DLP) for PE with international values is shown in Table 4.44. Table 4.48: Descriptive statistics of representative CTDIvol values for CT of pulmonary embolism indication Total AD CTDIvol (mGy) Min. median Max. median CTDIvol SD scanners CTDIvol (mGy) (mGy) 10 9.9 4.2 22.8 6.0 CT21 CT6 CT18 75th percentile CT4 =13.8 mGy CT9 25th percentile CT1 = 6.9 mGy CT7 CT24 CT26 CT23 0 5 10 15 20 25 CTDIvol (mGy) AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.22: Distribution of CTDIvol values for CT of pulmonary embolism indication 189 CT Scanners University of Ghana http://ugspace.ug.edu.gh Table 4.49: Descriptive statistics of representative DLP values for CT of pulmonary embolism indication Total AD DLP Min. median DLP Max. median DLP SD scanners (mGy.cm) (mGy.cm) (mGy.cm) 10 710.6 184.0 1111.7 330.4 CT9 CT6 th CT1 75 percentile = 942 mGy.cm CT4 CT21 25th percentile = 435.2 mGy.cm CT18 CT24 CT7 CT23 CT26 0 200 400 600 800 1000 1200 DLP (mGy.cm) The DLP values are derived from between 1-2 sequences (Pre-contrast phase and actual PE phase). The monitoring phases are not included as it is the situation for all established DRLs. AD: Achievable dose; CTDIvol: volume weighted computed tomography dose index, DLP: dose length product, SD: standard deviation, Min: minimum, Max: maximum. Figure 4.23: Distribution of DLP values for CT of PE indication Table 4.50: A comparison of indication-based DRL values for CT of pulmonary embolism indication with international values. Country 75th percentile Image quality CTDIvol Mean SNR DLP (mGy.cm) (mGy) Current study (Ghana) 14 942+ 12.5 ± 5.1 Public Health England, (2016), UK 13 440* - Foley et al., (2012), IR 13 432* - Kanal et al., (2017), USA 19 557* - Schegerer et al., (2017), DE 15 300* - Van der Molen et al., (2013), NL - 371* - Wachabauer et al., (2017), AU - 400* - *=Single sequence procedure; + =1-2 sequences (pre- and post-contrast). UK represents United Kingdom, IR represents Ireland, USA represent the United States of America, DE represents Germany, NL represents the Netherlands, AU represents Australia. SNR: Signal to noise ratio, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 190 CT Scanners University of Ghana http://ugspace.ug.edu.gh The DRL values for PE are presented in Figures 4.22 and 4.23, and the 75th and 25th percentile (CTDIvol and DLP) values have been displayed in red and blue dashed lines, respectively. The results showed that the proposed national CTDIvol DRL for PE was about 14 mGy. Approximately 10 mGy was found to be the achievable CTDIvol dose value for PE CT examination. In terms of the DLP descriptor, the proposed NDRL for CT of PE was approximately 942 mGy.cm. The achievable DLP for PE was also estimated to be approximately 711 mGy.cm. The image quality in terms of mean SNR value (12.5 ± 5.1) was considered acceptable across the images, although the dose levels varied by a factor 5.4. The projected CTDIvol DRL value was lower than those reported in the USA (19 mGy) and Germany (15 mGy), but higher than those reported in UK (13 mGy) and Ireland (13 mGy). The corresponding DLP was, however, higher than those reported in the UK (440 mGy.cm), Ireland (432 mGy.cm), USA (557 mGy.cm), Germany (300 mGy.cm), Netherlands (371 mGy.cm) and Australia (400 mGy.cm) (Public Health England, 2016; Foley et al., 2012; Kanal et al., 2017; Schegerer et al., 2017; Van der Molen et al., 2013; Wachabauer et al., 2017). The projected indication-based DRL values as outlined are recommended for adoption and implementation by the NRA in collaboration with relevant professional bodies for dose management and accountability of CT imaging in Ghana. The national values may be revised at regular intervals (every 5 years), especially when significant changes in technology and new imaging procedure standards and protocols are available. For an effective implementation of the proposed DRLs, a Microsoft Excel-based tool (BOTB) for inspection and monitoring of DRL compliance purposes was developed in this study. An interphase of the tool is presented in Figure 4.24. 191 University of Ghana http://ugspace.ug.edu.gh Figure 4.24: Interphase of a Microsoft Excel-based developed tool (BOTB) for inspection and monitoring of DRLs compliance purposes. 192 University of Ghana http://ugspace.ug.edu.gh The tool (BOTB) shown in Figure 4.24 has been developed for each indication-based DRL. The interphase enables input of demographic data (such as age, weight, etc.), and dose output/descriptor (CTDIvol, DLP) data. The dose output data entries are calculated automatically for 75th and 25th percentiles. The results are displayed in the ‘results’ section of the interphase. Results displayed in black and red are interpreted as pass (within National indication-based DRL value [NIBDRLV]) and fail (above NIBDRLV), respectively. A blue coloured displayed result indicates that a facility’s doses are below the 25th percentile and hence, image quality tests are required to ensure examinations of acceptable diagnostic image quality are produced. Once the developed indication-based DRLs in this study are approved and duly implemented by the appropriate authority, such as the NRA, this simple tool could be used as part of the NRA’s QMS to encourage DRLs compliance purposes. Crucially, individual facilities could also use this tool for internal audit purposes to monitor their compliance to NIBDRLV and where necessary, corrective actions could be taken when typical patient dose output values are higher than the indication-based DRLs. 193 University of Ghana http://ugspace.ug.edu.gh 4.5 PHASE 5: Dose Impact and Estimation of Cancer Risk Associated with the Doses Effective doses and cancer risks were estimated for purposes of understanding the impact of the dose associated with each of the indications. The results are presented in Tables 4.51-4.56. Table 4.51: Indication-specific effective doses and their equivalent background radiation levels Effective dose, ED, (mSv) Indications Non- Contrast Overall for Background contrast phase examination radiation phase equivalent* Mean± SD Mean± SD Mean± SD CVA/stroke 2.23 ± 0.80 - 2.23 ± 0.80 11 months Head trauma/Injury 2.73 ± 1.20 - 2.73 ± 1.20 13.7 months Brain tumour/ SOL 2.27 ± 0.90 2.26 ± 0.90 4.53 ± 1.70 22.7 months Lung tumour/cancer 5.50 ±1.51 5.10 ± 0.50 10.37 ± 3.10 4.3 years Chest lesion with CKD 5.40 ± 3.00 - 5.40 ± 3.00 27 months Abdomino-pelvic lesion 8.30 ± 5.24 8.87 ± 5.18 17.17 ±10.00 7.2 years Kidney stones 8.51 ± 5.01 - 8.51 ± 5.01 3.5 years Urothelial malignancy (CT-IVU) 6.71± 3.78 14.26 ± 8.74 20.09 ± 12.19 8.4 years Pulmonary angiogram 5.04 ± 4.50 6.56 ± 3.70 11.61± 6.5 4.8 years *Average yearly natural background radiation used ≈ 2.4 mSv (Canadian Nuclear Safety Commission, 2019). CVA: cerebrovascular accident, CKD: chronic kidney disease, SD: standard deviation, ED: effective dose. The estimated mean effective doses (Table 4.51) for single phase (non-contrast) procedures were generally lower than multiple phase (both contrast and non-contrast) procedures, due to the number of examination series. The result was consistent with that previously reported (Smith-Bindman et al., 2009). The highest (20.09 ± 12.19 mSv) and least (2.23 ± 0.80 mSv) mean effective doses were recorded for the CT-IVU examination for diagnosing urothelial malignancy, and CVA/stroke procedure, respectively. The doses associated with these two examinations differ by a factor of 9. The results further showed that the high dose examinations could subject a patient to radiation levels equivalent to 8.4 years of natural background radiation. Radiation dose optimisation is therefore very important in CT imaging. 194 University of Ghana http://ugspace.ug.edu.gh Table 4.52: Various organ doses associated with the CT imaging for the CVA, head trauma/Injury and brain tumour/SOL indications Organ doses (mGy) Organ CVA Head trauma/Injury Brain tumour/ SOL 20 yrs 40 yrs 60 yrs 20 yrs 40 yrs 60 yrs 20 yrs 40 yrs 60 yrs Mean± SD Mean±SD Mean± SD Mean± SD Mean± SD Mean± SD Mean± SD Mean± SD Mean± SD Brain 40.2 ± 17.3 39.2±14.4 39.9±14.2 40.2 ±13.4 42.6± 13.5 43.7 ±14.5 84.5± 27.9 81.9 ± 26.2 82.2 ± 31.1 Pituitary gland 32.9 ± 10.7 32.4±12.9 33.5±11.8 36.6 ±12.4 38.3± 13.1 37.1 ± 12.0 68.1± 24.5 67.7 ± 22.8 67.7 ± 23.7 Lens 46.7 ± 18.0 44.7±18.8 46.5±17.6 47.9 ±15.7 50.7± 16.2 51.2± 16.9 94.0 ±37.3 94.0 ± 33.6 95. 5 ± 34.8 Eyeballs 44.4 ± 17.4 43.0±16.7 44.5±16.0 45.6 ±15.0 48.1±15.4 48.7 ± 16.1 90.9 ±33.3 90.8± 29.4 90.9 ± 33.2 Salivary glands 21.9 ± 17.9 19.9±15.5 17.1±13.3 35.2 ±20.2 37.1± 21.4 32.3 ± 16.7 43.9 ±28.2 40.0± 24.8 42.5 ± 23.6 Oral cavity 14.5 ± 14.9 11.6 ±11.3 10.8±11.2 26.8±16.6 28.8± 18.4 24.8 ± 15.6 28.1± 21.4 25.5 ± 20.7 25.1±17.1 Spinal cord 3.7 ± 4.6 3.2 ± 1.8 2.5±2.7 7.0±6.3 6.9± 6.3 6.2 ± 5.1 8.2 ±9.1 5.4 ± 4.6 5.5 ± 3.9 Thyroid 2.3 ± 3.7 1.2±1.1 1.1±1.3 8.5±9.7 4.2± 6.3 5.8± 2.2 4.8 ±4.0 4.4 ± 2.4 2.4 ± 1.7 CVA: cerebrovascular accident, SOL: space occupying lesion, SD: standard deviation, Yrs: years 195 University of Ghana http://ugspace.ug.edu.gh The results in Table 4.52 indicated that varying amounts of organ doses were delivered to patients of different ages during CT imaging for various head indications. The lenses of the eye in particular, were impacted the most among all the organs in the head region. For the CVA indication, the mean lens doses were 46.7 mGy, 44.7 mGy and 46.5 mGy for patients aged 20, 40 and 60 years old, respectively. The upper range value was lower than the upper limit (41.7- 71.0 mGy) as reported by a study (Yamauchi-Kawara et al., 2010) for similar examination. The higher mean lens doses of 47.9 mGy, 50.7 mGy and 51.2 mGy were recorded for 20, 40- and 60-years old patients in head injury examinations, respectively. Brain tumour/ SOL imparted the highest lens dose of 94.0 mGy, 94.0 mGy and 95.5 mGy in patients of the stated age categories, respectively. The lens doses were relatively high because the gantry angulation needed to project the lenses out of the primary beam during head CT imaging was often not utilised at various facilities. According to a study (Nikupaavo et al., 2015), head CT imaging performed with gantry angulation of 15-20o to project the x-ray beam parallel to the infraorbitomeatal line effectively reduced lens doses. Apart from the lenses, other organs which were greatly impacted by head related CT radiation in a descending order were: eyeballs, brain, pituitary gland, salivary glands, oral cavity, spinal cord and thyroid. 196 University of Ghana http://ugspace.ug.edu.gh Table 4.53: Various organ doses associated with CT imaging for lung tumour/cancer, chest lesion with CKD and PE indications Organ doses (mGy) Organ Lung tumour/cancer Chest lesion with CKD PE 20 yrs 40 yrs 60 yrs 20 yrs 40 yrs 60 yrs 20 yrs 40 yrs 60 yrs Mean± SD Mean± SD Mean± SD Mean±SD Mean± SD Mean± SD Mean± SD Mean± SD Mean± SD Thyroid 29.4 ± 15.6 40.5 ± 24.8 38.1 ± 21.4 15.9 ± 9.8 20.0 ±11.5 20.4 ± 10.7 36.7 ± 32.9 22.9 ± 10.4 26.7 ± 13.9 Oesophagus 17.3 ± 8.7 25.6 ± 15.9 24.1 ± 13.8 9.2 ± 5.7 12.5 ± 7.1 12.9 ± 6.8 24.4 ± 21.9 14.5 ± 6.2 17.0 ± 8.8 Trachea 22.4 ± 12.0 30.81 ± 8.8 29.0 ± 16.3 12.1 ± 7.6 15.2 ± 8.8 15.5 ± 8.2 27.7 ± 24.8 17.5 ± 7.9 20.3 ± 10.4 Thymus 24.9 ± 13.9 33.7 ±20.3 31.9 ± 17.9 13.7 ± 8.7 16.7 ± 9.7 17.1 ± 9.0 29.2 ± 26.2 19.2 ± 8.9 22.1 ± 10.9 Lungs 21.3 ± 11.0 31.2± 19.0 29.6 ± 16.7 11.5 ± 7.2 15.2 ± 8.7 15.8 ± 8.3 28.8 ± 25.8 17.7 ± 7.7 20.5 ± 10.2 Breast 17.8 ± 11.4 29.2 ± 18.0 27.5 ± 15.5 9.7 ± 7.3 14.0 ± 8.5 14.7 ± 7.7 27.0 ± 24.2 16.5 ± 7.3 19.3 ± 10.0 Heart wall 21.7 ± 12.3 32.8 ± 20.0 31.1 ± 17.6 11.8 ± 7.7 15.9 ± 9.3 16.7 ± 8.8 30.3 ± 27.1 18.7 ± 8.1 21.4 ± 10.4 CKD: chronic kidney disease, PE: pulmonary embolism, SD: standard deviation, Yrs: years. 197 University of Ghana http://ugspace.ug.edu.gh Table 4.53 indicated that the highest organ dose during CT imaging for lung tumour/cancer, chest lesion with CKD and PE was received by the thyroid gland, which is sensitive to the carcinogenic effect of ionising radiation. The mean ranges of thyroid dose for the various indications were 20.4-40.5 mGy (lung tumour/cancer), 15.9-40.4 mGy (chest lesion with CKD), and 22.9-36.7 mGy (PE). Relatively lower mean organ doses in the ranges of 17.8- 29.2 mGy and 21.3-31.2 mGy were received by the breast and lungs in lung tumour/cancer examinations, respectively. A previous study (Saltybaeva et al., 2016) however, reported a lower lung dose of 8.6 mGy for 60-year-old patients undergoing lung tumour/cancer CT imaging. The organ doses for the breast and lungs ranged from 16.5-27.0 mGy and 17.7-28.8 mGy, and from 9.7-14.7 mGy and 11.5-15.8 mGy, for PE, and chest lesion with CKD indicated examinations, respectively. A recent study (Haspi et al., 2019) in Malaysia reported lower breast organ dose range of 10.9-23.8 mGy and lung organ dose range of 10.6- 23.4 mGy during PE examination. Other ograns such as oesophagus, trachea, thymus and heart wall also received mean organ doses in a range of 11.8-32.8 mGy across the indications in the chest region. 198 University of Ghana http://ugspace.ug.edu.gh Table 4.54: Various organ doses associated with CT imaging for AP lesion, kidney stones and urothelial malignancy indications Organ doses (mGy) Organ AP lesion Kidney stones Urothelial malignancy 20 yrs 40 yrs 60 yrs 20 yrs 40 yrs 60 yrs 20 yrs 40 yrs 60 yrs Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Stomach wall 39.3 ± 25.8 39.0 ± 22.8 38.3 ± 22.8 17.2 ±11.1 17.2 ± 10.4 16.3 ± 9.9 38.0 ± 24.1 49.5 ± 30.2 51.7 ± 30.2 Liver 37.9 ± 25.1 37.7 ± 22.2 37.2 ± 22.2 16.7 ±10.8 16.7 ± 10.2 15.6 ± 9.5 36.6 ± 23.3 49.1 ± 29.5 49.6 ± 28.9 Gall bladder 35.6 ± 22.8 35.5 ± 20.2 35.5 ± 21.1 15.6 ± 9.9 15.7 ± 9.0 15.1 ± 9.4 34.9 ± 21.9 46.4 ± 27.6 47.7 ± 28.3 Adrenals 32.7 ± 20.9 32.7 ± 18.5 32.7 ± 19.4 14.3 ± 9.0 14.4 ± 8.2 13.9 ± 8.7 32.1 ± 20.1 42.8 ± 25.6 44.0 ± 26.1 Spleen 36.6 ± 24.0 36.4 ± 21.3 36.1 ± 21.5 16.1 ±10.4 16.1 ± 9.7 15.2 ± 9.3 35.5 ± 22.5 47.3 ± 28.3 48.2 ± 28.2 Pancreas 38.0 ± 25.4 37.8 ±22.5 37.2 ± 22.3 16.7±10.9 16.7 ± 10.3 15.6 ± 9.5 36.7 ± 23.4 49.0 ± 29.4 49.5 ± 28.8 Kidney 39.9 ± 25.7 39.8 ±22.8 39.7 ± 23.6 17.5±11.1 17.6 ± 10.2 16.8 ± 10.4 39.0 ± 24.5 51.7 ± 30.7 53.3 ± 31.5 Small intestine 39.9 ± 25.1 40.2 ±22.3 40.4 ± 24.1 17.6±11.0 17.8 ± 9.9 17.3 ± 10.9 39.7 ± 24.8 52.0 ± 30.4 54.9 ± 33.0 Colon 41.7 ± 26.4 41.8 ±23.4 41.9 ± 24.9 18.3±11.5 18.5 ± 10.4 17.9 ± 11.2 41.1 ± 25.7 54.4 ± 32.1 56.7 ± 33.8 Rectosigmoid 29.8 ± 19.0 32.5 ±17.7 32.3 ± 20.0 13.9 ± 9.1 14.3 ± 7.8 13.6 ± 8.7 32.2 ± 20.2 39.7 ± 24.7 45.5 ± 28.4 U. Bladder 34.0 ± 21.9 37.4 ±21.2 36.2 ± 22.1 16.2±10.6 16.3 ± 9.5 15.0 ± 9.2 36.2 ± 23.0 43.4 ± 26.3 50.9 ± 30.9 Prostate 8.1 ± 10.9 11.0 ±14.2 12.1 ± 16.2 4.2 ± 6.1 4.6 ± 5.6 5.5 ± 7.0 11.3 ± 15.8 14.1 ± 21.3 22.6 ± 27.6 Uterus 13.9 ± 19.2 15.3 ±19.4 12.0 ± 17.1 7.0 ± 9.1 6.6 ± 9.2 3.8 ± 5.7 12.7 ± 18.6 16.2 ± 18.7 13.5 ± 19.4 Testes 1.6 ± 3.1 2.1 ± 3.8 3.8 ± 8.0 0.6 ± 0.9 1.0 ± 2.3 2.0 ± 5.7 1.6 ± 2.4 1.9 ± 3.4 5.0 ± 7.0 Ovary 14.9 ± 20.5 16.3 ±20.6 12.8 ± 18.2 7.5 ± 9.6 7.0 ± 9.8 4.1 ± 6.1 13.5 ± 19.8 16.1 ± 19.6 14.4 ± 20.7 Skin 11.2 ± 7.2 11.8 ± 6.5 12.0 ± 7.2 5.1 ± 3.4 5.3 ± 3.0 5.0 ± 3.0 11.6 ± 7.1 14.1 ± 8.5 16.5 ± 9.7 Muscle 12.7 ± 8.3 13.6 ± 7.5 13.8 ± 8.5 5.9 ± 4.0 6.1 ± 3.5 5.6 ± 3.5 13.4 ± 8.2 16.6 ± 9.3 19.1 ± 11.3 *Act. marrow 16.9 ± 11.0 17.5 ±10.1 17.2 ± 10.3 7.5 ± 5.0 7.6 ± 4.6 7.1 ± 4.4 17.0 ± 10.8 21.6 ± 12.5 23.4 ± 13.9 +S. marrow 14.5 ± 9.6 13.8 ± 8.1 14.2 ± 8.6 5.9 ± 3.1 6.4 ± 4.1 5.8 ± 3.5 14.1 ± 9.2 18.3 ± 10.7 19.3 ±11.3 *Act. = Active marrow; +S = Shallow; AP= abdominopelvic, SD = standard deviation, Yrs = years. 199 University of Ghana http://ugspace.ug.edu.gh Table 4.54 explains that many organ doses were estimated for CT examinations involving AP lesion, kidney stones and urothelial malignancy. These organs included very radiosensitive ones such as colon, uterus, testes, ovary, active marrow and shallow marrow, among others. The highest organ dose for CT imaging was received in the colon associated with AP lesion (41.7-41.9 mGy), kidney stones (17.9 - 18.5 mGy) and urothelial malignancy (41.1 - 56.7 mGy) across all ages (20, 40 and 60 years old). A study (Pai et al., 2018) also reported colon organ doses of 9.33-24.2 mGy in kidney stones indicated CT imaging, with an upper limit 30.8% higher than the observed value in this study. 200 University of Ghana http://ugspace.ug.edu.gh Table 4.55: Cancer and mortality risks associated with CT doses for lung tumour/cancer, lung lesion with CKD and PE indications. Lung tumour/cancer Chest lesion with CKD PE Cancer Ages Organ dose LARi/ LARm/ Organ LARi/ LARm/ Organ dose LARi/ LARm/ site (mGy) (yrs) 100,000 100,000 dose (mGy) 100,000 100,000 (mGy) 100,000 100,000 20 21.3 ± 11.0 52.7 48.6 11.5 ± 7.2 28.5 26.2 28.8 ± 25.8 71.3 65.7 Lung 40 31.2 ± 19.0 53.7 49.8 15.2 ± 8.7 26.1 24.2 17.7 ± 7.7 30.4 28.2 60 29.6 ± 16.7 42.9 40.8 15.8 ± 8.3 22.9 21.8 20.5 ± 10.2 29.7 28.3 20 29.4 ± 15.6 19.7 - 15.9 ± 9.8 10.7 - 36.7 ± 32.9 24.6 - Thyroid 40 40.5 ± 24.8 10.1 - 20.0 ± 11.5 5.0 - 22.9 ± 10.4 5.7 - 60 38.1 ± 21.4 20.6 - 20.4 ± 10.7 0.1 - 26.7 ± 13.9 0.2 - 20 17.8 ± 11.4 76.4 18.0 9.7 ± 7.3 41.6 10.0 27.0 ± 24.2 115.8 27.3 Breast 40 29.2 ± 18.0 41.2 10.2 14.0 ± 8.5 19.7 4.9 16.5 ± 7.3 23.3 5.8 60 27.5 ± 15.5 8.5 2.5 14.7 ± 7.7 5.3 1.3 19.3 ± 10.0 6.0 1.8 LARi: lifetime attributable risk of cancer incidence, LARm: Lifetime attributable risk of cancer mortality, CKD: chronic kidney disease, PE: pulmonary embolism, Yrs: years. 