IMAGE QUALITY AND LESION DETECTION IN MAMMOGRAPHY: A COMPARATIVE STUDY BETWEEN FULL-FIELD DIGITAL MAMMOGRAPHY AND COMPUTED RADIOGRAPHY DIGITAL MAMMOGRAPHY BY DESMOND BEDIAKO (10876873) THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN PARTIAL FULFILMENT OF THE REQUIREMENT FOR THE AWARD OF MPHIL IN MEDICAL PHYSICS DEGREE DECEMBER, 2022 University of Ghana http://ugspace.ug.edu.gh i DECLARATION Candidate’s Declaration I hereby declare that this thesis is a sole result of my personal and original research and no aspect of this work has been submitted to the University of Ghana for another degree or presented elsewhere for any other purposes. Candidate’s Signature: …………………………… Date: ……………………… Name: Desmond Bediako Supervisor’s Declaration We declare that the planning, execution and presentation of this thesis were supervised in strict guidance of supervision adhered to by the University of Ghana. Principal Supervisor’s Signature: ………………………… Date: …………………… Name: Dr Edem Kobla Sosu Co - Supervisor’s Signature: ………………………… Date: …………………… Name: Professor Mary Boadu 13-06-2023 13-06-2023 University of Ghana http://ugspace.ug.edu.gh EKSOSU Typewriter 13th June, 2023 ii ABSTRACT Image quality and lesion detection abilities are primary to accurate diagnosis in medical imaging; hence this study was aimed at examining the image quality and lesion detection abilities in Full- Field digital mammography and Computed Radiography digital mammography using the American College of Radiology Mammography Accreditation Phantom (ACR-MAP). Pre- exposure and exposure tests were conducted to establish the effective performance of the mammography systems used. DICOM images were obtained of the ACR-MAP at varying values mAs and kVp. Qualitative image quality assessment was made using the internationally recommended protocol for detection scoring. Quantitative image quality was also estimated using the ImageJ software and the Albert Rose Model to analyze image quality with reference to the Human Health Series numbers 2 and 17 of the International Atomic Energy Agency (HHS - IAEA). Results of the pre-exposure and exposure texts showed optimal and satisfactory performance of four of the systems. The half value layer result of the system D within the CRDM systems was below the recommended limit; hence the poor quality and detection exhibited by the machine of that facility. The obtained signal-to-noise ratio (SNR) and spatial resolution results indicated standard quality images were achievable at the 20 mm and 45 mm thicknesses within both systems but poor-quality images at 70 mm. Signal-to-noise ratio and spatial resolution decreased with increasing PMMA thickness. SNR was 16 % more in FFDM than that of CRDM, whiles the spatial resolution was 0.5 lp/mm, 1.0 lp/mm and 0.5 lp/mm more in in the FFDM systems compared to the CRDM systems within the respective PMMA thicknesses, indicating adequate quality within FFDM. Both FFDM and CRDM systems produced quality images proportional to increasing the detectability as the technique factors (kVp and mAs) increased with the ACR MAP, with the FFDM system’s average percent visibility at 89.05 % and that of CRDM at 75.00 %. The FFDM proved superiority in image quality and lesion detection over the CRDM. University of Ghana http://ugspace.ug.edu.gh iii ACKNOWLEDGEMENTS I am extremely thankful to God Almighty for His enabling strength, provisions and safety granted me throughout the period of my study and the execution of this work. I am sincerely grateful to my supervisors; Dr. Edem K. Sosu and Prof. Mary Boadu, for their continued inputs of directions, guidance and corrections throughout this work. I like to acknowledge the immediate past Head of Department, Prof. Francis Hasford, and the entire staff of the Medical Physics Department most especially Dr. Theresa Dery for their immense contributions and support throughout my study period in the department. My heartfelt appreciation goes to my uncle Dr. Winfred K. Kokor for his financial support on regular and timely basis. To my sweet mother, Madam Rose Kokor, I am deeply grateful to you for your love and constant encouragement and words of counsel, God bless you. I also wish to extend my appreciation to WO1(Rtd) George Bediako and his wife for hosting me throughout the period. To my siblings, Racheal, Kingsley and Godfred, I am utterly amazed by your individual contributions to this journey, I thank you. GOD BLESS YOU ALL University of Ghana http://ugspace.ug.edu.gh iv DEDICATION To my beloved mother, Madam Rose Korkor University of Ghana http://ugspace.ug.edu.gh v TABLE OF CONTENTS DECLARATION ........................................................................................................................... i ABSTRACT ................................................................................................................................... ii ACKNOWLEDGEMENTS ........................................................................................................ iii DEDICATION ............................................................................................................................. iv TABLE OF CONTENTS ............................................................................................................. v LIST OF ACRONYMS ............................................................................................................. xiii CHAPTER ONE ........................................................................................................................... 1 INTRODUCTION ........................................................................................................................ 1 1.1 Background ........................................................................................................................... 1 1.2 Statement of the Research Problem ...................................................................................... 4 1.3 Objectives .............................................................................................................................. 4 1.4 Relevance and Justification ................................................................................................... 5 1.5 Scope and Limitation ............................................................................................................ 6 1.6 Organization of the study ...................................................................................................... 6 CHAPTER TWO .......................................................................................................................... 7 LITERATURE REVIEW ............................................................................................................ 7 2.1 Introduction ........................................................................................................................... 7 2.2 The Female Breast ................................................................................................................. 7 2.3 Breast Cancer ........................................................................................................................ 9 2.4 Symptoms of Breast Cancer ................................................................................................ 10 2.5 Risk Factors/Causes of Breast Cancer ................................................................................ 11 University of Ghana http://ugspace.ug.edu.gh vi 2.5.1 Gender .......................................................................................................................... 11 2.5.2 Age................................................................................................................................ 12 2.5.3 Breast Cancer History ................................................................................................... 12 2.5.4 Race/Ethnicity .............................................................................................................. 12 2.5.5 Height ........................................................................................................................... 