201 University of Ghana http://ugspace.ug.edu.gh The risks of radiation-induced lung, thyroid and breast cancer incidence and mortality from CT imaging of the investigated indications were relatively low as upper LAR values were within 1 in 10,000 to 1 in 1,000 of the population (Table 4.55). However, the risk of PE radiation-induced breast cancer was moderate (1 in 1,000 to 1 in 500) (Varghese et al, 2019). In particular, the LAR of lung cancer incidence for patients aged 20, 40 and 60 years old due to CT scans for lung tumour/cancer detection were 52.7, 53.7 and 42.9 in 100,000 patients, respectively. Their corresponding risk of death was 48.6, 49.8 and 40.8 in 100,000 patients, respectively. For patients who received radiation for chest lesion with CKD indication, the populations in 100,000 patients likely to develop lung cancer were 28.5 (among 20 years old), 26.1 (among 40 years old) and 22.9 (among 60 years old); and possibly deaths were 26.2, 24.2 and 21.8, respectively. The risk of developing PE examination-induced lung cancers and resulting in death were more likely in the 20 years group compared with their counterparts who received radiation from other indications in the chest region. Numerically, 71.3 (among 20- years old), 30.4 (among 40 years old) and 29.7 (among 60 years old) were likely to develop lung cancer, from which 65.7, 28.2 and 28.3 were likely to die, respectively. The number of patients at risk of developing thyroid cancer from CT of lung tumour/cancer, chest lesion with CKD and PE ranged from 10.1-20.6, 0.1-10.7 and 0.2-24.6 in 100,000 patients, respectively. Radiation induced-breast cancer from CT of lung tumour/cancer was also likely in 76.4 (among 20 year old), 41.2 (among 40 years old) and 8.5 (among 60 years old) patients among a population of 100,000, while, 18.0, 10.2 and 2.5 were likely to die, respectively. Radiation subjected to patients for chest lesion with CKD and PE examinations were likely to induce breast cancer in 5.3-41.6 (1 in 18,868 to 1 in 2,404) and 6.0-115.8 (1 in 16,667 to 1 in 857) patients per 100,000 procedures, respectively. The corresponding likely radiation-induced cancer deaths were estimated at 1 in 76,923 to 1 in 10,000 patients for chest lesion with CKD and 1 in 55,556 to 1 in 3,663 patients for PE examinations. 202 University of Ghana http://ugspace.ug.edu.gh Table 4.56: Cancer and mortality risks associated with CT doses for AP lesion, kidney stones and urothelial malignancy AP lesion Kidney stone Urothelial malignancy/CT-IVU Cancer Age Organ LARi/ LARm Organ LARi/ LARm Organ LARi/ LARm site (yrs) dose (mGy) 100,000 /100,000 dose (mGy) 100,000 /100,000 dose (mGy) 100,000 /100,000 20 39.3 ± 25.8 18.1 9.8 17.2 ± 11.1 7.9 4.3 38.0 ±24.1 17.5 9.5 Stomach 40 39.0 ± 22.8 10.5 6.6 17.2 ± 10.4 4.6 3.0 49.5 ±30.2 13.4 8.7 60 38.3 ± 22.8 7.7 3.0 16.3 ± 9.9 3.3 2.2 51.7 ±30.2 10.3 7.0 20 37.9 ± 25.1 8.3 6.6 16.7 ±10.8 3.7 2.9 36.6 ±23.3 8.1 6.4 Liver 40 37.7 ± 22.2 7.9 4.5 16.7 ± 10.2 3.5 2.0 49.1 ±29.5 10.3 5.9 60 37.2 ± 22.2 5.2 3.5 15.6 ± 9.5 2.2 1.5 49.6 ±28.9 6.9 4.7 20 41.7 ± 26.4 59.8 28.6 18.3 ±11.5 26.3 12.5 41.1 ±25.7 59.0 28.2 Colon 40 41.8 ± 23.4 51.0 20.3 18.5 ± 10.4 22.6 9.0 54.4 ±32.1 66.4 26.4 60 41.9 ± 24.9 39.4 16.8 17.9 ± 11.2 16.8 7.2 56.7 ±33.8 53.3 22.7 20 8.1 ± 10.9 3.8 0.7 4.2 ± 6.1 2.0 0.4 11.3 ±15.8 5.4 1.0 Prostate 40 11.0 ± 14.2 3.9 0.7 4.6 ± 5.6 1.6 0.3 14.1 ±21.3 5.0 0.8 60 12.1 ± 16.2 3.2 0.9 5.5 ± 7.0 1.4 0.4 22.6 ±27.6 6.0 1.6 20 16.9 ± 11.0 14.1 10.0 7.5 ± 5.0 14.6 10.3 17.0 ±10.8 14.4 10.1 Leukaemia 40 17.5 ± 10.1 12.8 10.4 7.6 ± 4.6 12.8 10.5 21.6 ±12.5 12.5 10.2 60 17.2 ± 10.3 12.0 11.0 7.1 ± 4.4 15.0 13.8 23.4 ±13.9 16.3 15.0 20 34.0 ± 21.9 37.1 18.4 16.2 ± 10.6 17.7 4.4 36.2 ± 23.0 39.5 9.8 Urinary 40 37.4 ± 21.2 29.5 7.5 16.3 ± 9.5 12.9 3.3 43.4 ± 26.3 34.3 8.7 Bladder 60 36.2 ± 22.1 23.5 7.1 15.0 ± 9.2 9.8 2.9 50.9 ± 30.9 33.1 9.9 20 13.9 ± 19.2 3.6 0.8 7.0 ± 9.1 1.8 0.42 12.7 ± 18.6 3.3 0.76 Uterus 40 15.3 ± 19.4 2.4 0.6 6.6 ± 9.2 1.1 0.31 16.2 ± 18.7 2.6 0.65 60 12.0 ± 17.1 1.9 0.4 3.8 ± 5.7 0.3 0.11 13.5 ± 19.4 1.2 0.41 20 14.9 ± 20.5 7.5 4.2 7.5 ± 9.6 3.8 2.1 13.5 ± 19.8 6.8 3.8 Ovary 40 16.3 ± 20.6 5.1 3.3 7.0 ± 9.8 2.2 1.4 16.1 ± 19.6 5.0 3.2 60 12.8 ± 18.2 2.3 1.9 4.1 ± 6.1 0.7 0.6 14.4 ± 20.7 2.6 2.2 LARi: lifetime attributable risk of cancer incidence, LARm: Lifetime attributable risk of cancer mortality, AP: abdomino-pelvic, CT-IVU: computed tomography intravenous urography, Yrs: years. 203 University of Ghana http://ugspace.ug.edu.gh Among the organs subjected to CT exposures for various indications in the abdominal and pelvic regions, radiation-induced colon cancer risks were highest and likely among 39.4 - 59.8 patients in 100,000 AP lesion procedures. The risk was even higher in CT-IVU examinations, in a range of 53.3- 66.4 patients in 100,000 procedures but were less (16.8-26.3 patients) in kidney stone procedures. Accordingly, the risk of radiation-induced colon mortality was common in CT-IVU than AP lesion and kidney stone procedures. Although the organ dose likely to cause radiation-induced leukaemia was lower than that received by the urinary bladder across all indications, the LAR of leukaemia mortality was higher than those observed in the urinary bladder. This suggests a high likelihood of leukaemia mortality. The ovaries were likely to develop AP lesion radiation-induced cancer in 7.5/100,000 (1 in 13,000), 5.1/100000 (1 in 19,608), and 2.3/100,000 (1 in 43,478) people at age 20, 40 and 60 years, respectively. Kidney stone examinations were also likely to induce ovarian cancer in 3.8, 2.2 and 0.7 patients in a pool of 100,000 procedures at ages of 20, 40 and 60 years, respectively. More so, about 1 in 38, 462 to 1 in 14,706 patients were also likely to develop ovarian cancer due to CT-IVU examinations in Ghana. Since the ovaries contain reproductive information, it is indicative that hereditary effects were also possible in patients who received high ovarian doses. Although all the upper LAR values for all the indications were within 1 in 10,000 to 1 in 1,000 of the population and suggested a low radiation risk, there was a need for further optimisation to reduce the dose levels and risks, as noted by Varghese et al., (2019). This is because small individual risks applied to a large population could lead to a public health issue some years in the future (IAEA, 2018; Do, 2016). 204 University of Ghana http://ugspace.ug.edu.gh 4.6 PHASE 6: Optimisation Methods As observed in the descriptive comparison, the developed NDRLs were relatively higher than some European values. Optimisation interventions were investigated to manage the observed dose levels to further reduce the radiation risks. 4.6.1 Optimisation method 1: Dose reduction through optimisation of scan length The tolerable additional scan length (along the z-axis) provided above upper target (DAUT) and below lower target (DBLT) were first investigated and the results are presented in Table 4.57. Table 4.57: Descriptive statistics of average extra scan length allowed above and below the target anatomic regions/areas. Indication Average extra scan length (mm) Total DAUT DBLT DAUT+ DBLT Routine CVA 8.1 ± 9.4 20.1 ± 13.7 28.2± 21.3 Head trauma/Injury 9.0 ± 12.9 40.1 ± 33.5 49.1 ± 25.4 Brain tumour/ SOL 7.1 ± 8.0 20.7 ± 14.8 27.8 ± 10.7 Lung tumour/cancer 24.1 ± 16.1 34.0 ± 17.0 58.1 ± 16.6 Chest lesion with CKD 22.0 ± 14.1 30.9 ± 15.2 52.9 ± 14.9 Abdominopelvic lesion 25.1 ± 16.4 23.6 ± 17.3 46.7 ± 17.1 Kidney stones 24.5 ± 17.8 20.1 ± 16.4 44.6 ± 17.3 Urothelial malignancy 27.7 ± 18.6 21.9 ± 16.6 49.6 ± 16.9 Pulmonary angiogram 30.0 ± 12.3 33.3 ± 19.0 63.3 ± 14.5 DAUT= Distance above upper target; DBLT= Distance below lower target; CVA= cerebrovascular accident; CKD =chronic kidney disease. Table 4.57 shows that varying DAUT and DBLT values were used across the facilities for each indication. These have the potential to cause detrimental effects to patients. For CVA, the average DAUT and DBLT were 8.1± 9 mm and 20.1±13.7 mm, respectively, making a total of 28.2 ± 21.3 mm extra scan coverage along the z-axis and considered as unnecessary radiation 205 University of Ghana http://ugspace.ug.edu.gh coverage. The estimated total average extra scan length ranged from 27.8 to 63.3 mm across all the indications. The largest extra total scan coverage of 63.3 ± 14.5 mm was recorded for PE examinations, while brain tumour/SOL (27.8 ± 10.7 mm) followed by routine CVA (28.2 ± 21.3 mm) were scanned with smaller extra total scan coverage. Moreover, the DBLT values recorded the highest extra scan coverage along the z-axis which suggested that more extra scan coverage was allowed at the distal part of the organ compared to the proximal part. The difference between the smallest and the largest extra total scan coverage was 2.3 folds. The results indicated the need to explore optimal scan coverage for indication-based CT procedures. The results of a CT phantom-based study that explored optimal scan coverage for routine CVA imaging are presented in Table 4.58. Evaluated extra scan length (z-axis), mean scores for subjective image quality analysis, and level of agreement between raters are also presented. 206 University of Ghana http://ugspace.ug.edu.gh 4.6.1.1 Phantom-Based Optimisation study Table 4.58: Extra scan length (z-axis) evaluated, mean scores for subjective image quality analysis, and level of agreement between raters in head region examination; routine CVA Protocol/ Average extra scan length CTDIvol DLP SNR % DLP Subjective image quality scores image ID DAUT (mm) DBLT (mm) (mGy) (mGy.cm) reduction over Mean SD ICC p-value current practice value A 8.1 ± 9.4 20.1 ±13.7 21.28 385.58 5.3 - 3.92 0.74 B 0 0 21.28 323.87 5.3 16.0 3.92 0.67 C 10* 0 21.28 307.17 5.4 22.0 3.92 0.74 0.78 < 0.001 D 10 10 21.28 366.43 5.3 5.0 3.89 0.67 E 0 5 21.28 335.58 5.1 13.0 3.92 0.67 F 5 5 21.28 345.15 5.2 10.5 3.90 0.67 Key: * = below the vertex, A= Average extra scan used across the CT facilities (approximately 8 mm above the vertex and 20 mm below the base of skull; B= No extra scan length above the vertex and below base of skull; C= 10 mm below the skull vertex and 0 mm below the base of skull at foramen magnum; D=10 mm above vertex and 10 mm below the base of skull; E= No extra scan lengths above the vertex and 5 mm extra scan lengths below the base of skull; F=5 mm extra scan length above the vertex and below the base of skull; DAUT= Distance above upper target; DBLT= Distance below lower target; ICC = Intraclass Correlation Coefficients; CTDIvol = volume weighted computed tomography dose index; DLP= dose length product, SD =Standard deviation. 207 University of Ghana http://ugspace.ug.edu.gh Among the scan coverage protocols (A-D) investigated, Table 4.58 indicates that image sets A, B, C and E had the highest subjective image quality with mean score of 3.92 each out of a maximum of 5. This indicated that they presented the best image quality among all evaluated protocol-based images. Image D also had a mean lower score of 3.89. Intraclass correlation coefficient tests were run using SPSS version 23.0 (SPSS Inc., Chicago, IL USA) to determine statistical agreement between the two raters about the image quality scores. Studies (Koo and Li, 2016; McHugh, 2012), have indicated that an ICC value of 0.78, p < 0.001 suggests a good agreement between the two raters. Hence, protocol C (a brain scan covering 10 mm below the vertex and 0 mm below the base of skull at foramen magnum) was considered a very suitable intervention for dose reduction in CVA CT imaging, since it produced a very good objective (SNR = 5.3) and subjective image qualities similar to the original protocol, and also generated the best dose output (DLP) reduction of 22.0% (Table 4.58). 4.6.1.2 Patient-Based Optimisation Study A patient-based optimisation study was undertaken to evaluate the application of phantom results in clinical practice. The demographic characteristics of the patients used in this particular evaluation are presented in Table 4.59. 208 University of Ghana http://ugspace.ug.edu.gh Table 4.59: Demographics and clinical history of the patients Variables Categories n % Mean ± SD (yrs) Range (yrs) Age - - 57.92 ±15.9 28 – 92 Gender Male 54 54 - - Female 46 46 - - Weight - - - 70.8 ± 10.3 50-90 Clinical CVA- Suspicion of ischemic 41 41 based History infarctions Suspicion of recurrent bleed 15 15 Transient ischemic attack 7 7 Hemiparesis 5 5 Recurrent CVA 14 14 Haemorrhagic stroke 10 10 Hypertensive, mild stroke 8 8 n= frequency, % = percentage, SD=standard deviation The ages of the sample size of 100 patients used in the patient-based interventional study ranged from 28 to 92 years with a mean age of 57.92 ± 15.9 years. The weight range was 50 - 90 kg with a mean of 70.8 ± 10.3 kg. There were more males (54.0%) than females (46.0%). Ischemic infarction was presented as the clinical history of majority (41%) of them while 5% presented with signs of hemiparesis, which are both conditions relating to poor blood supply to the brain. Ischemic infarction is common because it accounts for about 68% of all strokes globally, while haemorrhagic types are 32% prevalent (Chugh, 2019). Table 4.60 shows the results of the measured SNR and subjective image quality scores for full and reduced CVA CT procedures. 209 University of Ghana http://ugspace.ug.edu.gh Table 4.60: Measured SNR and subjective image quality scores for full range and reduced range CVA CT procedures Protocol/ image ID SNR Subjective image quality scores Mean ± SD, Range Mean ± SD ICC value p-value Full range CT 8.4 ± 1.8 (5.1-15.3) 4.6 ± 0.56 0.9 0.001 Reduced range CT 8.4 ± 1.8 (5.1-15.3) 4.6 ± 0.54 SNR: Signal to noise ratio, SD: standard deviation, ICC: interclass correlation coefficient; CT: computed tomography. A subjective analysis of the resultant images using radiologists indicated a similar (4.6 out of 5) subjective image quality scores for both full and reduced range CT scanners. The agreement between the two raters were statistically similar (ICC value: 0.9; p = 0.001). Similarly, a mean SNR value of 8.4 was observed across the image set which was 68% above the needed value to distinguish image features at 100% certainty (Bushberg et al., 2011). These results suggested that the reduced protocol was capable of achieving similar diagnostic quality as the full range CT for purposes of identifying CVA pathologies which constitute the second leading cause of death and a major cause of disability worldwide (Katan and Luft, 2018). The dose output characteristics in terms of DRL, effective dose (ED) and organ dose for both full range and reduced range CVA CT procedures are presented in Table 4.61. Comparative descriptive statistics of dose impact of full range CVA CT and reduced range CVA CT procedures are presented in Table 4.61. Figure 4.27 provides the pictorial presentation of the organ doses for the reduced and full range CVA CT scans. 210 University of Ghana http://ugspace.ug.edu.gh Table 4.61: Comparison of dose impact of full range and reduced range CVA CT procedures Parameter Full range Reduced DR p-value t-value Mean 95% Confidence CT range CT (%) difference interval of difference DLP (mGy.cm) 926.7 ± 280 715.1 ± 213 22.8 <0.001 21.21 211.65 (191.850, 231.450) Effective dose (ED)(mSv) 2.1 ± 0.6 1.6 ± 0.5 23.8 <0.001 21.21 0.487 (0.441, 0.532) Brain 39.3 ±13.8 35.0 ±11.5 10.9 <0.001 13.34 4.285 (3.648, 4.923) Pituitary gland 33.3 ±10.3 23.2 ± 7.9 30.2 <0.001 10.76 10.04 (8.193, 11.896) Lens 46.1 ±15.4 31.3 ± 15.6 32.0 <0.001 6.98 14.76 (10.56, 18.950) Organ dose Eyeballs 43.9 ±14.7 30.2 ± 13.4 31.1 <0.001 7.44 13.64 (10.000, 17.280) (mGy) Salivary glands 21.3 ± 9.3 5.3 ± 4.8 75.0 <0.001 17.68 15.982 (14.189, 17.776) Oral cavity 14.4 ± 7.7 3.5 ± 2.9 75.6 <0.001 16.09 10.907 (9.562, 12.253) Spinal cord 2.9 ± 1.6 0.8 ± 0.6 70.7 <0.001 15.64 2.016 (1.760, 2.272) Thyroid 1.1 ± 0.7 0.5 ± 0.3 57.2 <0.001 14.58 0.637 (0.550, 0.723) DR: dose reduction, DLP: dose length product; CT: computed tomography. 211 University of Ghana http://ugspace.ug.edu.gh 50.0 45.0 Full range CT 40.0 Reduced range CT 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 Brain Pituitary Lens Eye balls Salivary Oral Spinal Thyroid gland glands cavity cord Ograns Figure 4.25: Pictorial presentation of the organ doses for reduced and full range CVA CT scans. Table 4.61 and Figure 4.25 indicated that when the reduced range intervention was used in patient-based stroke/CVA examinations across the CT facilities in Ghana, the DLP and ED values reduced by 22.8% and 23.8%, respectively. In terms of organ doses, the brain, pituitary gland, eye lens and the eyeballs doses reduced by 10.9%, 30.2%, 32.0% and 31.1%, respectively. Moreover, the doses to the salivary glands, oral cavity, spinal cord and thyroid reduced by 75.0%, 75.6%, 70.7% and 57.2%, respectively. All the dose reductions were found to be statistically significant. A recent study (Zinsser et al., 2019) which also explored reduced scan length in patients with suspected acute appendicitis demonstrated that a judicious use of scan range could reduce testicular, breasts and red marrow doses by 81%, 97.4% and 29.6%, respectively. Therefore, an implementation of the developed reduced scan range protocol in CT facilities could reduce radiation exposure levels in routine CVA examinations. This could potentially reduce the population organ dose, since CVA is the most common indication in CT imaging in Ghana. 