13 2.5.6 Denser Breast ................................................................................................................ 13 2.5.7 Early menstrual periods and later menopause .............................................................. 13 2.5.8 Overweight ................................................................................................................... 14 2.5.9 Chest Exposure/radiation .............................................................................................. 14 2.5.10 Childbirth .................................................................................................................... 15 2.5.11 Breastfeeding .............................................................................................................. 15 2.5.12 Physical Activities ...................................................................................................... 15 2.5.13 Alcohol ....................................................................................................................... 15 2.5.14 Family Related Factors ............................................................................................... 16 2.5.15 Environmental Factors ................................................................................................ 16 2.5.16 Genetic Factors ........................................................................................................... 17 2.5.17 Hormonal Factors ....................................................................................................... 17 2.5.18 Ductal Carcinoma in Situ (DCIS) ............................................................................... 18 2.5.19 Benign Breast Conditions ........................................................................................... 18 2.6 Breast Cancer Screening ..................................................................................................... 18 2.7 Mammography .................................................................................................................... 19 2.8 Computed Radiology Digital Mammography ..................................................................... 20 2.9 Full-Field Digital Mammography ....................................................................................... 20 University of Ghana http://ugspace.ug.edu.gh vii 2.10 Image Quality and Lesion detectability ............................................................................ 21 2.11 Breast Compression........................................................................................................... 23 2.12 Spatial Resolution ............................................................................................................. 23 2.13 Noise.................................................................................................................................. 24 2.14 Artefacts ............................................................................................................................ 25 2.15 The Automatic Exposure Control ..................................................................................... 26 2.16 X-ray Spectrum ................................................................................................................. 26 2.17 Signal-to-Noise Ratio (SNR) ............................................................................................ 27 CHAPTER THREE .................................................................................................................... 29 MATERIALS AND METHODS ............................................................................................... 29 3.1 Overview ............................................................................................................................. 29 3.2 Materials .............................................................................................................................. 29 3.2.1 Mammography Machine............................................................................................... 29 3.2.2 Polymethylmethacrylate (PMMA) Slabs ...................................................................... 31 3.2.3 The TORMAS Phantom ............................................................................................... 32 3.2.4 American College of Radiology Mammography Accreditation Phantom (ACR MAP) ............................................................................................................................................... 33 3.2.5 Digital Imaging and Communications in Medicine (DICOM) .................................... 34 3.2.6 Piranha Multimeter ....................................................................................................... 35 3.2.7 Ocean 2014 Software.................................................................................................... 36 3.2.8 Microsoft Excel 2016 ................................................................................................... 37 3.2.9 ImageJ Software ........................................................................................................... 38 University of Ghana http://ugspace.ug.edu.gh viii 3.3 Methodology ....................................................................................................................... 39 3.3.1 Compression Test ......................................................................................................... 40 3.3.2 Compression Force ....................................................................................................... 40 3.3.3 Compression Thickness ................................................................................................ 41 3.3.4 Compression Alignment ............................................................................................... 42 3.3.5 Accuracy and Repeatability of Tube Voltage (kVp) .................................................... 43 3.3.6 Output Linearity and Repeatability and Timer Repeatability ...................................... 44 3.3.7 Tube Voltage (kVp) and Time Linearity ...................................................................... 45 3.3.8 Half Value Layer (HVL) .............................................................................................. 46 3.3.9 Qualitative Image Quality Tests (ACR MAP) ............................................................. 47 3.3.10 Quantitative Image Quality Tests (PMMA and Aluminum Filter) ............................ 50 3.3.11 Spatial Resolution Test ............................................................................................... 51 CHAPTER FOUR ....................................................................................................................... 53 RESULTS AND DISCUSSIONS ............................................................................................... 53 4.1 Overview ............................................................................................................................. 53 4.2 Quality Control or Pre – Exposure Performance Tests ....................................................... 53 4.2.1 Compression Tests ........................................................................................................ 53 4.2.2 Compression Thickness Test ........................................................................................ 54 4.2.3 Compression Alignment Test ....................................................................................... 55 4.2.4 Compression Force Test ............................................................................................... 56 4.2.5 Tube Voltage Accuracy and Repeatability ................................................................... 56 4.2.6 Output Repeatability and Linearity .............................................................................. 57 4.2.7 Half Value Layer test .................................................................................................... 