212 Organ dose (mGy) University of Ghana http://ugspace.ug.edu.gh The effect of the proposed reduced scan protocol on diagnosing CVA-related and other conditions through radiologists’ reports are presented in Table 4.62. Table 4.62: Diagnoses based on radiologists’ reports Diagnosis Full Reduced Sensitivity of range range reduced range detection detection n N (%) Normal CT brain 34 34 100 CVA-related diagnosis Thalamus infarcts 3 3 100 Small vessel ischemic disease 23 23 100 Posterior limb of the internal capsule infarcts 11 11 100 Bilateral cerebral infarcts 14 14 100 Bilateral subacute cerebral and pontine infarcts 5 5 100 Lacunar infarct 13 13 100 Chronic occipital lobe deep white matter infarct 4 4 100 Midbrain infarcts 3 4 100 Frontal lobe infarct 12 12 100 Parieto-temporal infract 6 6 100 Subarachnoid haemorrhage 2 2 100 Intracerebral haemorrhage 9 9 100 Subdural haemorrhage 9 9 100 Incidental findings Maxillary sinus disease 12 6 50 Ethmoid sinus disease 4 1 25 Sphenoid sinus disease 3 0 0 Frontal sinus disease 2 2 100 Pneumatized mastoid air cells 1 1 100 Brain atrophy 51 51 100 Note: There were multiple disease conditions in some patients, CVA: cerebrovascular accident 213 University of Ghana http://ugspace.ug.edu.gh During the subjective image analysis, radiologists reported on the diagnosable conditions on both the reduced and full range images. Table 4.62 shows that the reduced range images had the same sensitivity (100%) as the full range regarding the identification of normal brain and CVA- related pathologies. This is expected as the CVA-related pathologies are found mainly within the brain region (Mayo Clinic, 2020). However, for incidental findings, the sensitivity was 100%, 100%, 50%, 25% and 0% for detection of brain atrophy, frontal sinus disease, maxillary sinus disease, ethmoid sinus disease and sphenoid sinus diseases, respectively. The results further clarify that the proposed reduced range could provide 100% results on normal brain and CVA-related diagnoses. However, it could not be utilised to provide accurate diagnostic information on some conditions (such as sphenoid, maxillary, and ethmoid sinus diseases) outside the brain tissue. Therefore, this reduced range protocol is a recommended procedure targeted for only CVA-related diagnosis. 4.6.2 Regression Modelled Equations From the patient-based study data, equations (4.1-4.8) were produced for easy estimation of organ doses from CVA examinations using CTDIvol, slice thickness (ST) and scan length (SL) parameters. These are supportive tools for optimisation of radiation dose levels. All variables used in modelling the equations passed normality tests [The significant levels, p-values, of Shapiro- Wilk and Grubbs' tests for outlies were all > 0.05). The optimum multicollinearity tests (VIF ranged from 1.767 – 2.065] (Appendices X-XII). Mathematical algorithms developed for estimating brain dose (BrainD), salivary gland dose (S-glandD), eye lens dose (E-lensD) and eyeballs dose (E-ballsD) have been expressed in equations 4.1, 4.2, 4.3 and 4.4, respectively. Other equations for oral cavity dose (O-cavityD), 214 University of Ghana http://ugspace.ug.edu.gh spinal cord dose (S-cordD), thyroid dose (ThyroidD) and pituitary gland dose (P-glandD) have been also been expressed in equations 4.5 to 4.8, respectively. BrainD (mGy) = -5.19 + 0.7218 CTDIvol + 0.285 ST + 0.0241 SL (4.1) (SD=2.12, R-Sq=97.7%, p-value = <0.001). S-glandD (mGy) = -104.5 + 0.3874 CTDIvol - 0.358 ST + 0.6158 SL (4.2) (SD=6.22, R-Sq= 56.09%, p-value =<0.001) E-lensD (mGy) = -12.35 + 0.8392 CTDIvol - 0.148 ST + 0.0764 SL (4.3) (SD=2.50, R-Sq=97.44%, p-value =<0.001). E-ballsD (mGy) = -10.79 + 0.7970 CTDIvol - 0.102 ST + 0.0668 SL (4.4) (SD=2.24, R-Sq=97.74%, p-value =<0.001). O-cavityD (mGy) = -98.91 + 0.2738 CTDIvol - 0.117 ST + 0.5747 SL (4.5) (SD=4.49, R-Sq=66.66%, p-value =<0.001). S-cordD (mGy) = -18.15 + 0.05423 CTDIvol + 0.0935 ST + 0.1031 SL (4.6) (SD=1.13, R-Sq=50.81%, p-value =<0.001). ThyroidD (mGy) = -5.04 + 0.02214 CTDIvol + 0.0836 ST + 0.02715 SL (4.7) (SD=0.54, R-Sq=32.8%, p-value =<0.001). P-glandD (mGy) = -19.69 + 0.5890 CTDIvol - 0.778 ST + 0.1363 SL (4.8) (SD=3.05, R-Sq=91.51%, p-value =<0.001). 215 University of Ghana http://ugspace.ug.edu.gh From equations 4.1 to 4.8, the respective R-squares (R-Sq) indicated the influence of the independent variables (CTDIvol, ST and SL) in the model on the respective organ doses. Per equations 4.1, 4.3 and 4.4, the models suggested that over 97% of the respective organ doses were predicted by CTDIvol, ST and SL. In the case of equations 4.2, 4.5, 4.6, 4.7 and 4.8, the variation in respective doses were accounted for by 56.1%, 66.7%, 50.8%, 32.8% and 91.5% of the respective independent variables. Models with low R-square values, particularly, equation 4.7 must be used with caution because the model could not entirely predict the respective organ doses. 4.6.3 Optimisation Method 2: The role of AEC Utilisation In the Phase 5 study, it was observed that some of the CT scanners were operated without activating their AEC systems. The impact of AEC systems was investigated in those facilities. A pictorial diagram with four plates (A-D) showing tube loading variations across different z- positions for different scanners operating with and without the activation of AEC systems is presented in Figure 4.26. 216 University of Ghana http://ugspace.ug.edu.gh Figure 4.26: AEC modulated tube current and fixed tube current profiles across different z-positions As illustrated in plate A-D, the blue and red lines show the variation of the tube loading with and without activation of AEC systems, respectively, as the phantom was scanned from the dome of the liver to the pubic symphysis. The above shows that with the use of AEC (of various models), the tube loading varied with the density of the anatomical region along the z-axis. However, constant tube loading was used along the entire z-axis positions for fixed-mAs procedures. This phenomenon was same for all the types of AECs used in the study, hence, for 217 University of Ghana http://ugspace.ug.edu.gh examinations without AEC activation, both smaller and bigger abdominal regions along the z-axis would receive the same dose without any optimisation. The dose output results of the study undertaken to compare AEC and fixed tube current systems (in facilities operating without AEC systems) for optimisation actions have been presented in the Table 4.63. Table 4.63: Comparison of dose output for abdomino-pelvic procedures undertaken with and without AEC systems. Scanner AEC setting Mean CTDIvol DLP DLP Mean noise Mean models mAs (mGy) (mGy.cm) DR (%) (SD) SNR Siemens AEC: off 240 16.1 731.0 - 7.2 7.3 Somatom CareDose4D 166 6.9 312.0 57.3 11.1 12.1 Emotion GE AEC: off 200 12.5 572.1 - 5.6 6.4 Lightspeed AutomA 3D 105 6.6 306.8 46.4 8.6 9.7 Pro 16 Toshiba AEC: off 187 22.3 1027.6 - 6.1 5.3 Aquilion SureExposure 54.45 9.3 428.2 58.3 9.1 7.9 16 TSX 3D Philips AEC: off 355 14.1 637.0 - 5.3 5.6 Brilliance DoseRight Z- 278.9 5.9 284.0 55.4 7.8 12.2 64 DOM, ACS Cumulative comparative results AEC 7.2 ± 0.9 332.8 ± 64.8 55.2, 6.5 ± 0.9 Without AEC 16.3 ± 1.4 741.9 ± 29.6 p=0.01 10.5 ± 2.1 DR: dose reduction: AEC: automatic exposure control: CTDIvol: computed tomography volume weighted dose index, DLP: dose length product, SD: standard deviation. ACS: automatic current selection. 218 University of Ghana http://ugspace.ug.edu.gh Table 4.63 shows the cumulative mean dosimetric results of AP CT imaging with AEC and fixed mAs. The cumulative mean CTDIvol values were 7.2 ± 0.9 mGy and 16.3 ± 1.4 mGy, respectively. The corresponding mean DLP values were 332.8 ± 64.8 mGy.cm and 741.9 ± 296.0 mGy.cm, respectively. The results indicated that the non-AEC scanning CT facilities could reduce CT radiation dose output (DLP) by a factor of 2.2 (55.2%) while ensuring acceptable image quality if the AEC systems were activated and utilised. The DLP dose reduction found across the four facilities varied from 46.4-58.3% which was closely similar to values (36.8 -73.8%) reported by Sulemana et al. (2020) in a single centre. Moreover, the dose reduction benefits observed with respective AEC systems were even higher compared with values (about 20–40%) reported by Higaki et al. (2019) and Merzan et al. (2017). It suggested lack of training, policy on CT infrastructure and proper organisational programmes driving CT purchasing, installation, acceptance testing, commissioning and operations are some of the drawbacks in the utilisation of AEC systems. Due to some of these challenges, CT machines are sometimes purchased and commissioned for work without any training offered by application specialists, thereby, creating a knowledge gap on the best ways to utilise equipment configurations such as the AEC. Therefore, training of radiographers on technical application of AEC systems, particularly, before equipment commissioning is absolutely crucial for dose management in Ghana. Moreover, strict adherence to policies regarding AEC usage is necessary in order to ensure that dose optimisation benefits are derived from the available CT scanners with AEC activated systems. As part of the study, all the facilities operating fixed-mAs settings have been assisted to activate their AEC systems. 219 University of Ghana http://ugspace.ug.edu.gh CHAPTER FIVE SUMMARY, CONCLUSION AND RECOMMENDATIONS 5.0 Overview The summary, conclusion, challenges/limitations and recommendations from this study are presented in this chapter. 5.1 Summary and Conclusion The main objective of this study was to develop national indication-based DRL values for common and prioritised indications of the adult human body for clinical application in Ghana, and assess the risk of undertaking each indication-based CT examination, and also propose steps for dose optimisation. Studies on CT infrastructure and common indications as well as assessment of QMS were undertaken. The study found that there were 28 functional CT scanners out of the 31 that participated in the study. Most of the scanners (58.1%) were installed in the Greater Accra Region, and all those included in the DRL study (n=25) passed the QC tests. However, the needed infrastructure for QMS was unavailable in some of the CT facilities. Particularly, more than half (51.6%) of the facilities lacked a documented protocol for CT scanning for some indications. Radiologists were, therefore, used mainly to define the basic diagnostic requirements of each indication, prior to collecting the needed dose descriptors for the DRL study. The observed common indications and their projected DRL values in terms of CTDIvol (mGy) and DLP (mGy.cm) were CVA/stroke (77 mGy; 1313 mGy.cm), head trauma/injury (76 mGy; 1596 mGy.cm), brain tumour/SOL (77 mGy; 2696 mGy.cm), lung tumour/cancer (12 mGy; 828 mGy.cm) and chest lesion with chronic kidney disease (13 mGy; 467 mGy.cm). Others were AP lesion (17 mGy; 1299 mGy.cm), kidney stones (15 mGy; 731 220 University of Ghana http://ugspace.ug.edu.gh mGy.cm), urothelial malignancy/CT-IVU (11 mGy; 1449 mGy.cm) and PE (14 mGy; 942 mGy.cm). These values could be used as a QA/QC tool and a trigger to detect CT imaging centres utilising remarkably high doses (outliers) in a particular indication-based radiological task, for which optimisation measures are required. It is also anticipated that these benchmarks would guide stakeholders to further unearth better ways of optimising radiation doses in CT imaging. For an effective implementation of the proposed DRL values, a Microsoft Excel-based tool (BOTB) for inspection and monitoring of DRLs’ compliance was developed in this study. The tool had a capability for internal audit purposes to monitor individual facility’s compliance to the national indication-based DRL values and where necessary, corrective actions could be taken to promote dose management. The estimated risk of PE radiation-induced breast cancer ranged from 6-115.8 people in 100,000 procedures. Moreover, the risk of CT-IVU radiation-induced colon cancer ranged from 53.3-66.4 people in 100,000 patients. About 1 in 38, 462 to 1 in 14,706 patients were also likely to develop ovarian cancer due to CT-IVU examinations in Ghana. The risk of leukaemia mortality as a result of radiation accruing from CT-IVU for urothelial malignancy, also ranged from 1 in 1,901 to 1 in 6,667 patients. Although the LARi and LARm were considered minimal to moderate levels across the various indications, there was a need for further optimisation measure in CT imaging to reduce the dose levels and radiation risks. The study found a novel protocol that could be used to scan stroke/CVA related conditions with optimal image quality, while reducing facilities’ radiation dose levels. In particular, it found that using a scan range covering 10 mm below the vertex and 0 mm below the base of skull at foramen magnum could reduce the mean effective dose of the facilities by 23.8%, and organ doses 221 University of Ghana http://ugspace.ug.edu.gh by 32% (eye lens), 70.7% (spinal cord), 57.2% (thyroid), 75.6% (oral cavity), 10.9% (brain), 31.1% (eyeballs) and 30.2% (pituitary gland) with negligible effect on the image quality. However, this protocol was not 100% sensitive in diagnosing incidental findings that were not within the brain region. Therefore, it was strictly recommended for stroke/CVA-related CT imaging. Eight organ dose equations were also developed to aid in dose management. It is believed that CT imaging facilities could utilise these equations to estimate organ doses with ease, and guide in appropriate decisions. Finally, the study concluded that if AEC systems were used in facilities operating with fixed tube loading systems, radiation doses could also be reduced by a range of 46.4-58.3% without any significant compromise on image quality. Therefore, there was a need to ensure proper utilisation of AEC systems in all the CT facilities in Ghana. 5.2 Challenges/Limitations Many challenges were encountered during the study. Some of them include: i. Though the study attempted to include all the CT facilities in the study, four of them did not to participate. These few numbers did not affect the generalisation of the findings. However, their inclusion would have provided full information on CT scanners and dose levels for their inclusion in the DRL values and implementation of optimisation measures where necessary. ii. Long equipment downtimes prevented the inclusion of broken-down CT scanners in the study, although several follow-ups were made to inquire about their status and possible inclusion in the study. 222 University of Ghana http://ugspace.ug.edu.gh iii. In Ghana, patient dose information recording, and automatic dose tracking systems are not mandatory, and are not readily available in most facilities. In the absence of these, several days were spent in the facilities collecting data. iv. The study was solely focused on adults. Paediatrics were therefore not considered. 5.3 Recommendations 5.3.1 Nuclear Regulatory Authority i. The developed indication-based DRL values are recommended for adoption and implementation by the NRA, in collaboration with the relevant professional bodies for dose management and accountability of CT imaging in Ghana. The national values may be revised at regular intervals (every 5 years), especially when significant changes in technology, and new imaging procedure standards and protocols are available. ii. The developed monitoring tool (BOTB) in this study could be used by the NRA for inspection and monitoring of DRLs compliance purposes, if the proposed DRL values are implemented. iii. In view of the observed lack of QMS in the various CT facilities, the NRA should liaise with the Ministry of Health and all appropriate stakeholders to ensure that all policies needed for CT QMS in Ghana are implemented, and adhere to accordingly. 223 University of Ghana http://ugspace.ug.edu.gh 5.3.2 Managers and Health Professionals in CT Imaging i. Managers and professionals in CT imaging such as radiologists, radiographers and medical physicists may use the proposed indication-based values as a country-based guide to regulate and optimise their CT practices in Ghana. ii. A scan range covering from 10 mm (1 cm) below the vertex and 0 mm below the base of skull at foramen magnum is recommended in CT imaging of stroke/CVA specific conditions or indications in order to significantly optimise radiation dose, while ensuring good image quality. 5.3.3 Future Research i. Future research focusing on the development of paediatric indication-based diagnostic reference levels and optimisation methods for computed tomography examinations in Ghana is recommended. ii. Future research focusing on the development of adult indication-based diagnostic reference levels for indications not considered in the study (non-common indications), and optimisation methods for computed tomography examinations in Ghana is also recommended. 224 University of Ghana http://ugspace.ug.edu.gh REFERENCES 1. AAPM, (2007). The Measurement, Reporting and Management of Radiation Dose in CT: Report of the AAPM Task Group 23. Report No. 96, American Association of Physicists in Medicine, New York. 2. AAPM, (2008). AAPM Report No. 96. The Measurement, Reporting, and Management of Radiation Dose in CT Report of AAPM Task Group 23: CT Dosimetry. New York: American Association of Physicists in Medicine. 3. AAPM, (2014). "Use of water equivalent diameter for calculating patient size and size- specific dose estimates (SSDE) in CT: The report of AAPM Task Group 220. Los Angeles: American Association of Physicists in Medicine. 4. Abdulkadir, K.M, Schandorf, C. and Hasford, F. (2016). Determination of Computed Tomography Diagnostic Reference Levels in North-Central Nigeria. The Pacific Journal of Science and Technology, 17(2), 341-348. 5. Addo, P. (2016). Computed tomography (CT) radiation dose in children: A survey to propose regional diagnostic reference levels in greater Accra Ghana. MPhil Thesis: University of Ghana. http//ugspace.ug.edu.gh. 6. Adejoh, T., Nwogu B.N, Anene, N.C, Onwujekwe, C.E, Imo, S.A, Okolo, C.J, Chiegwu, U. H, Nzotta, C.C. (2017). Computed Tomography Scanner Distribution and Downtimes in Southeast Nigeria. Journal of The Association of Radiographers of Nigeria, 31(1), 8-15 7. Akyea-Larbi, K.O., Schandorf, C., Hasford, F., Inkoom, S., Sackey, T. A. and Acquah, G. F. (2016). Assessment of cancer incidence and mortality risks associated with effective dose of computed tomography examinations. Journal of Applied Science and Technology, 20(1 and 2),17 – 22. 225 University of Ghana http://ugspace.ug.edu.gh 8. Alessio, A.M., Farrel, M.B., Fahey, F.H. (2015). Role of reference levels in nuclear medicine: a report of the SNMMI dose optimisation task force. Journal of Nuclear Medicine, 5(12), 1960-1964. 9. Ali, M.U. (2015). Role of CTDI to estimate patient dose from a single slice to multi slice CT scanners. International Journal of Multidisciplinary Research and Development, 2 (10), 262-266. 10. Alkhorayef, M., Babikir, E., Alrushoud, A., Al-Mohammed, H, Sulieman, A. (2017) Patient radiation biological risk in computed tomography angiography procedure. Saudi Journal of Biological Sciences, 24, (1), 235–240. 11. Amaoui, B., Semghouli, S., Kharras, A.E., Fahssi, E.M. (2020). Medical exposure and risk estimation during routine radio-diagnostic examinations in Agadir City in 2012. Journal of Radiation Research and Applied Sciences, 13(1), 69-72. 12. American Association of Physicists in Medicine (AAPM), (1992). Recommendations on performance characteristics of diagnostic exposure meters. AAPM REPORT NO. 35. Medical Physics. 19(1), 231-241. 13. American College of Radiology, (2010). Image filters improve image quality and lower patient radiation dose associated with CT scans. ScienceDaily. Retrieved from www.sciencedaily.com/releases/2010/05/100503161221.htm 14. American College of Radiology, (2014). ACR–AAPM practice parameter for diagnostic reference levels and achievable doses in medical x-ray imaging. Retrieved from www.acr.org/~/media/796DE35AA407447DB81CEB5612B4553D.pdf. Accessed on: 25th August, 2017. 226 University of Ghana http://ugspace.ug.edu.gh 15. American College of Radiology, (2018). ACR-AAPM practice parameter for diagnostic reference levels and achievable doses in medical x-ray imaging. Retrieved from http://www.acr.org/~/media/ACR/Documents/PGTS/guidelines/Reference_Levels_Diagn ostic_Xray.pdf. Published 2013. Revised 2018. Accessed 4/11/ 2019. 16. American Society for Quality (2019). what is a quality management system (QMS)? -- ISO 9001 and other quality management system. Retrieved from https://asq.org/quality- resources/iso-9000. Accessed on:9/4/2019. 17. Anim-Sampong, S., Antwi, A.K., Botwe, O.B., Boateng, S.R. (2016). Comparison of 640- slice Aquilon ONE CT scanner’s measured dosimetric parameters with ICRP dose reference levels for head, chest and abdominal CT examinations. Safety in Health, (2016) 2(7),1-8. DOI 10.1186/s40886-016-0019-4. 18. Ataç, G.K, Parmaksız, A., İnal, T., Bulur, E., Bulgurlu, F., Öncü, T., Gündoğdu, S. (2015). Patient doses from CT examinations in Turkey. Diagnostic and Interventional Radiology. 2015 Sep; 21(5), 428–434. 19. Australian Government, (2016), National Diagnostic Reference Level Fact Sheet... ARPANSA. Retrieved from https://www.arpansa.gov.au/research/surveys/national- diagnostic-reference-level-service/frequently-asked-questions Accessed on: 25th August, 2017. 20. Australian Radiation Protection and Nuclear Safety Agency, [ARPANSA] (2018). Current Australian national diagnostic reference levels for multi detector computed tomography. Retrieved from https://www.arpansa.gov.au/research-and- expertise/surveys/national-diagnostic-reference-level-service/current-australian-drls- update/mdct. Accessed: 7/11/2019. 227 University of Ghana http://ugspace.ug.edu.gh 21. Aweda, M.A, Arogundade, R.A. (2007). Patient dose reduction methods in computerized tomography procedures: A review. International Journal of Physical Sciences, 2 (1), 001- 009. 22. Bauhs JA, Vrieze TJ, Primak AN, Bruesewitz MR and McCollough CH. (2008). CT Dosimetry: Comparison of Measurement Techniques and Devices. RadioGraphics. 28(1), 245-253. 23. Bellolio, M.F., Heien, H.C., Sangaralingham, L.R., Jeffery, M.M., Campbell, R.L., Cabrera, D., … Hess, E.P. (2017). Increased Computed Tomography Utilisation in the Emergency Department and Its Association with Hospital Admission. Western Journal of Emergency Medicine, 18(5), 835–845. http://doi.org/10.5811/westjem.2017.5.34152. 24. de González, A.B., Mahesh, M., Kim, K.P, Bhargavan, M., Lewis, R., Mettler, F., Land, C. (2009). Projected cancer risks from computed tomographic scans performed in the United States in 2007. Archives of internal medicine, 169 (22), 2071-2077. doi: 10.1001/archinternmed.2009.440. 25. Boone J., Strauss K., Cody D., McCollough C., McNitt-Gray M., Toth T., … Frush D. (2011). “Size-specific dose estimates (SSDE) in paediatric and adult body CT examinations,” Report of AAPM Task Group 204, 2011. 26. Boone, J.M., Brink, J.A., Edyvean, S., Huda, W., Leitz, W., McCollough, C.H., and McNitt-Gray, M.F. (2012). 4. Overview of Existing CT-Dosimetry Methods. Journal of the ICRU, 12(1), 35–45. doi:10.1093/jicru/nds004. 27. de Basea, M., Salotti, J.A., Pearce, M. S., Muchart, J., Riera, L., Barber, I., … Cardis, E. (2016). Trends and patterns in the use of computed tomography in children and young 228 University of Ghana http://ugspace.ug.edu.gh adults in Catalonia - results from the EPI-CT study. Pediatric Radiology, 46(1), 119–129. doi:10.1007/s00247-015-3434-5. 28. Bowling, A., (2014). Research Methods in Health: Investigating Health and Health Services (UK Higher Education OUP Humanities and Social Sciences Health). 4th eds. London: Open University Press. 29. Brady, Z., Ramanauskas, F., Cain, T.M., and Johnston, P.N. (2012). Assessment of paediatric CT dose indicators for the purpose of optimisation. The British Journal of Radiology, 85(1019), 1488–1498. doi:10.1259/bjr/28015185. 30. Brat, H., Zanca, F., Montandon, S., Racine, D., Rizk, B., Meicher, E., and Fournier, D. (2019). Local clinical diagnostic reference levels for chest and abdomen CT examinations in adults as a function of body mass index and clinical indication: a prospective multicenter study. European Radiology, 29 (12), 6794-6804 doi:10.1007/s00330-019-06257- 31. Brenner, D.J., Hall, E.J. (2007). Computed tomography — an increasing source of radiation exposure. The New England Journal of Medicine 357(1),2277–2284. 32. Brink, J.A., Miller, D.L. (2015). U. S national diagnostic reference levels: Closing the gap. Radiology, 277(1), 3–6. 33. Bujila, R., Kull, L., Danielsson, M., and Andersson, J. (2018). Applying three different methods of measuring CTDIfree air to the extended CTDI formalism for wide-beam scanners (IEC 60601-2-44): A comparative study. Journal of Applied Clinical Medical Physics, 19(4), 281–289. doi:10.1002/acm2.12363. 34. Bushberg, J.T., Seibert, J. A, Leidholt, M, E., Boone, J.M. (2011). The essential physics of medical imaging. 2nd edition, London: Lippincott Williams and Wilkins; 2011. 229 University of Ghana http://ugspace.ug.edu.gh 35. Butler, P. F., Kanal, K. M. (2018). Diagnostic reference levels for adult patients in the United States. Journal of the American College of Radiology, 15(6), 932–933. doi: 10.1016/j.jacr.2017.12.012. 36. Canadian Nuclear Safety Commission (2019) Radiation doses. Retrieved from http://nuclearsafety.gc.ca/eng/resources/radiation/introduction-to-radiation/radiation- doses.cfm#targetText=A%20typical%20annual%20dose%20received,energy%20workers %20is%2050%20mSv. Accessed on: 30/10/19. 37. Cannata, V., Genovese., E., Longo, M. (2017). Radiation risk. In MC Garganese and GFL D’Errico (Eds.), Conventional Nuclear Medicine in Pediatrics (pp.11-18). Cham: Springer International Publishing. 38. Mitchell, O. (2015). Experimental Research Design. Philadelphia: John Wiley & Sons, Inc. https://doi.org/10.1002/9781118519639.wbecpx113. 39. Castellano, I.A, Nicol, E.D, Bull, R.K, Roobottom, C.A, Williams, M.C, Harden, S.P. (2017). A prospective national survey of coronary CT angiography radiation doses in the United Kingdom. Journal of Cardiovasc Comput Tomography.11(4), 268–73. 40. Charan, J., and Biswas, T. (2013). How to calculate sample size for different study designs in medical research? Indian Journal of Psychological Medicine, 35(2),121–126. doi:10.4103/0253-7176.116232. 41. Cheung, L.C., Katki, H.A., Chaturvedi, A.K., Jemal, A., Berg, C.D. (2018). Preventing lung cancer mortality by computed tomography screening: The Effect of Risk-Based Versus U.S. Preventive Services Task Force Eligibility Criteria, 2005–2015. Annals of Internal Medicine. 168(1),229–232. doi: 10.7326/M17-2067. 230 University of Ghana http://ugspace.ug.edu.gh 42. Choudhurg, A. (2017). Questionnaire Method of Data Collection: Advantages and Disadvantages. Retrieved from http://www.yourarticlelibrary.com/social-research/data- collection/questionnaire-method-of-data-collection-advantages-and-disadvantages/64512. Accessed on:17/11/17. 43. Chugh, C. (2019). Acute Ischemic Stroke: Management Approach. Indian Journal of Critical Care Medicine, 23(Suppl 2), S140–S146. 44. Council of the European Union, (2014) Council directive 2013/59/ Euratom of 5 December 2013 laying down basic safety standards for protection against the dangers arising from exposure to ionising radiation, and repealing Directives 89/618/Euratom, 90/641/ Euratom, 96/29/Euratom, 97/43/Euratom and 2003/122/Euratom. In: Official Journal of the European Union, 2014; 13. 45. Damilakis, J., Frija, G., Hierath, M., Jaschke, W., Mayerhofer-Sebera, U., Paulo, G., Repussard, J. … Verius, M. (2018). European Study on Clinical Diagnostic Reference Levels for x-ray Medical Imaging Most studies conducted in CT DRLs have been Anatomic DRLs. Retrieved from http://www.eurosafeimaging.org/wp/wp- content/uploads/2017/09/D2.1_Report-and-review-on-existing-clinical DRLs_final_published-on-website.pdf 46. Danish Health Authority, (2015). CT Reference doser. Copenhagen; Danish Health Authority. 47. Dauer, L.T, Hricak, H. (2014). Addressing the challenge of managing radiation use in medical imaging: Paradigm shifts and strategic priorities. Oncology-Ny. 28(3),243-244. 231 University of Ghana http://ugspace.ug.edu.gh 48. Delis, H., Christaki, K., Healy, B., Loreti, G., Poli, G.L., Toroi, P., Meghzifene, A. (2017). Moving beyond quality control in diagnostic radiology and the role of the clinically qualified medical physicist. Physica Medica, 41 (1), 104-108. 49. DenOtter, T.D, Schubert, J. (2019). Hounsfield Unit. In: StatPeals (internet). Treasure Island (FL): StatPeals Publishing. 50. Desouky, O., Ding, N., Zhou, G. (2015). Targeted and non-targeted effects of Ionising radiation. Journal of Radiation Research and Applied Sciences, 8(2), 247-254. 51. Disher, B., Lenarduzzi, L., Lewis, B., and Teeuwen, J. (2006). The Physics of Computed Tomography. Retrieved from https://web2.uwindsor.ca/courses/physics/high_schools/2006/Medical_Imaging/ctphysics .html. A ccesed on: 29/11/2019. 52. Dixon, M.T., Loader, R.J., Stevens, G.C., and Rowles, N.P. (2016). An evaluation of organ dose modulation on a GE optima CT660-computed tomography scanner. Journal of Applied Clinical Medical Physics, 17(3), 380–391. doi:10.1120/jacmp.v17i3.5724. 53. Dixon, R.L., Ballard, A. C. (2007). Experimental validation of a versatile system of CT dosimetry using a conventional ion chamber: Beyond CTDI100. Medical Physics, 34(8), 3399-3413. 54. Do, K. (2016). General principles of radiation protection in fields of diagnostic Medical Exposure. Journal of Korean Medical Science, 2016; 31(1), S6-9. 55. Duncan, J.R.,, Panahipour, S. (2014). Tissue attenuation of X-rays. Retrieved from http//www.ImageWisely.org 56. Edmonds, K.D. (2009). Diagnostic reference levels as a quality assurance tool. The Radiographer, 56 (3), 32–37. 232 University of Ghana http://ugspace.ug.edu.gh 57. Ekpo, E.U. Adejoh, T. Erim, E.A. (2019). Dose benchmarks for paediatric head computed tomography examination in Nigeria. Radiation Protection Dosimetry. 185(4), 464-471. https://doi.org/10.1093/rpd/ncz036 58. Ekpo, E.U., Adejoh, T., Akwo, D.J., Emeka, C.O., Modu, A.A., Abba, M… Chiegwu, H.U. (2018). Diagnostic reference levels for common computed tomography (CT) examinations: results from the first Nigerian nationwide dose survey. Journal of Radiological Protection. 38 (2018), 525-535. 59. eSurvey, (2017). Features Overview. Retrieved from https://esurv.org/docs/?Features_Overview. Accessed on 2/12/17. 60. European Association of Nuclear Medicine, (EANM), European Federation of Organisations for Medical Physics (EFOMP), European Federation of Radiographer Societies (EFRS), European Society of Radiology (ESR) and European Society for Radiotherapy and Oncology (ESTRO). (2017). Common strategic research agenda for radiation protection in medicine. Insights Imaging; 8(1),183–197. 61. European Commission, (1999). European guidelines on quality criteria for computed tomography. EUR 16262 EN. Office for Official Publications of the European Communities, Luxembourg. 62. European Commission, (2014). Radiation Protection N° 180. Diagnostic Reference Levels in Thirty-six European Countries. Luxembourg: Publications Office of the European Union, 2014. 63. European Commission, (2015). European Guidelines on DRLs for Paediatric Imaging Retrieved from 233 University of Ghana http://ugspace.ug.edu.gh file:///C:/Users/user/Downloads/Livelli_di_dose_di_riferimento_in_Radiologia_Pediatric a_(Draft_2015).pdf. Accessed on: 25th Sep, 2017. 64. European Society of Radiology, (2014). Renewal of radiological equipment. Insights Imaging. 5(5),543-546. 65. European Society of Radiology, (2017). Diagnostic Reference Levels. Retrieved from http://www.eurosafeimaging.org/webinars/diagnostic-reference-levels. Accessed on: 19/02/18. 66. European Society of Radiology, (2018). EuroSafe Imaging Call for Action 2018. Retrieved from http://www.eurosafeimaging.org/about/call-for-action. Accessed on: 14 Nov, 2018 67. European Union, (2014). Medical Radiation Exposure of the European Population (Part 1/2), and Diagnostic Reference Levels in Thirty-six European Countries. Luxembourg: European Union. 68. Ferreira, T., Rasband, W. (2012). ImageJ User Guide-IJ1.46r. Retrieved from http://imagej.nih.gov/ij/docs/guide. Accessed on: 16 august, 2018. 69. Fertikh, D. (2015). Head Computed Tomography Scanning. Available at:https://emedicine.medscape.com/article/2110836-overview. Accessed on: 7/01/2020. 70. Fillon, M., Si-Mohamed, S., Coulon, P., Vuillod, A., Klahr, P., Boussel, L. 92018). Reduction of patient radiation dose with a new organ based dose modulation technique for thoraco abdominopelvic computed tomography (CT) (Liver dose right index). Diagnostic and Interventional Imaging. 99(7-8),483-492. 71. Foley, S.J., McEntee, M.F., Rainford, L.A. (2012). Establishment of CT diagnostic reference levels in Ireland. British Journal of Radiology. 85(1018),1390–1397. 234 University of Ghana http://ugspace.ug.edu.gh 72. Fukuda, A., Lin, P.J., Matsubara, K., Miyati, T. (2014). Measurement of table feed speed in modern CT. Journal of Applied Clinical Medical Physics, 15(3),275–281. doi:10.1120/jacmp.v15i3.4703. 73. Fukushima, Y., Tsushima, Y., Takei, H., Taketomi-Takahashi, A., Otake, H., Endo, K. (2012). Diagnostic reference level of computed tomography (CT) in Japan. Radiation Protection Dosimetry. 151(1),51–57. 74. Furugori, S., Kato, M., Abe, T., Iwashita, M., Morimura, N. (2018). Treating patients in a trauma room equipped with computed tomography and patients' mortality: a non-controlled comparison study. World journal of emergency surgery, 13, (16), 1-11. doi:10.1186/s13017-018-0176-3. 75. Gao, Y., Quinn, B., Mahmood, U., Long, D., Erdi, Y., St. Germain, J., ……..Dauer, L.T. (2017). A comparison of pediatric and adult CT organ dose estimation methods. BMC Medical Imaging 17(28),2-17. DOI 10.1186/s12880-017-0199-3 76. Garcia, E.M., Camacho, M.A., Karolyi, D.R., Kim, D. H., Cash, B.D., Chang, K. J., … Carucci, L.R. (2018). ACR Appropriateness Criteria® Right Lower Quadrant Pain- Suspected Appendicitis. Journal of the American College of Radiology, 15(11), S373– S387. doi: 10.1016/j.jacr.2018.09.033 77. García-Mónaco, R.D., Andisco, D.E., Blanco, S.A. A., Ulla, M. (2019). Clinical diagnostic computed tomography reference levels (DRLs) for adults and children at Hospital Italiano de Buenos Aires. EuroSafe Imaging 2019 Poster No: ESI- 0077http://dx.doi.org/10.26044/esi2019/ESI-0077. 235 University of Ghana http://ugspace.ug.edu.gh 78. Gedel, A.M., Gablah, P.G. (2014). Management of radiation dose to pediatric patients undergoing CT examination at Korle-Bu Teaching Hospital, Accra – Ghana. Advances in Physics Theories and Applications, 37(1),30-38. 79. German Federal Office for Radiation Protection (2016). Bundesamt für Strahlenschutz. Diagnostische Referenzwerte für diagnostische und interventionelle Röntgenanwendungen. Bundesanzeiger. AT 15.07.2016: B8. 80. Ghana Statistical Services (2019). Population and Housing Census projection 2019. Available at: http://www.statsghana.gov.gh/. Accessed on: 7/3/2019. 81. Goodman, T.R. (2018). Ionising Radiation Effects and Their Risk to Humans. Retrieved from https://www.imagewisely.org/Imaging-Modalities/Computed- Tomography/Imaging-Physicians/Articles/Ionising-Radiation-Effects-and-Their-Risk-to- Humans. Accessed on 13/1/2018. 82. Gould, M.K., Donington, J., Lynch, W.R., Mazzone, P.J., Midthun, D. E., Naidich, D. P., and Wiener, R.S. (2013). Evaluation of individuals with pulmonary nodules: When is It lung cancer? Chest, 143(5), e93S–e120S. doi:10.1378/chest.12-2351 83. Greffier, J., Pereira, F., Macri, F., Beregi, J-P., Larbi, A. (2016). CT dose reduction using Automatic Exposure Control and iterative reconstruction: A chest paediatric phantoms study. Physica Medica. 32(4),582-589. 84. Guberina, N., Lechel, U., Forsting, M., Ringelstein, A. (2016). Efficacy of high-pitch CT protocols for radiation dose reduction. Journal of Radiological Protection. 36(4),N57-N66. 85. Halid B., Karim M.K.A., Sabarudin A., Bakar K.A., Shariff N.D. (2018) Assessment of Lifetime Attributable Risk of Stomach and Colon Cancer During Abdominal CT Examinations Based on Monte Carlo Simulation. In: Vo Van T., Nguyen Le T., Nguyen 236 University of Ghana http://ugspace.ug.edu.gh Duc T. (eds) 6th International Conference on the Development of Biomedical Engineering in Vietnam (BME6). BME 2017. IFMBE Proceedings, vol 63. Springer, Singapore 86. Hall, E.J, Brenner, D.J. (2008). Cancer Risks from Diagnostic Radiology. Br J Radiol. 81(965),362–78. 87. Hamed, S.A., Mekkawy, M.A., and Abozaid, H. (2015). Differential diagnosis of a vanishing brain space occupying lesion in a child. World Journal of Clinical Cases, 3(11),956–964. doi:10.12998/wjcc.v3.i11.956. 88. Hart, D., Wall, B.F. (2004). “U.K. Population dose from medical X-ray examinations,” European Journal of Radiology, 50(3),285-291. 89. Hart, D., Wall, BF, Hillier, M.C., Shrimpton, P.C. (2008). Frequency and collective dose for medical and dental X-ray examinations in the UK, 2008. Oxfordshire: National Radiological Protection Board; 2010. 90. Hasford, F., Wyk, B.V., Mabhengu,T., Vangu, M.D.T., Kyere, A.K., Amuasi J.H. (2015). Determination of dose delivery accuracy in CT examinations. Journal of Radiation Research and Applied Sciences, 8 (2015),489-492. 91. Haspi, H.H., Abdul, K.M.K., Abbas, Z., Muniandy, S.C., Sabarudin, A., Hoong, N.K. (2019). Patient-specific radiation dose and cancer risk estimate in computed tomography pulmonary angiography examinations based on lung effective diameter. Preprints.org. DOI: 10.20944/preprints201907.0269.v1. 92. Hausleiter, J., Meyer, T., Hermann, F., Hadamitzky, M., Krebs, M., Gerber, T.C., McCollough, C., … Achenbach, S. (2009). Estimated radiation dose associated with cardiac CT angiography. The Journal of the American Medical Association, 301(5),500- 507. 237 University of Ghana http://ugspace.ug.edu.gh 93. Heale, R., Twycross, A. (2015). Validity and reliability in quantitative studies. Evidence Based Nursing, 18(3),66-67. 10.1136/eb-2015-102129. 94. Healthmanagement, (2018). Clinical diagnostic reference levels in medical imaging. Retrieved from https://healthmanagement.org/c/imaging/issuearticle/clinical-diagnostic- reference-levels-in-medical-imaging. Accessed on: 24/10/2019. 95. Higaki, T., Nakamura, Y., Fukumoto, W., Honda, Y., Tatsugami, F., Awai, K. (2019). Clinical application of radiation dose reduction at abdominal CT. European Journal of Radiology, 111(1),68–75. 96. Huda, W. (2011). Radiation dosimetry in CT: the role of the manufacturer. Imaging in Medicine, 3(2),247–259. 97. IAEA GSR Part 3. (2014). Radiation protection and Safety of radiation sources: International basic Safety standards. Vienna: IAEA. 98. IAEA, (2012). IAEA human health series No. 19: Quality Assurance Programme for Computed Tomography: Diagnostic and Therapy Applications. Vienna: IAEA. Retrieved from https://www-pub.iaea.org/MTCD/Publications/PDF/Pub1557_web.pdf 99. IAEA, (2013). Diagnostic Reference Levels (DRLs) in Medical Imaging. Retrieved from https://rpop.iaea.org/RPOP/RPoP/Content/InformationFor/HealthProfessionals/1_Radiolo gy/Optimisation/diagnostic-reference-levels.htm. Accessed on: 20/10/2017. 100. AEA, (2018). IAEA Safety Standards for protecting people and the environment: Radiation Protection and Safety in Medical Uses of Ionising Radiation. Vienna: International Atomic Energy Agency. IAEA Safety standards Series No. SSG-46. 238 University of Ghana http://ugspace.ug.edu.gh 101. IAEA, (2019). Referring medical practitioners. Retrieved from https://www.iaea.org/resources/rpop/health-professionals/other-specialities-and-imaging- modalities/referring-medical-practitioners. Accessed on: 20/10/19. 102. ICRP, (1991). 1990 Recommendations of the International Commission on Radiological Protection. ICRP Publication 60. Ann. ICRP, 21 (1-3). 103. ICRP, (1996). Radiological Protection and Safety in Medicine (Report 73). Ann ICRP, 26(2), 1-31. 104. ICRP, (2000). Managing patient dose in computed tomography. A report of the International Commission on Radiological Protection. Task Group on Control of Radiation Dose in Computed Tomography. Ann ICRP, 30(4),7-45. 105. ICRP, (2001). Diagnostic reference levels in medical imaging: review and additional advice. Ann ICRP, 31:33-52. 106. ICRP, (2007). The 2007 recommendations of the international commission on radiological protection. ICRP publication 103. Ann ICRP, 37(2-4),1-332. 107. ICRP, (2017). Diagnostic reference levels in medical imaging. ICRP Publication 135. Ann. ICRP, 46(1). https://journals.sagepub.com/doi/pdf/10.1177/0146645317717209 108. ImageQC, (2017). Image analysis for quality control. Retrieved from https://www.imageqc.com/. Accessed on: 3/01/18. 109. Imaginis, (2017). Brief History of CT. Retrieved from http://www.imaginis.com/ct- scan/brief-history-of-ct. Accessed on: 13/12/17. 110. Imai, S., Akahane, M., Konishi, Y., Imamura, T. (2018). Benefits of computed tomography in reducing mortality in emergency medicine. Open Medicine 13(1),394-401. doi:10.1515/med-2018-0058. 239 University of Ghana http://ugspace.ug.edu.gh 111. Inkoom., S., Schandorf, C., Boadu, M., Emi-Reynolds, G. Nkansah, A. (2014). Adult medical x-ray dose assessments for computed tomography procedures in Ghana - a review paper. Journal of Applied Science and Technology, 19 (1 and 2),1 – 9. 112. Institute of Physics and Engineering in Medicine, (2005). IPEM report 91: recommended standards for the routine performance testing of diagnostic x - ray imaging systems. London: U.K; IPEM. 113. International Electrotechnical Commission, (2011). Medical Electrical Equipment – Part 2-44: Particular requirements for the safety of X-ray equipment for computed tomography. International Electrotechnical Commission Standard 60601-2-44: 2009+A1. Geneva, IEC. 114. Jaffe, A.T, Hoang, J.K, Yoshizumi, T.T., Toncheva, G., Lowry, C., and Ravin, C. (2010). Radiation Dose for Routine Clinical Adult Brain CT: Variability on Different Scanners at One Institution. American Journal of Roentgenology, 195(1),433-438. 10.2214/AJR.09.3957. 115. Japan Network for Research and Information on Medical Exposures (2015). Diagnostic reference levels based on latest surveys in Japan — Japan DRLs 2015. Retrieved from http://www.radher.jp/J-RIME/report/DRLhoukokusyoEng.pdf Accessed on: 28/8/18. 116. Järvinen, H., Seuri, R., Kortesniemi, M., Lajunen, A., Hallinen, E., Savikurki- Heikkilä, P., Laarne, P., Perhomaa, M., Tyrväinen, E. (2015). Indication-based national diagnostic reference levels for paediatric CT: a new approach with proposed values. Radiation Protection Dosimetry, 165(1-4),86-90. 240 University of Ghana http://ugspace.ug.edu.gh 117. Järvinen, H., Vassileva, J., Samei, E., Wallace, A., Vano, E., Rehani, M. (2017). Patient dose monitoring and the use of diagnostic reference levels for the optimisation of protection in medical imaging: current status and challenges worldwide. Journal of medical imaging (Bellingham, Wash.), 4(3), 031214. doi:10.1117/1.JMI.4.3.031214 118. Kaasalainen, T., Palmu, K., Reijonen, V., Kortesniemi, M. (2014). Effect of patient centering on patient dose and image noise in chest CT. American Journal of Roentgenology, 203(1),123-130. 119. Kalra, M.K., Maher, M.M., Toth, T.L., Hamberg, L.M., Blake, M.A., Shepard, J.- A., Saini, S. (2004). Strategies for CT Radiation Dose Optimisation. Radiology, 230(3), 619–628. doi:10.1148/radiol.2303021726 120. Kanal, K.M., Butler, P.F., Sengupta, D., Bhargavan-Chatfield, M., Coombs, L.P., Morin, R.L. (2017). U.S. diagnostic reference levels and achievable Doses for 10 adult CT examinations. Radiology, 284(1),120-133. 121. AL-Jasim A.K., Hulugalle, S.N.C.W.M.P.S.K., Al-Hamadani, H.K. (2017). A quality control test for general X-ray machine. World Scientific News, 90 (1); 11-30. 122. Katan, M., Luft, A. (2018). Global Burden of Stroke. Seminars Neurology, 38(2),208-211. doi: 10.1055/s-0038-1649503. Epub 2018 May 23. 123. Kim, M., Lee, J.M., Yoon, J.H., Son, H., Choi, J.W., Han, J.K., Choi, B.I. (2014). Adaptive iterative dose reduction algorithm in CT: effect on image quality compared with filtered back projection in body phantoms of different sizes. Korean Journal of Radiology, 15(2),195–204. doi:10.3348/kjr.2014.15.2.195 241 University of Ghana http://ugspace.ug.edu.gh 124. Koo, T.K., Li, M.Y. (2016). A guideline of selecting and reporting intraclass correlation coefficients for reliability research. Journal of Chiropractic Medicine, 15(2),155–163. doi:10.1016/j.jcm.2016.02.012 125. Korir G.K., Wambani, J.S., Korir, I.K. (2012). Patient doses using multidetector computed tomography scanners in Kenya. Radiation Protection Dosimetry, 151(1),267– 271. 126. Korir, G. K., Wambani, J. S., Korir, I. K., Tries, M. A., and Boen, P. K. (2016). National diagnostic reference level initiative for computed tomography examinations in Kenya. Radiation Protection Dosimetry, 168(2),242–252. doi:10.1093/rpd/ncv020 127. Korir, G.K., Wambani, J.S., Korir, I.K., Tries, M., Mulama, B. (2013). Quality management systems in radiology. South African Journal of Radiology. 17(3):84-88. DOI:10.7196/SAJR.886. 128. Kruskal B.J., Eisenberg, R., Sosna, J., Yam, S.C., Kruskal, J.D., Boiselle, MP. (2011). Quality improvement in radiology: Basic principles and tools required to achieve success. RadioGraphics, 31(1),1499–1509. 129. Kruskal, J.B., Anderson, S., Yam, C.S., Sosna, J. (2009). Strategies for Establishing a comprehensive quality and performance improvement program in a Radiology Department. Radiographics, 29(2),315-330. 130. Lajunen, A. (2015). Indication-based diagnostic reference levels for adult CT- examinations in Finland. Radiation Protection Dosimetry. 165(1-4),95-97. 131. Lee, C., Kim, K.P., Long, D.J., Bolch, W.E. (2012). Organ doses for reference pediatric and adolescent patients undergoing computed tomography estimated by Monte Carlo simulation. Medical Physics, 39(4),2129–2146. doi:10.1118/1.3693052. 242 University of Ghana http://ugspace.ug.edu.gh 132. Lee, C., Kim, K.P., Long, D., Fisher, R., Tien, C., Simon, S.L, … Bolch, W.E. (2011). Organ doses for reference adult male and female undergoing computed tomography estimated by Monte Carlo simulations. Medical. Physics. 38(1),1196–206. 133. Lee, C., Kim, P., K., Bolch, E., W., Moroz1 B., E., and Folio, L. (2015). NCICT: a computational solution to estimate organ doses for pediatric and adult patients undergoing CT scans. J. Radiol. Prot. 35 (2015), 891–909. 134. Liang, C.R., Chen, P.X.H., Kapur, J., Ong, M.K.L., Quek, S.T., and Kapur, S.C. (2017). Establishment of institutional diagnostic reference level for computed tomography with automated dose‐tracking software. Journal of Medical Radiation Sciences, 64(2),82– 89. http://doi.org/10.1002/jmrs.210 135. Lim, H., Choi, J., Kim, J.-H., Cheong, H.-K., Ha, M. (2018). Estimation of cancer incidence and mortality risks attributed to diagnostic medical radiation exposure in Korea, 2013. Journal of Korean Medical Science, 33(29),1-11. doi:10.3346/jkms.2018.33.e211. 136. Lira, D., Padole, A., Kalra, M.K., Singh, M. (2015). Tube potential and CT Radiation dose optimisation. American Journal of Roentgenology, 204(1),W4-W10. 137. Liu, H., Gu, J., Caracappa, P.F Xu, X.G. (2010). Comparison of two types of adult phantoms in terms of organ doses from diagnostic CT procedures. Physics in Medicine & Biology. 55(1),1441–51. 138. Lobiondo-Wood, G, Haber, J. (2013). Nursing research in Canada. Methods, critical appraisal, and utilisation. 3rd Canadian edn. Toronto: Elsevier. 139. MacGregor, K., Li, I., Dowdell, T., Gray, B.G. (2015). Identifying institutional diagnostic reference levels for CT with radiation dose index monitoring software. Radiology, 276(2),507-515. 243 University of Ghana http://ugspace.ug.edu.gh 140. Mafalanka, F., Etard, C., Rehel, J.L., Pesenti-Rossi, D., Amrar-Vennier, F., Baron, N., … Aubert, B. (2015). Establishment of diagnostic reference levels in cardiac CT in France: A need for patient dose optimisation. Radiation Protection Dosimetry, 164(1- 2),116–119. 141. Manghani K. Quality assurance: Importance of systems and standard operating procedures. Perspect Clin Res. 2011; 2(1),34-7. 142. Marder, C. P., Narla, V., Fink, J. R., Tozer Fink, K. R. (2014). Subarachnoid Hemorrhage: Beyond Aneurysms. American Journal of Roentgenology, 202(1), 25– 37. doi:10.2214/ajr.12.9749. 143. Martin, C.J. (2016). The application of diagnostic reference levels for optimisation of x-ray imaging in the UK. Radiation Protection Dosimetry, 169(1-4),211-6. 144. Martin, C.J. and Vano, E., (2018). Diagnostic reference levels and optimisation in radiology: where do we go from here? Journal of Radiological Protection. 38 (1),E1–E4 145. Maskell, G., Smith, K., Keevil, S. (2015). CT equipment, operations, capacity and planning in the NHS. Retrieved from https://www.sor.org/sites/default/files/document- versions/ct_equipment_in_the_nhs_report_-_cib_may_2015_sk_v2.0_1.pdf.Acess on:26/2/2019. 146. Matthews K, Brennan P.C. (2009). The Application of diagnostic reference levels: General principles and an Irish perspective. Radiography,15(2),171-178. 147. Mayo Clinic, (2020). Stroke. Retrieved from https://www.mayoclinic.org/diseases- conditions/stroke/symptoms-causes/syc-20350113. Accessed on: 09/01/2020. 244 University of Ghana http://ugspace.ug.edu.gh 148. Mazrani, W., McHugh, K., Marsden, P. J. (2007). The radiation burden of radiological investigations. Archives of Disease in Childhood, 92(12),1127–1131. doi:10.1136/adc.2006.101782 149. McCollough, C.H. (2017). Diagnostic Reference Levels. Retrieved from http://www.imagewisely.org/Imaging-Modalities/Computed-Tomography/Medical- Physicists/Articles/Diagnostic-Reference-Levels. Accessed on: 25th August, 2017. 150. McCollough, C.H., Leng, S., Yu, L., Cody, D.D., Boone, J.M., McNitt‐Gray, M.F. (2011). CT dose index and patient dose: they are not the same thing. Radiology, 259(1),311–316. 151. McElroy, L.M., Ladner, D.P. (2014). Defining the Study Cohort: Inclusion and Exclusion Criteria. In: Pawlik T., Sosa J. (eds) Success in Academic Surgery: Clinical Trials. Success in Academic Surgery. London: Springer. 152. McHugh, M.L. (2012). Interrater reliability: the kappa statistic. Biochemia Medica, 22(3),276–282. 153. McLean, D., Chapple C.L. (2015). CT dosimetry. In: C.J Martin, and D.G Sutton (Eds). Practical Radiation Protection in Healthcare (2 ed). Oxford: Oxford University Press. 154. Merzan, D., Nowik, P., Poludniowski, G., Bujila, R. (2017). Evaluating the impact of scan settings on automatic tube current modulation in CT using a novel phantom. British Journal of Radiology, 90(20160308),1-14. 155. Miglioretti, D.L., Johnson, E., Williams, A., Greenlee, R.T., Weinmann, S., Solberg, L.I., … Smith-Bindman, R. (2013). The use of computed tomography in pediatrics and the associated radiation exposure and estimated cancer risk. The Journal of the 245 University of Ghana http://ugspace.ug.edu.gh American Medical Association pediatrics, 167(8),700–707. doi:10.1001/jamapediatrics.2013.311 156. Minami, Y. Kudo, M. (2015). Imaging modalities for assessment of treatment response to nonsurgical hepatocellular carcinoma therapy: Contrast-enhanced US, CT, and MRI. Liver Cancer, 4(2),106–114. 157. Ministry of Health (2014). Health Ministry to supply new equipment to 85 hospitals. Retrieved from https://www.ghanaweb.com/GhanaHomePage/NewsArchive/Health-Ministry-to-supply- new-equipment-to-85-hospitals-296975. Accessed on: 20/18/18 158. Muhogora, W.E. Ahmed, N.A., Beganovic, A., Benider, A., Ciraj-Bjelac, O., Gershan, V., …Rehani, M.M. (2009). Patient doses in CT examinations in 18 countries: initial results from international atomic energy agency projects. Radiation Protection Dosimetry, 136 (2),1–9. 159. Muhogora, W.E., Ahmed, N.A., AlSuwaidi, J.S., Beganovic, A., Ciraj-Bjelac, O., Gershan, V., … Rehani, M.M. (2010). Paediatric CT examinations in 19 developing countries: frequency and radiation dose. Radiation Protection Dosimetry, 140(1),49–58. doi:10.1093/rpd/ncq015 160. Mullenders, L., Atkinson, M., Paretzke, H., Sabatier, L., Bouffler, S. (2009). Assessing cancer risks of low-dose radiation. Nature Reviews Cancer, 9(1),596–604. 161. National Cancer Institute, (2017). NCI Dictionary of Cancer Terms. Retrieved from https://www.cancer.gov/publications/dictionaries/cancer-terms?cdrid=348991. Accessed on; 21/08/2017. 246 University of Ghana http://ugspace.ug.edu.gh 162. National Cancer Institute, (2019). Computed Tomography (CT) Scans and Cancer. Retrieved from https://www.cancer.gov/about-cancer/diagnosis-staging/ct-scans-fact- sheet. Accessed on: 24/01/20. 163. National Council on Radiation Protection and Measurements, (1993). Ionising radiation exposure of the population of the United States. NCRP, Bethesda, MD 164. National Council on Radiation Protection and Measurements, (2012). Reference levels and achievable doses in medical and dental imaging: recommendations for the United States. Bethesda, MD: NRCP, 2012: report 172 165. National Research Council, (2006). Health Risks from Exposure to Low Levels of Ionising Radiation: BEIR VII Phase 2. Washington, DC: The National Academies Press. https://doi.org/10.17226/11340. BEIR VII Report Pg 259-267 (2006). 166. Neuburger, J., Walker, K., Sherlaw-Johnson, C., van der Meulen, J., Cromwell, D.A. (2017). Comparison of control charts for monitoring clinical performance using binary data. BMJ Quality and Safety, 26(11),919–928. doi:10.1136/bmjqs-2016-005526 167. Ngoya, P.S, Muhogora, W.E, Pitcher, R.D. (2016). Defining the diagnostic divide: an analysis of registered radiological equipment resources in a low-income African country. The Pan African Medical Journal, 25(99),1-7. doi:10.11604/pamj.2016.25.99.9736 168. Ngoye, W.M, Motto, J.A, Muhogora, W.E. (2015). Quality control measures in Tanzania: Is it done? Journal of Medical Imaging and Radiation Sciences, 46(3),S23 - S30. 169. Nikupaavo, U., Kaasalainen, T., Reijonen, V., Ahonen, S.-M., and Kortesniemi, M. (2015). Lens dose in routine head CT: Comparison of different optimisation methods with 247 University of Ghana http://ugspace.ug.edu.gh anthropomorphic phantoms. American Journal of Roentgenology, 204(1),117– 123. doi:10.2214/ajr.14.12763 170. O’Connel, W.J, (2015). Diagnostic Reference Levels and Achievable Doses in Medical Imaging. GE Healthcare. Retrieved from http://www3.gehealthcare.co.uk/~/media/downloads/uk/services/dose%20management/ec r%20highlights/diagnostic%20reference%20levels%20and%20achievable%20doses.pdf. Accessed on: 31/August, 2017 171. OECD, (2019). Computed tomography (CT) exams (indicator). doi: 10.1787/3c994537-en (Accessed on 08 July 2019). Retrieved from https://data.oecd.org/healtheqt/computed-tomography-ct-scanners.htm. 172. Ofori, E.K, Antwi, W.K, Scutt, D.N. (3013). Current status of quality assurance in diagnostic imaging departments in Ghana. The South African Radiographer, 51(2),19-25. 173. Ozdiev, A.H., Afornu, B.K., and Sednev, D.A. (2018). Optimised filtered back- projection tomographic reconstruction algorithm for the step-shift scanning of a sample. Research in Nondestructive Evaluation, 30(3),179-187. doi:10.1080/09349847.2018.1498960 174. Padole, A., Khawaja, A.D.R., Kalra, M.K, Singh, S. (2015). CT Radiation Dose and Iterative Reconstruction Techniques. American Journal of Roentgenology, 204(4), W384-W392. 175. Pai, R., Modh, R., Lamoureux, R.H., Deitte, L., Wymer, D.C., Mench, A., … Canales, B.K. (2018). Image quality and patient-specific organ doses in stone protocol CT: A comparison of traditional CT to low dose CT with iterative reconstruction. BioMed Research International, 2018(1),1–6. doi:10.1155/2018/5120974 248 University of Ghana http://ugspace.ug.edu.gh 176. Palorini, F., Origgi, D., Granata, C., Matranga, D., Salerno, S. (2014). Adult exposures from MDCT including multiphase studies: First Italian nationwide survey. European Radiology. 24(2),469–83. 177. Pan, T. (2009). Computed tomography: from photon statistics to modern cone- beam CT. Journal of Nuclear Medicine, 50(7),1194– 1194. doi:10.2967/jnumed.109.064501 178. Pandharipande, P.V, Reisner, A.T, Binder, W.D, Zaheer, A., Gunn, M.L, ….. Gazelle, G.S. (2016). CT in the emergency department: a real-time study of changes in physician decision making. Radiology, 278(3),812–821 179. Patschan, D., Buschmann, I., Ritter, O. (2018). Contrast-induced nephropathy: update on the use of crystalloids and pharmacological measures. International Journal of Nephrology, 2018(1),1–8. doi:10.1155/2018/5727309 180. Pearce, M.S., Salotti, J. A., Little, M.P., McHugh, K., Lee, C., Kim, K.P., … de González, A.B. (2012). Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study. Lancet, 380(9840),499– 505. 181. Pedersen, C.C.E., Hardy, M., Blankholm, A.D. (2018). An Evaluation of Image Acquisition Techniques, Radiographic Practice, and Technical Quality in Neonatal Chest Radiography. Journal of Medical Imaging and Radiation Sciences, 49(3),257– 264. doi:10.1016/j.jmir.2018.05.006 182. Public Health England, (2016). National Diagnostic Reference Levels (NDRLs) Vol. 11. Retrieved from https://www.gov.uk/government/publications/diagnostic- 249 University of Ghana http://ugspace.ug.edu.gh radiology-nationaldiagnostic-reference-levels-ndrls/national-diagnostic-reference- levelsndrls#fnref:2 Accessed on 21/11/2016. 183. Purysko, C.P., Renapurkar, R., Bolen, A.M. (2016). When does chest CT require contrast enhancement? Cleveland Clinic Journal of Medicine, 83(6),423-426. 184. Pyfferoen, L., Mulkens, T.H., Zanca, F., De Bondt, T., Parizel, P.M., and Casselman, J.W. (2017). Benchmarking adult CT-dose levels to regional and national references using a dose-tracking software: a multicentre experience. Insights into Imaging, 8(5),513–521. http://doi.org/10.1007/s13244-017-0570-5 185. Qureshi, Z. (2019). The Role of Public Policy in Sustainable Infrastructure. Retrieved from https://www.brookings.edu/wp-content/uploads/2016/07/public-policy- sustainable-infrastructure-qureshi-1.pdf. Accessed on: 5/4/19 186. Radiation and Nuclear Safety Authority, (2013). Reference levels for patient radiation exposure in computed tomography examinations of adults. Helsinki: Radiation and Nuclear Safety Authority. 2013. 187. Radiologycafe, (2019). Production of X-rays. Retrieved from https://www.radiologycafe.com/radiology-trainees/frcr-physics-notes/production-of-x- rays. Accessed on: 29/11/2019. 188. Rahman, A.O.R (2018). Principles and Applications in Nuclear Engineering Radiation Effects. London: IntechOpen. 189. Raman, S.P., Mahesh, M., Blasko, R.V., and Fishman, E.K. (2013). CT Scan Parameters and Radiation Dose: Practical Advice for Radiologists. Journal of the American College of Radiology, 10(11),840–846. 250 University of Ghana http://ugspace.ug.edu.gh 190. Razali, M.A.S.M., Ahmad, M.Z., Roslee, M.A.A.M., Osman, N.D. (2019). Establishment of institutional diagnostic reference level for CT imaging associated with multiple anatomical regions. Journal of Physics: Conference Series, 1248(1),1-5. doi:10.1088/1742-6596/1248/1/012067 191. Rehani, M.M. (2009). Patient doses in CT examinations in 18 countries: Initial results from international atomic energy Agency projects. Radiation Protection Dosimetry, 136(2),118-26. doi: 10.1093/rpd/ncp144. 192. Rehani, M.M. (2015). Limitations of diagnostic reference level (DRL) and introduction of acceptable quality dose (AQD). British Journal of Radiology. 88(1045), 1- 3. doi: 10.1259/bjr.20140344 193. Ria, F., Davis, J.T., Solomon, J.B., Wilson, J.M., Smith, T.B., Frush, D.P. (2019). expanding the concept of diagnostic reference levels to noise and dose reference levels in CT. American Journal of Roentgenology, 213(4),889-894. doi: 10.2214/AJR.18.21030. Epub 2019 Jun 10. 194. Ridley, E.L. (2016). DRLs should be tied to clinical indication, not anatomy Retrieved from http://www.auntminnie.com/index.aspx?sec=roadandsub=ris_2016andpag=disanditemId =115607. Accessed on:12/12/17. 195. Rogers, A.T. (2014). An approach to local diagnostic reference levels (DRL's) in the context of national and international DRL's. IAEA-CN-85-77. 440-443. Retrieved from http://www.iaea.org/inis/collection/NCLCollectionStore/_Public/32/039/32039907.pdf. Accessed on: 29/08/2017. 251 University of Ghana http://ugspace.ug.edu.gh 196. RTI Group (2016). RTI CT Dose Profiler. Retrieved from http https://rtigroup.com/accessories/detail/rti-ct-dose-profiler [Last accessed on May 2017]. 197. Rusandu, A., Ødegård, A., Engh, G.C., Olerud, H.M. (2018). The use of 80 kV versus 100 kV in pulmonary CT angiography: An evaluation of the impact on radiation dose and image quality on two CT scanners. Radiography, 25(1),58–64. 198. Sackey, T.A., Schandorf, C., Fletcher, J. J., Mensah, Y.B., Shirazu1, I., Akyea- Larbi, K.O., … Odonkor S.T. (2018). Cancer risk assessment of patients undergoing computed tomography examination at the Korle-Bu Teaching Hospital. International Journal of Scientific Research in Science and Technology, 4(2),861-868. 199. Salama, D.H., Vassileva, J. Mahdaly, G., Shawki, M., Salama, A., Gilley, D., Rehani, M.M. (2017). Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Physica medica, 39(1),16–24. 200. Saltybaeva, N., Martini, K., Frauenfelder, T., Alkadhi, H. (2016). Organ dose and attributable cancer risk in lung cancer screening with low-dose computed tomography. PLoS ONE, 11(5),e0155722. doi:10.1371/journal.pone.0155722 201. Santos, J., Foley, S., Paulo, G., McEntee, M., Rainford, L. (2014). The establishment of computed tomography diagnostic reference levels in Portugal. Radiation Protection Dosimetry, 2014;158(3),307-317. 202. Sanuade, O.A., Dodoo, F.N-A., Koram, K., de-Graft Aikins, A. (2019). Prevalence and correlates of stroke among older adults in Ghana: Evidence from the study on global ageing and adult health (SAGE). PLoS ONE, 14(3),1-17 e0212623. https://doi.org/10.1371/journal.pone.0212623 252 University of Ghana http://ugspace.ug.edu.gh 203. Saravanakumar, A., Vaideki, K., Govindarajan, K.N., Jayakumar, S., Devanand, B. (2017). Assessment of regional pediatric computed tomography dose indices in Tamil Nadu. Journal of Medical Physics, 42(1),48–54. doi:10.4103/0971-6203.202425 204. Schandorf, C., Tetteh, G.K. (1998). Analysis of the status of X-ray diagnosis in Ghana. The British Journal of Radiology. 71(850),1040-1048. 205. Schegerer, A.A., Nagel, H.D., Stamm, G., Adam, G., Brix, G. (2017). Current CT practice in Germany: Results and implications of a nationwide survey. European Journal of Radiology. 90(1),114–28. 206. Sergieva, S., Mihaylova, I., Alexandrova, E., Dimcheva, M., Mansi, L. (2015). SPECT-CT in Radiotherapy Planning, with Main Reference to Patients with Breast Cancer. Current Radiopharmaceuticals, 8(1),9–18. https://doi.org/10.2174/1874471008666150316221722 207. Shah, D.J., Sachs R.K., Wilson D.J. (2012). Radiation-induced cancer: a modern view. British Journal of Radiol., 85(1020),1166–1173 208. Shefer, E., Altman, A., Behling, R., Goshen, R., Gregorian, L., Roterman, Y., … Zarchin, O. (2013). State of the art of CT detectors and sources: A literature review. Current Radiology Reports, 1(1),76–91. 209. Shirazu, I. Mensah, Y.B. Schandorf, C. Mensah, S.Y. (2017a). Estimate of reference effective dose and renal dose during abdominal CT scan for dose optimisation procedures in Ghana. International Journal of Scientific and Technology Research, 6(2),215-224 253 University of Ghana http://ugspace.ug.edu.gh 210. Shirazu, I., Mensah, Y. B., Schandorf, C., Mensah, S., Y. (2017b). Determination of Standard Reference Body Indices for Clinical Application in Ghana. International Journal of Scientific and Technology Research, 6(2),225-231. 211. Shrimpton, P, Wall B, Jones D, Fisher E, Hillier M, Kendall G, et al. A national survey of doses to patients undergoing a selection of routine X-ray examinations in English hospitals. NRPB Report R200. Chilton, UK: National Radiological Protection Board; 1986 212. Silvestrin, C. (2016). Europe’s Looming Radiology Capacity Challenge A Comparative Study. TMC. Retrieved from: https://www.telemedicineclinic.com/wp- content/uploads/2016/11/Europes_looming_radiology_capacity_challenge- A_comparitive_study.pdf. Accessed on: 11/1/2020. 213. Simantirakis, G., Hourdakis, C.J., Economides, S., Kaisas, I., Kalathaki, M., Koukorava, C., … Dimitriou, P. (2015). Diagnostic reference levels and patient doses in computed tomography examinations in Greece, Radiation Protection Dosimetry, 163(1), 319–24. 214. Skinner, S. (2015). Guide to thoracic imaging. The Royal Australian College of General Practitioners, 44(8),558-563. 215. Smith-Bindman, R., Miglioretti, D.L. (2011). CTDIvol, DLP, and effective dose are excellent measures for use in CT quality improvement. Radiology, 261(3),999. https://doi.org/10.1148/radiol.11111055 216. Smith-Bindman, R., Lipson, J., Marcus, R., Kim, K.P., Mahesh, M., Gould, R., … Miglioretti, D.L. (2009). Radiation dose associated with common computed tomography examinations and the associated lifetime attributable risk of Cancer. Archives of Internal Medicine, 169(22),2078–2086. 254 University of Ghana http://ugspace.ug.edu.gh 217. Smith-Bindman, R., Wang, Y., Chu, P., Chung R., Einstein, A. J.J., Balcombe, J… Miglioretti, D.L. (2019). International variation in radiation dose for computed tomography examinations: prospective cohort study. British Medical Journal, 364(k4931), 1-11 218. Sprawls, 2019. Computed Tomography Image Formation. Retrieved from http://www.sprawls.org/resources/CTIMG/module.htm#beginning. Accessed on: 2/12/19 219. Statista, Number of computer tomography (CT) scanners in selected countries as of 2017 (per million population). Retrieved from https://www.statista.com/statistics/266539/distribution-of-equipment-for-computer- tomography/. Accessed on 1/3/2019. 220. Statisticshowto, (2019). Calculating Test-Retest Reliability Coefficients. Retrieved from https://www.statisticshowto.datasciencecentral.com/test-retest-reliability/. Accessed on: 4th /06/19. 221. Stoltzfus, J.C. (2011). Logistic regression: a brief primer. Academic Emergency Medicine, 18(10),1099-104. doi: 10.1111/j.1553-2712.2011.01185.x. 222. Storrs, C. (2013). How Much Do CT Scans Increase the Risk of Cancer? Retrieved from https://www.scientificamerican.com/article/how-much-ct-scans-increase-risk- cancer/ Accessed on: 8/08/2017. 223. Stradiotti, P., Curti, A., Castellazzi, G., andamp; Zerbi, A. (2009). Metal-related artifacts in instrumented spine. Techniques for reducing artifacts in CT and MRI: state of the art. European Spine Journal, 18(Suppl 1),102–108. 224. Strauss, K.L., Somasundaram, E., Sengupta, D., Marin, J.R., Brady, S.L. (2019). Radiation Dose for Pediatric CT: Comparison of Pediatric versus Adult Imaging Facilities. Radiology. 291(1), 158–167. 255 University of Ghana http://ugspace.ug.edu.gh 225. Sulemana, H., Inkoom, S., Sosu, E.K, Schandorf, C. (2020). Estimation of absorbed and effective doses in organs through computed tomography examinations using automatic exposure control and fixed tube current techniques: A phantom case study. Iranian Journal of Medical Physics, 17(1),58-65. 10.22038/ijmp.2019.34196.1432 226. Sutton, D.G., McVey, S., Gentle, D., Hince, A.J., MacDonald, N., McCallum, S. (2014). CT chest abdomen pelvis doses in Scotland: has the DRL had its day? The British Journal of Radiology, 87(1041), 20140157. http://doi.org/10.1259/bjr.20140157 227. Tabachnick, B.G, Fidell, L.S. (2007). Using Multivariate Statistics. 5th ed. Boston, MA: Pearson Education, Inc 228. Taylor, C.R. (2017). Abdominal Computed Tomography Scanning. Retrieved from https://emedicine.medscape.com/article/2114236-overview. Accessed on: 7/01/2020 229. Teach Nuclear, (2019). Effects of Ionising Radiation on DNA. Retrieve from: https://teachnuclear.ca/all-things-nuclear/radiation/biological-effects-of-radiation/effects- of-Ionising-radiation-on-dna/ 230. Trattner, S., Pearson, G., Chin, C., Cody, D.D., Gupta, R., Hess, C.P., … Einstein, A. J. (2014). Standardisation and optimisation of CT protocols to achieve low dose. Journal of the American College of Radiology, 11(3),271–278. doi: 10.1016/j.jacr.2013.10.016 231. Treier, R., Aroua, A., Verdun, F.R., Samara, E., Stuessi, A., Trueb, P.R. (2010). Patient doses in CT examinations in Switzerland: implementation of national diagnostic reference levels. Radiation Protection Dosimetry. 142(2–4),244–54. 232. Tsukagoshi, S., Ota T., Fujii, M., Kazama, M., Okumura, M., Johkoh, T (2007). Improvement of spatial resolution in the longitudinal direction for isotropic imaging in helical CT. Physics in Medicine and Biology, 52 (3),791-80. 256 University of Ghana http://ugspace.ug.edu.gh 233. Tsukamoto, A. (2017). Explanation of the DRLs 2015 Published in Japan. Japanese Journal of Radiological Technology, 73(7),565-570. 234. Tzedakis, A., Damilakis, J, Perisinakis, K. and Stratakis, J. (2005). The effect of z overscanning on patient effective dose from multidetector helical computed tomography examinations. Medical Physics, 32(6),1621-1629. 235. United Nations Scientific Committee on the Effect of Atomic Radiation (UNSCEAR), (2000). Sources and effects of Ionising radiation. United Nations Scientific Committee on the Effects of Atomic Radiation UNSCEAR 2000 Report to the General Assembly. Volume 1, New York: United Nations. 236. United State Environmental Protection Agency, (1999). Estimating radiographic cancer risks. Addendum: uncertainty analysis. Washington, DC: US EPA. 237. United States Food and Drug Administration (1984). Diagnostic X-Ray Systems and Their Major Components. Code of Federal Regulations. 21 CFR 1020.33 238. United States Food and Drug Administration, (2017). What are the Radiation Risks from CT?. Retrieved from https://www.fda.gov/radiation-emitting-products/medical-x- ray-imaging/what-are-radiation-risks-ct. accessed on: 12/05/2017. 239. UNSCEAR (2012). Sources and effects of Ionising radiation. United Nations Scientific Committeeon the Effects of Atomic Radiation, Report to General Assembly. Annex B Exposure from Natural Radiation Sources. 2012. New York United Nations. 240. van der Molen, A.J., Schilham, A., Stoop, P., Prokop, M., Geleijns, J. (2013). A national survey on radiation dose in CT in The Netherlands. Insights into Imaging, 4(3),383–390. doi:10.1007/s13244-013-0253-9 257 University of Ghana http://ugspace.ug.edu.gh 241. Vanckavicienea, A., Starkienec, L., Macijauskiene, J. (2014). Supply and demand for radiographers in Lithuania: A prognosis for 2012–2030. European Journal of Radiology, 83(7),1292-1300. 242. Vano, E., Lau, L., Kwan-Hoong, Ng. (2014). How radiation protection influences quality in radiology. In Lau L, Kwan-Hoong Ng (eds). Radiological Safety and Quality: Paradigms in Leadership and Innovation (pp 35-54). London: Springer. 243. Varghese, A., Keshava, S.N., Moses, V., Koshy, G., Mammen, S., Ahmed, M., Livingstone, R.S. (2019). Radiation dose reference card for interventional radiology procedures: Experience in a tertiary referral centre. The Indian Journal of Radiology and Imaging, 29(3),247–252. doi:10.4103/ijri.IJRI_35_19 244. Vassileva, J. and Rehani, M. (2015). Diagnostic reference levels. American Journal of Roentgenology, 204(1),W1-W3. 245. Vassileva, J., Rehani, M., Kostova-Lefterova, D., Al-Naemi, H.M., Al Suwaidi, J. S., Arandjic, D., … Zaman, A. (2015). A study to establish international diagnostic reference levels for paediatric computed tomography. Radiation Protection Dosimetry, 165(1-4),70–80. doi:10.1093/rpd/ncv116 246. Vassileva, J., Rehani, M.M (2011). IAEA survey of pediatric CT practice in 40 countries in Asia, Europe, Latin America, and Africa: Part 1, frequency and appropriateness. American Journal of Roentgenology, 198(1),1021–1031. 247. Vawda, Z., Pitcher, R., Akudugu, J., Groenewald, W. (2015). Diagnostic reference levels for paediatric computed tomography. South African Journal of Radiology, 19(2),1- 4. 258 University of Ghana http://ugspace.ug.edu.gh 248. Vock, P., Frija, G. (2016). Diagnostic reference levels based on clinical indications. Poster Prsentation. EuroSafe Imaging Congress. No ESI-0034. DOI: 10.1594/esi2016/ESI-0034 249. Vom, J., Williams, I. (2017). Justification of radiographic examinations: What are the key issues?. Journal of Medical Radiation Sciences, 64(3), 212–219. doi:10.1002/jmrs.211 250. Wachabauer, D., Röthlin, F. (2017). Aktualisierung der diagnostischen Referenzwerte für Österreich. Wien; 2017. Retrieved from https://jasmin.goeg.at. Accessed on: 27/05/2018. 251. Wagner, F., Bize, J., Racin, D., Le Coultre, R., Verdun, F., Trueb, P.R., Treier, R. (2018). Derivation of new diagnostic reference levels for neuro-paediatric computed tomography examinations in Switzerland. Journal of Radiological. Protection, 38 (2018),1013–1036. 252. Wambani, J.S., Korir, G.K., Onditi, E.G., Korir, I.K (2010). A survey of computed tomography imaging techniques and patient dose in Kenya. East African Medical Journal, 87(10),400 – 407. 253. Wardman, P. (2009). The importance of radiation chemistry to radiation and free radical biology (The 2008 Silvanus Thompson memorial lecture). British Journal of Radiology, 82(974),89-104. 254. Watson, D. J. Coakley, K.S. (2010). Paediatric CT reference doses based on weight and CT dosimetry phantom size: local experience using a 64-slice CT scanner. Pediatric Radiology, 40(5),693-703. doi: 10.1007/s00247-009-1469-1. 259 University of Ghana http://ugspace.ug.edu.gh 255. WHO, (2018). Diagnostic imaging. Retrieved from http://www.who.int/diagnostic_imaging/imaging_modalities/dim_plain- radiography/en/index4.html. Accessed on: 7th April, 2018. 256. WHO, (2019). Stroke: a global response is needed. Retrieved from https://www.who.int/bulletin/volumes/94/9/16-181636/en/. Accessed on 16/July/2019. 257. Widmark, A. (2018). Representative doses in Norway - 2017. Results from reporting and auditing and the establishment of new ones national reference values. Radiation Protection Report 2018: 3, Østerås: The Norwegian Radiation Protection Agency, Østerås 2018. 258. Willson, T., Larsen, B., Blecha, M., Connolly, M., Podbielski, F. (2014). Computed Tomography of the Chest: Indications and Utilisation in the Community Hospital Emergency Department. Chest, 145(3),533A. doi:10.1378/chest.1836673 259. World Health Organisation (WHO), (2016). Ionising radiation: Chapter 1: Scientific background. Retrieved from http://www.who.int/Ionising_radiation/pub_meet/radiation-risks-paediatric-imaging/en/ Accessed on: 29/July/2017 260. World Medical Association Declaration of Helsinki, (2001). Ethical Principles for Medical Research Involving Human Subjects. Bulletin of the World Health Organisation. 79(4), 373-374. 261. Yamauchi-Kawara, C., Fujii, K., Aoyama, T., Yamauchi, M., Koyama, S. (2010). Radiation dose evaluation in multidetector-row CT imaging for acute stroke with an anthropomorphic phantom. The British Journal of Radiology, 83(996),1029– 1041. doi:10.1259/bjr/52267127 260 University of Ghana http://ugspace.ug.edu.gh 262. Yeh, D.M., Tsai, H.Y., Tyan, Y.S., Chang, Y.C., Pan, L.K., Chen, T.R. (2016). The population effective dose of medical computed tomography examinations in Taiwan for 2013. PLoS ONE, 11(10), e0165526. https://doi.org/10.1371/journal.pone.0165526 263. Yu, L., Liu, X., Leng, S., Kofler, J.M., Ramirez-Giraldo, J.C., Qu, M., … McCollough, C.H. (2009). Radiation dose reduction in computed tomography: techniques and future perspective. Imaging in medicine, 1(1), 65–84. doi:10.2217/iim.09.5 264. Zarb, F., Rainford, L., McEntee, M.F. (2010). Image quality assessment tools for optimisation of CT images. Radiography, 16(2),147–153. doi:10.1016/j.radi.2009.10.002 265. Zhou, D-D., Sun, P., Jia, Z., Zhu, W., Shi, G., Kong, B., … Zhang, H. (2019). Multisection computed tomography: Results from a Chinese survey on radiation dose metrics. Journal of the Chinese Medical Association, 82(2),155–160 DOI: 10.1097/JCMA.0000000000000019 266. Zinsser, D., Maurer, M., Do, P.-L., Weiß, J., Notohamiprodjo, M., Bamberg, F., Othman, A.E. (2019). Reduced scan range abdominopelvic CT in patients with suspected acute appendicitis - impact on diagnostic accuracy and effective radiation dose. BMC Medical Imaging, 19(1),1-7. doi:10.1186/s12880-019-0304-x 267. Zira, D.J. Nzotta, C.C. (2016). The need to establish national dose reference levels for radiological examinations in Nigeria: radiographer's role. Nigerian Journal of Medical Imaging and Radiation Therapy, 5(1),25-40. 261 University of Ghana http://ugspace.ug.edu.gh APPENDICES APPENDIX I: NUMBER OF SCANNERS ON NRA’S RECORDS 262 University of Ghana http://ugspace.ug.edu.gh APPENDIX II: INTRODUCTORY LETTER FROM NRA 263 University of Ghana http://ugspace.ug.edu.gh APPENDIX III: PARTICIPANT INFORMATION SHEET 264 University of Ghana http://ugspace.ug.edu.gh 265 University of Ghana http://ugspace.ug.edu.gh 266 University of Ghana http://ugspace.ug.edu.gh APPENDIX IV: CONSENT FORM 267 University of Ghana http://ugspace.ug.edu.gh APPENDIX V: QUESTIONNAIRE A PART 1 (To be completed by a radiographer, radiologist or a medical physicist in charge as the technical head) 1. Your professional grade/rank……………………Gender:……………….. 2. Name of Centre/Hospital: …………………………………….…………………………… 3. Ownership: Private Public 4. Region of location Location………………………………….…………………................. 5. CT Type/Model: …..……………………………………………………………………….. 6. Manufacturer and Year of manufacture:…………………..……………………. 7. Year of installation:………………………………………………….…………………….. 8. Number of CT Slice and detector row: ……………………………………………………. 9. Scanning Mode:…………………………………………………………………………… 10. Is the CT equipment working?……………………………………………………………... 11. How many CT scanners do you have in your facility? ……………………………………. 12. Is your equipment having an automatic exposure control/tube current modulation system? Yes No 13. Number of CT operating radiographers. ……………………………...……………………. 14. Number of CT attending/reporting radiologists:………………………… ……….…. 15. Number of CT attending medical physicists……...............................................…………… 16. Are data on the following dose descriptors displayed on the control console? (i) CTDIvol Yes No (ii) DLP Yes No 17. Is there any AEC system incorporated in the CT in your facility? Yes No 18. How long does it normally take to repair your equipment when it is broken down? .......................... 19. Overall average number of CT cases per a year: …………………………………………. 20. Average number of CT cases per a year for the following: Head: ………………………………………………………………………………….…… Chest: ……………………………………………………………………………………… Abdomen: …………………………………………………………………………………. Pelvis……………………………………………………………………………………….. 268 University of Ghana http://ugspace.ug.edu.gh Abdominopelvic …………………………………………………………………………… Lumbar spine……………………………………………………………………………… Others…………………………………………………………………….. 21. Indicate the commonest indications and their average frequencies (example, 2 a day or 100 a year) for which adult patients come for Head CT examinations [Please select more than one(1) but not more than four (4)] For the selected ones, use alphabets (A-Z) to represents those that are scanned with uses similar scanning protocol. I. CVA/Stroke Frequency …………………… II. Head injury/trauma Frequency …………………… III. Tumour(Metastasis/cancers/lesion) Frequency …………………… IV. Sinusitis Frequency …………………… V. Facial bone injuries/trauma Frequency …………………… VI. Blurred vision Frequency …………………… VII. Dizziness Frequency …………………… VIII. Psychiatric disorders Frequency …………………… IX. Others and their frequencies, please specify ……………………………………………………………………………………… ……………………………………………………………………………………… …………………………………………………………………………………….... 22. Indicate the commonest indications and their average frequencies (example, 2 a day or 100 a year) for which adult patients come for Chest CT examinations [Please select more than one (1) but not more than three (3)] For the selected ones, use alphabets (A- Z) to represents those that are scanned with uses similar scanning protocol. I. Chest tumour (nodules/cancer/abscess) Frequency …………………… II. Chest injury/trauma Frequency …………………… III. Plueral effusion/Airway assessment Frequency …………………… IV. Interstitial lung diseases Frequency …………………… 269 University of Ghana http://ugspace.ug.edu.gh V. Pulmonary embolism Frequency …………………… VI. Others and their frequencies, please specify ……………………………………………………………………………………… ……………………………………………………………………………………… …………………………………………………………………………………….... 23. Indicate the commonest indications and their average frequencies (example, 2 a day or 100 a year) for which adult patients come for CT examinations involving the abdomino- pelvic region of the body [Please select more than one (1) but not more than five (5)] For the selected ones, use alphabets (A-Z) to represents those that are scanned with uses similar scanning protocol. I. Liver metastases Frequency …………………………………………….… II. Kidney stone/colic Frequency ………………………………………….…… III. Abdomino-pelvic abscess/tumour/cancer Frequency …………………………. IV. Pelvic tumour Frequency …………………………………….. V. Pancreatic tumour Frequency……………………………………… VI. Chrohn’s disease Frequency …. ………………………………………………. VII. Bowel obstruction Frequency …. …………………………………………… VIII. Ischemia Frequency …. …………………………………………… IX. Urothelial malignancy (CT-IVU) Frequency …. ………………………. X. CT Colonoscopy (for Polys/tumour) Frequency.………………………. XI. Others and their frequencies, please specify ……………………………………………………………………………………… ……………………………………………………………………………………… ……………………………………………………………………………………… 24. Which indication-based diagnostic reference levels (DRLs) do you compare your CT dose parameters with? Please specify.............................................................................. 25. Do you think there is the need to develop Indication-Based Diagnostic Reference Levels (IBDRLs) for CT examinations in Ghana? Yes No 270 University of Ghana http://ugspace.ug.edu.gh PART 2 i. Quality Assurance structures available at the CT facilities in Ghana Quality Assurance structures Availability Yes No 26. Availability of QA or QC Committee 27. Availability of a written down protocol for CT scanning 28. Post-major repair QC assessment records 29. Do you have records of regular QC checks 30. Certification/authorisation by Nuclear Regulatory Authority 31. Availability of established acceptance testing procedure 32. Availability of effective planned maintenance schedules 33. Equipment performance record keeping 34. Availability of radiation protection devices 35. Availability of systems for justifying CT exposures 36. Availability of accepted or established DRLs 37. Availability of scheduled dose optimisation programmes 38. Keeping of patient dose records 39. Availability of an established patient dose and image quality audit programmes 40. Planned schedules for frequent cleaning of equipment 41. Documented training program and records 4 .42. If yes to question 41, please indicate the frequency of the QC test……………… 271 University of Ghana http://ugspace.ug.edu.gh ii. Basic Quality improvement structures in the CT facilities Quality improvement structures Availability Yes No 43. Do you have institutional leadership and support toward quality improvement? 44. Do you have regular meetings involving all the stakeholders to communicate QC results? 45. Has your facility establish a culture of quality in your practice? 46. Has your facility established an improvement team? 47. Do you have a quality implementation protocol or manual 48. Do you have a system that engages all the professional groups in the department on quality improvements? 49. Do you have a system that regularly receives and analyses feedback from customers and stakeholders? 50. Do you have surveillances system for monitoring quality improvements indicators? 51. Does your facility have a system to reward hard work associated with quality improvement? 52. Do you have an educational programme on quality improvement? 272 University of Ghana http://ugspace.ug.edu.gh iii. Policy infrastructure and their availability in Ghana Please indicate whether or not the following are available in your facility or you are aware of it in the country by indicating yes or no against each statement Policy infrastructure Availability Yes No 53. Policy on CT authorisation for use in Ghana 54. Policy driving CT infrastructure and distribution 55. Policy on operation and maintenance 56. Policy and availability of quality management systems 57. Policy on purchasing, construction and installations 58. Policy on decommission of a CT facility 59. Policy on education and training 60. Policy and availability of diagnostic reference levels 61. Policy on CT referrer guidelines 62. Policy on recommended frequency for QC tests 63. Standardised policy on AEC application 64. Policy on patient dose optimisation, image quality and audit programmes 65. Policy on acceptance testing and record keeping 66. Policy on human resources in CT facilities 67. Policy on CT maintenance systems 273 University of Ghana http://ugspace.ug.edu.gh APPENDIX VI: QUESTIONNAIRE B (To be completed by a radiologist) Please tick (✔) the basic requirement that pertains to your facility regarding each of the routine indications i. Basic diagnostic imaging requirements for CVA/stroke, head injury and brain tumour/SOL procedures in Ghana. CVA/stroke Head injury Brain tumour/SOL Diagnostic imaging requirements Yes No Yes No Yes No Scan coverage: Scan should cover from just below the base of skull to the vertex Scan should cover from C3 to the vertex if the base of skull injury is suspected Scan series: None contrast Once (only IV contrast phase) Twice (both non-contrast and contrast phases) Image quality should be acceptable to the reporting radiologists and should meet national/international standards Slice thickness- 5 mm-10 mm <5 mm Scan mode: Helical only Axial only Both Scan Technique: Low dose Optimised dose High dose AEC usage: Yes No C3= Third Cervical vertebrae; CVA: cerebrovascular accident, IV: intravenous; SOL: space occupying lesion. 274 University of Ghana http://ugspace.ug.edu.gh Please tick (✔) the basic requirement that pertains to your facility regarding each of the routine indications ii. : Basic diagnostic imaging requirements for CT lung tumour, Chest lesion with CKD and PE procedures in Ghana Lung tumour CL CKD PE Diagnostic imaging requirements Yes No Yes No Yes No Scan coverage: Scan should cover from just above lung apices to below lung bases Scan series: Once (only non- contrast phase) Once (only IV routine contrast phase) Once (only IV contrast “angiogram” phase) Twice (both non-contrast and contrast “angiogram” phases Twice (both non-contrast and contrast phases) Twice (IV contrast scan of the lung filed and scan of the liver) Image quality should be acceptable to the reporting radiologists and should meet national/international standards Slice thickness- < 5 mm 5 mm-10 mm Scan mode: Helical only Axial only Both Scan Low dose technique: Optimised dose High dose AEC usage: Yes No Key: Chest lesion with CKD= CL CKD, PE: pulmonary embolism, IV: intravenous. 275 University of Ghana http://ugspace.ug.edu.gh Please tick (✔) the basic requirement that pertains to your facility regarding each of the routine indications iii. Basic diagnostic imaging requirements for abdomino-pelvic lesion, kidney stone and urothelial malignancy indication (CT-IVU) examinations AP lesion Kidney stone UM (IVU) Diagnostic imaging requirements Yes No Yes Yes No Yes Scan coverage: Scan should cover from just top of higher hemidiaphragm to below the ischium or symphysis pubis Scan series: Once (Non-contrast phase) - Once (only contrast phase; “oral with IV”) Twice (both non-contrast and contrast phases) *3-4 phases (pre-contrast and 2-3 other post IV contrast phases - including nephrographic, corticomedullary* and excretory phases) 2 phases (slip bolus technique involving nephrographic and excretory phases) Image quality should be acceptable to the reporting radiologists and should meet national/international standards Slice thickness: ≤ 5 mm 7-10 mm Scan mode: Helical only Axial only Both Scan technique: Low dose Optimised dose High dose AEC usage: Yes No Key: IV represents intravenous; UM=Urothelial malignancy. IVU: intravenous urography 276 University of Ghana http://ugspace.ug.edu.gh APPENDIX VII: CT DOSE PARAMETERS DATA SHEET CT DOSE PARAMETERS DATA SHEET (1) Name of Institution....................................................................................................CT MANUFACTURER/ MODEL.......................................................................................... Indication: CVA / STROKE Patient Age Sex Wt kVp mAs CTDIvol DLP SE P RT No. ST A SM C SCAN COVERAGE ID (mGy) (mGy.cm) of (mm) Scan length/range (mm) slices 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt. =weight, kVp= tube potential, mAs=milliampere second, SE= number of sequences, P=pitch, RT=rotation time, ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage 277 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (2) Name of Institution....................................................................................................CT MANUFACTURER/MODEL.................................................................................. Indication: HEAD INJURY / TRAUMA Patient Age Sex Wt kVp mAs CTDIvol DLP SE P RT No. ST A S C SCAN COVERAGE ID (mGy) (mGy.cm) of (mm) M Scan length/range (mm) slices 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt =weight, kVp= tube potential, mAs=milliampere second, SE= number of sequences, P=pitch, RT=rotation time, ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage 278 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (3) Name of Institution....................................................................................................CT MANUFACTURER/MODEL........................................................................................ Indication: HEADACHES? SOL/TUMOUR Patient Age Sex Wt kVp mAs CTDIvol DLP SE P RT No. of ST A SM C SCAN ID (mGy) (mGy.cm) Slices (mm) COVERAGE per SE SL (mm) Pre Post Pre Post Total CE CE CE CE DLP 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt =weight, kVp= tube potential, mAs=milliampere second, CE= contrast, SE= number of sequences, P=pitch, RT=rotation time,? = suspicion of, ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage, SL= scan length. 279 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (4) Name of Institution....................................................................................................CT MANUFACTURER/ MODEL......................................................................................... Indication: COUTH and CHEST PAIN? LUNG CANCER/TUMOUR Patient Age Sex Wt kVp mAs CTDIvol DLP SE P RT No. of ST A SM C SCAN ID (mGy) (mGy.cm) Slices (mm) COVERAGE Per SE SL (mm) Pre Post Pre Post Total CE CE CE CE DLP 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt =weight, kVp= tube potential, mAs=milliampere second, CE= contrast, SE= number of sequences, P=pitch, RT=rotation time,? = suspicion of, ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage, SL= scan length. 280 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (5) Name of Institution....................................................................................................CT MANUFACTURER /MODEL...................................................................................... Indication: SUSPICION OF LUNG LESION WITH CKD Patient Age Sex Wt kVp mAs CTDIvol DLP SE P RT No. Slice ST A SM C SCAN COVERAGE ID (mGy) (mGy.cm) of thickness (mm) Scan length/range slices (cm) (mm) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt =weight, kVp= tube potential, mAs=milliampere second, SE= number of sequences, P=pitch, RT=rotation time, ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage 281 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (6) Name of Institution....................................................................................................CT MANUFACTURER/ MODEL.......................................................................................... Indication: ? ABDOMINO-PELVIC TUMOUR/ LESION Patient Age Sex Wt. kVp mAs CTDIvol DLP SE P RT No. of ST A S C SCAN COVERAGE ID (mGy) (mGy.cm) Slices per (mm) M SL (mm) SE Pre Post Pre Post Total CE CE CE CE DLP 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt =weight, kVp= tube potential, mAs=milliampere second, CE= contrast, SE= number of sequences, P=pitch, RT=rotation time, ? = suspicion of, ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage, SL= scan length. 282 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (7) Name of Institution....................................................................................................CT MANUFACTURER/MODEL................. ................................................................... Indication: ? KIDNEY STONES Patient Age Sex Wt kVp mAs CTDIvol DLP SE P RT No. of ST A SM C SCAN COVERAGE ID (mGy) (mGy.cm) Slices per SE (mm) SL (mm) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt =weight, kVp= tube potential, mAs=milliampere second, SE= number of sequences, P=pitch, RT=rotation time, ? = suspicion of ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage, SL= scan length. 283 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (8) Name of Institution....................................................................................................CT MANUFACTURER/MODEL......................................................................................... Indication: ? UROTHELIAL MALIGNANCY (CT-IVU) ID Age Sex Wt kVp mAs CTDIvol DLP SE P RT No. of ST A SM C SCAN . (mGy) (mGy.cm) Slices (mm) COVERAGE per SE SL (mm) Pre Post CE series Pre Post CE series Total CE 1 2 3 4 5 CE 1 2 3 4 5 DLP 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Key: Wt =weight, kVp= tube potential, mAs=milliampere second, CE= contrast, SE= number of sequences, P=pitch, RT=rotation time, ? = suspicion of. ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage, SL= scan length. 284 University of Ghana http://ugspace.ug.edu.gh CT DOSE PARAMETERS DATA SHEET (9) Name of Institution................................... CT MANUFACTURER/ MODEL.................................................... Indication: ? PULMONARY EMBOLISM (Pulmonary Angiogram) Pati Age Sex Wt SE mAs kVp CTDIvol DLP P R No. ST A S C SCAN ent (mGy) (mGy.cm) T of M COVERAGE ID slices SL Total SE DLP DLP 1 PRE CE 1st SE 2nd SE 3rd SE 4th SE 5th SE 2 PRE CE 1st SE 2nd SE 3rd SE 4th SE 5th SE 3 PRE CE 1st SE 2nd SE 3rd SE 4th SE 5th 4 PRE CE 1st SE 2nd SE 3rd SE 4th SE 5th SE 5 PRE CE 1st SE 2nd SE 3rd SE 4th SE 5th SE 3rd SE 4th SE 5th SE This data sheet continues to page 20. Key: Wt =weight, kVp= tube potential, mAs=milliampere second, CE= contrast, SE= number of sequences, P=pitch, RT=rotation time,? = suspicion of. ST= slice thickness, A= AEC, SM=scan mode, C=contrast usage, SL= scan length. 285 University of Ghana http://ugspace.ug.edu.gh APPENDIX VIII: CONTROL CHART FOR ALL INDICATION DATA SETS (CTDIVOL AND DLP) 140 120 100 80 60 40 20 0 -20 CTDIvol Mean UCL LCL Figure 1: Control chart for CVA data sets (CTDIvol and DLP) 3000 2500 2000 1500 1000 500 0 -500 DLP MEAN UCL LCL Figure 2: Control chart for head injury/trauma data sets (CTDIvol and DLP) 140 120 100 80 60 40 20 0 -20 CTDIvol MEAN UCL LCL Figure 3: Control chart for brain tumour/SOL data sets (CTDIvol and DLP) 286 1 33 1 65 33 97 65 129 97 161 129 193 161 225 193 257 225 289 257 321 289 353 321 385 353 417 385 449 417 481 449 481 1 35 69 103 137 171 205 239 273 307 341 375 409 443 477 University of Ghana http://ugspace.ug.edu.gh 30 2000 25 1500 20 1000 15 10 500 5 0 0 -500 -5 -10 -1000 CTDIvol MEAN UCL LCL DLP MEAN UCL LCL Figure 4: Control chart for lung tumour/cancer data sets (CTDIvol and DLP) Figure 5: Control chart for chest lesion with CKD data sets (CTDIvol and DLP) Figure 6: Control chart for abdominopelvic lesion data sets (CTDIvol and DLP) 287 1 30 59 88 117 146 175 204 233 262 291 320 349 378 407 436 1 32 63 94 125 156 187 218 249 280 311 342 373 404 435 University of Ghana http://ugspace.ug.edu.gh Figure 7: Control chart for kidney stone data sets (CTDIvol and DLP) 30 6000 25 5000 20 4000 15 3000 10 2000 5 1000 0 0 -5 -1000 -10 -2000 CTDIvol MEAN UCL LCL DLP MEAN UCL LCL Figure 8: Control chart for Urothelial malignancy data sets (CTDIvol and DLP) 40 2500 2000 30 1500 20 1000 10 500 0 0 -500 -10 -1000 CTDIvol MEAN UCL LCL DLP MEAN UCL LCL Figure 9: Control chart for PE data sets (CTDIvol and DLP) 288 1 1 15 25 29 49 43 73 97 57 121 71 145 85 169 99 193 113 217 127 241 141 265 155 289 169 313 183 337 197 361 1 1 15 27 29 53 43 79 57 105 71 131 85 157 99 183 113 209 127 235 141 261 155 287 169 313 183 339 197 365 University of Ghana http://ugspace.ug.edu.gh APPENDIX IX: DISTANCES COVERED ABOVE AND BELOW UPPER TARGETS Name of Institution....................................................................................................CT MANUFACTURER/ MODEL............................ .............................................................. Patient CVA Head injury Headaches? Couth and Suspicion of ? Abdomino- ? Kidney ? Urothelial ID / trauma SOL/tumour chest pain? lung lesion with pelvic tumour/ stones malignancy PE Lung CKD lesion (CT-IVU) cancer/tumour DAUT DBLT DAUT DBLT DAUT DBLT DAUT DBLT DAUT DBLT DAUT DBLT DAUT DBLT DAUT DBLT DAUT DBLT (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) (cm) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 etc Key: DAUT=distance above upper target, DBLT=distance below lower target. 289 University of Ghana http://ugspace.ug.edu.gh APPENDIX X: TESTING OF NORMALITY Kolmogorov-Smirnova Shapiro-Wilk Statistic df Sig. Statistic df Sig. BrainO .220 39 .060 .832 39 .060 BrainR .185 39 .052 .890 39 .061 EDoriginal .210 39 .060 .900 39 .052 EDreduced .116 39 .200* .925 39 .073 Reduced .116 39 .200* .925 39 .073 ORIGINAL .210 39 .060 .900 39 .052 P.glandO .215 39 .075 .865 39 .070 Eye lensO .217 39 .057 .849 39 .060 EyeballsO .218 39 .076 .845 39 .076 S.glandO .222 39 .089 .924 39 .067 O.cavityO .215 39 .068 .871 39 .067 S.cordO .235 39 .076 .863 39 .068 ThyriodO .255 39 .056 .803 39 .091 p.glandR .179 39 .083 .874 39 .067 Eye lensR .245 39 .087 .875 39 .085 EyeballsR .203 39 .052 .905 39 .067 S.glandR .212 39 .056 .791 39 .067 O.cavityR .323 39 .080 .534 39 .089 S.cordR .308 39 .060 .553 39 .123 ThyroidR .236 39 .000 .762 39 .057 DLP .240 39 .100 .872 39 .089 N.Slices .241 39 .006 .837 39 .097 CTDIvol .224 39 .060 .801 39 .060 kVp .475 39 .070 .522 39 .062 mAs .187 39 .071 .856 39 .067 P .347 39 .080 .768 39 .087 Thickness .369 39 .052 .728 39 .056 SL .145 39 .057 .928 39 .016 DAUT .151 39 .025 .916 39 0.17 DBLT .184 39 .002 .897 39 0.07 BOTHDAUTDBLT .103 39 .200* .940 39 0.08 *. This is a lower bound of the true significance. a. Lilliefors Significance Correction 290 University of Ghana http://ugspace.ug.edu.gh APPENDIX XI: TESTING FOR OUTLIES Grubbs' Test Variable N Mean StDev Min Max G P SNR 100 9.274 3.401 5.437 15.760 1.91 1.000 Wt. 100 72.780 9.138 51.000 90.000 2.38 1.000 Age 100 57.92 15.87 28.00 92.00 2.15 1.000 kVp 100 121.90 3.94 120.00 130.00 2.05 1.000 mAs 100 223.3 102.5 100.0 400.0 1.72 1.000 CTDIvol 100 54.59 18.99 30.27 86.00 1.65 1.000 DLP 100 1028.1 336.7 486.6 1699.6 1.99 1.000 P 100 0.6883 0.1692 0.4000 1.0000 1.84 1.000 N. slices 100 48.44 15.76 29.00 80.00 2.00 1.000 N. slices 100 48.44 15.76 29.00 80.00 2.00 1.000 SL 100 171.87 14.21 135.00 202.32 2.59 0.845 DAUT 100 8.065 7.078 0.000 32.000 3.38 0.052 DBLT 100 20.21 11.33 6.00 50.30 2.66 0.697 DAUT+ABLT 100 28.28 13.55 6.00 60.60 2.39 1.000 Original SL 100 926.7 279.7 486.6 1582.4 2.34 1.000 ED full range 100 2.1315 0.6434 1.1192 3.6395 2.34 1.000 10* 100 715.1 213.3 395.8 1238.4 2.45 1.000 ED reduced 100 1.6447 0.4905 0.9103 2.8483 2.45 1.000 BrainO 100 39.31 13.75 21.27 65.44 1.90 1.000 P. glandO 100 33.27 10.30 17.67 57.03 2.31 1.000 Eye lensO 100 46.09 15.41 25.90 75.93 1.94 1.000 Eye ballsO 100 43.85 14.67 24.58 71.97 1.92 1.000 S. glandO 100 21.297 9.251 5.600 52.810 3.41 0.086 O.cavityO 100 14.420 7.663 3.600 41.930 3.59 0.721 S. cordO 100 2.852 1.581 0.950 9.340 4.10 0.562 ThyroidO 100 1.1141 0.6527 0.4500 3.5500 3.73 0.081 BrainR 100 35.02 11.53 20.09 61.28 2.28 1.000 P. glandR 100 23.224 7.949 1.790 47.970 3.11 0.144 Eye lensR 100 31.33 15.65 3.53 69.08 2.41 1.000 Eye ballsR 100 30.22 13.36 5.03 64.59 2.57 0.901 S. glandR 100 5.314 4.846 1.160 29.060 4.90 0.090 O.cavityR 100 3.513 2.888 1.300 23.080 6.78 0.060 S. cordR 100 0.8362 0.6362 0.3200 5.1400 6.77 0.064 ThyroidR 100 0.4773 0.2837 0.1700 1.9600 5.23 0.067 Oeff.dose brain 100 0.3931 0.1375 0.2127 0.6544 1.90 1.000 Reffe.dose brain 100 0.3502 0.1153 0.2009 0.6128 2.28 1.000 Oeffec.dose Thyroid 100 0.04456 0.02611 0.01800 0.14200 3.73 0.071 Reffec.dose Thyroid 100 0.01909 0.01135 0.00680 0.07840 5.23 0.670 OEffective dosesalivary gland 100 0.21297 0.09251 0.05600 0.52810 3.41 0.076 REffective dosesalivary gland 100 0.05314 0.04846 0.01160 0.29060 4.90 0.700 291 University of Ghana http://ugspace.ug.edu.gh APPENDIX XII: TEST FOR MULTICOLLINEARITY Ist test for multicollinearity Coefficientsa Model Collinearity Statistics Tolerance VIF CTDIvol .474 2.110 P .716 1.396 Slice thickness .211 4.730 1 SNR .179 5.587 SL .558 1.791 kVp .273 3.658 a. Dependent Variable: LensO Final test for multicollinearity after removal of some highly corelated variables Coefficientsa Model Collinearity Statistics Tolerance VIF CTDIvol .566 1.767 1 Slice thickness .484 2.065 SL .559 1.789 a. Dependent Variable: LensO 292 University of Ghana http://ugspace.ug.edu.gh APPENDIX XIII: ETHICAL APPROVAL-UNIVERSITY OF GHANA ETHICS COMMITTEE FOR BASIC AND APPLIED SCIENCES (ECDAS) 293 University of Ghana http://ugspace.ug.edu.gh APPENDIX XIV: ETHICAL APPROVAL-GHANA HEALTH SERVICE ETHICS REVIEW COMMITTEE 294 University of Ghana http://ugspace.ug.edu.gh APPENDIX XV: ETHICAL APPROVAL -THE KORLE BU TEACHING HOSPITAL INSTITUTIONAL REVIEW BOARD (KBTH) 295 University of Ghana http://ugspace.ug.edu.gh 296 University of Ghana http://ugspace.ug.edu.gh APPENDIX XVI: OTHER PERMISSION LETTERS 297 University of Ghana http://ugspace.ug.edu.gh 298 University of Ghana http://ugspace.ug.edu.gh 299 University of Ghana http://ugspace.ug.edu.gh 300 University of Ghana http://ugspace.ug.edu.gh APPENDIX XVII: SUPPORTING DOCUMENT SHOWING AN EXPANDED VERSION OF THE METHODOLOGICAL/CONCEPTUAL FRAMEWORK (FIGURE 3.1) USED FOR CONDUCTING PHASES 1 TO 4 STUDIES. Status of CT Identification of Definition of basic infrastructure indications. eg. diagnostic imaging common ones. requirements of indications -prioritised Selection of facilities -eg. adult, eg. -scan & equipment children coverage -series image quality QC & validation of console displayed Use of -Radiologists dose descriptors -Radiographers /quantities -Protocols Population Fail Pass considerations eg. Weight, 50- 90kg -Application of correction factor if Final inclusion of Development of necessary scanners dose survey > 20-30 or 30-50% collection tool -Exclusion of all Median values Random selection from each facility Collection of 20-50 Selection of Development of dose quantity data patients’ image Indication -based sets (eg. CTDIvol folder- registry and DRL value using -DLP) for each th image quality 75 percentile rule indication per each assessment scanner Statistical data Exclusion criteria control eg. repeats QC: quality control, DRL: diagnostic reference level, CT: computed tomography, CTDIvol: volume weighted computed tomography dose index, DLP: dose length product. 301 University of Ghana http://ugspace.ug.edu.gh APPENDIX XVIII: PUBLICATION LISTS 1. Authors: Botwe, B., Schandorf, C., Inkoom S., Faanu, A. (2020). An investigation into the infrastructure and management of Computerized Tomography units in Ghana. Journal of Medical Imaging and Radiation Sciences, 51(2020), 165-172. 2. Authors: Botwe, B., Schandorf, C., Inkoom S., Faanu, A. (2020). Status of Quality Management Systems in Computed Tomography Facilities in Ghana. Radiologic Technology, 91(4), 324-332. 302 University of Ghana http://ugspace.ug.edu.gh APPENDIX XIX: CONFERENCE AND POSTER PRESENTATION 1. Botwe, B., Schandorf, C., Inkoom S., Faanu, A, Rolstadaas L, Goa, P.E (2019). CT indication-based diagnostic reference levels and dose optimisation steps in Ghana. 10th Pan African Congress of Radiology and Imaging Conference. 14th -16th February 2019. 2. Botwe, B., Schandorf, C., Inkoom S., Faanu, A (2019). An investigation into the infrastructure and management of Computerized Tomography units in Ghana. KBTH Research day. Research Unit Korle Bu Teaching Hospital. 4th December 2019.  3. Botwe, B., Schandorf, C., Inkoom S., Faanu, A, Rolstadaas L, Goa, P.E (2019). Indication-Based Diagnostic Reference Levels: Maiden Values from Ghana. Ghana- Norway Summer School under the theme Quality 3D Imaging and Applications in Radiotherapy Treatment Delivery; 24 - 28 June 2019; KNUST, Kumasi. 4. Botwe, B., Schandorf, C., Inkoom S., Faanu, A, Rolstadaas L, Goa, P.E (2018). Comparison of Indication-Based Diagnostic Reference Level Values: A case of Norway and Ghana. Norwegian University of Science and Technology (NTNU), Physics and Radiography departments. (Student Research Conference, 6th December 2018). 5. Botwe, B., Schandorf, C., Inkoom S., Faanu, A, (2018). Preliminary findings on CT Indication-Based Diagnostic Reference Levels in Ghana. Ghana Norway Summer School under the theme; MRI, Ultrasound and X-Ray Imaging. 25 - 29 June 2018; UDS, Tamale. 303 University of Ghana http://ugspace.ug.edu.gh APPENDIX XX: AWARDS ASSOCIATED WITH THE STUDY The research won the International Society of Radiographers and Radiological Technologists (ISRRT) 2019 Chesney Award (Winner) for the best research likely to improve quality of services in practice in the discipline of medical imaging and/or radiation therapy. 304 University of Ghana http://ugspace.ug.edu.gh The candidate received an exchange studentship at the Norwegian University of Science and Technology as part of the PhD studies and thesis write-up under the NORPART PROJECT. 305 University of Ghana http://ugspace.ug.edu.gh Appendix XXI: COPIES OF PUBLISHED ARTICLES First article link: https://pubmed.ncbi.nlm.nih.gov/32057744/ 306 University of Ghana http://ugspace.ug.edu.gh 307 University of Ghana http://ugspace.ug.edu.gh 308 University of Ghana http://ugspace.ug.edu.gh 309 University of Ghana http://ugspace.ug.edu.gh 310 University of Ghana http://ugspace.ug.edu.gh 311 University of Ghana http://ugspace.ug.edu.gh 312 University of Ghana http://ugspace.ug.edu.gh 313 University of Ghana http://ugspace.ug.edu.gh Second article link: https://pubmed.ncbi.nlm.nih.gov/32102860/ 314 University of Ghana http://ugspace.ug.edu.gh 315 University of Ghana http://ugspace.ug.edu.gh 316 University of Ghana http://ugspace.ug.edu.gh 317 University of Ghana http://ugspace.ug.edu.gh 318 University of Ghana http://ugspace.ug.edu.gh 319 University of Ghana http://ugspace.ug.edu.gh 320 University of Ghana http://ugspace.ug.edu.gh 321 University of Ghana http://ugspace.ug.edu.gh 322