58 4.2.8 Timer Repeatability, Time and Tube Voltage Linearity .............................................. 59 University of Ghana http://ugspace.ug.edu.gh ix 4.3 Qualitative Image Quality and Lesion Detection Assessment ............................................ 60 4.4 Quantitative Image quality Assessment .............................................................................. 67 4.4.1 Detectability and Spatial Resolution ............................................................................ 67 4.4.2 Spatial Resolution at Varying kVp ............................................................................... 72 4.4.3 Signal-to-Noise Ratio Analysis .................................................................................... 75 CHAPTER FIVE ........................................................................................................................ 79 CONCLUSION AND RECOMMENDATIONS ...................................................................... 79 5.1 Pre-Exposure Performance .................................................................................................. 79 5.2 Qualitative Image Quality Assessment ............................................................................... 79 5.3 Quantitative Image Quality Assessment ......................................................................... 7979 5.4 Recommendations ............................................................................................................... 80 5.4.1 Ministry of Health/Ghana Health Service and Hospital Authorities .......................... 800 5.4.2 Medical physicists ........................................................................................................ 81 5.4.3 Radiographers ............................................................................................................. 811 5.4.4 Further Studies ............................................................................................................ 811 REFERENCES .......................................................................................................................... 822 APPENDICES ......................................................................................................................... 1011 APPENDIX A ............................................................................................................................ 101 Ethical Clearance from ECBAS .............................................................................................. 101 APPENDIX B ............................................................................................................................ 102 Ethical Clearance from 37 Military Hospital .......................................................................... 102 APPENDIX C ............................................................................................................................ 103 Data for estimating Compression Thickness ........................................................................... 103 University of Ghana http://ugspace.ug.edu.gh x APPENDIX D ............................................................................................................................ 103 Data for estimating Compression Force .................................................................................. 103 APPENDIX E ............................................................................................................................ 104 Data for estimating Compression Alignment .......................................................................... 104 APPENDIX F ............................................................................................................................ 105 Data for estimating KVP Accuracy and Repeatability............................................................ 105 APPENDIX G ............................................................................................................................ 106 Data for estimating the Output Repeatability and Linearity ................................................... 106 APPENDIX H ............................................................................................................................ 107 Data for estimating Half Value Layer ..................................................................................... 107 APPENDIX I ............................................................................................................................. 108 Data for estimating the Timer and KVP Linearity .................................................................. 108 APPENDIX J ............................................................................................................................. 109 Data for estimating the Timer Repeatability ........................................................................... 109 APPENDIX K ............................................................................................................................ 110 Data for estimating Lesion Detection in ACR-MAP .............................................................. 110 APPENDIX L ............................................................................................................................ 111 Data for estimating the Signal-to-Noise-Ratio ........................................................................ 111 University of Ghana http://ugspace.ug.edu.gh xi LIST OF TABLES Table 3.1: Particulars of mammography machines utilized ......................................................... 30 Table 4.1: Results of the compression thickness test for all five (5) systems .............................. 54 Table 4.2: Results of the compression alignment test for all five (5) systems ............................. 55 Table 4.3: Results of the compression force test for all five (5) systems ..................................... 56 Table 4.4: Results of the tube voltage accuracy and repeatability for the systems ...................... 57 Table 4.5: Results of the output repeatability and linearity test for the mammography systems . 58 Table 4.6: Results of HVL tests for the mammography systems ................................................. 59 Table 4.7: Results of other QC tests on the mammography systems ........................................... 60 Table 4.8: Results of FFDM and CRDM fiber detections ............................................................ 62 Table 4.9: Results of FFDM and CRDM specks detections ......................................................... 63 Table 4.10: Mass detection within the FFDM and CRDM systems ............................................. 64 Table 4.11: Overall summary score of both the FFDM and CRDM systems .............................. 66 Table 4.12: Results of the spatial resolution test at 29 kVp and 20 mm ...................................... 69 Table 4.13: Results of the spatial resolution test at 29 kVp and 45 mm ...................................... 70 Table 4.14: Results of the spatial resolution test at 29 kVp and 70 mm ...................................... 71 Table 4.15: Results of spatial resolution at 50 mAs and 20 mm. ................................................. 73 Table 4.16: Results of spatial resolution at 50 mAs and 45 mm. ................................................. 74 Table 4.17: Results of spatial resolution at 50 mAs and 70 mm. ................................................. 75 Table 4.18: Calculated SNR for the five (5) systems used ........................................................... 76 University of Ghana http://ugspace.ug.edu.gh xii LIST OF FIGURES Figure 2.1. Anatomy of the breast (Johns Hopkins Medicine 2019) .............................................. 8 Figure 3.1 Image of a Mammography machine ............................................................................ 31 Figure 3.2. Image of the PMMA slab ........................................................................................... 32 Figure 3.3: The TORMAS Phantom ............................................................................................. 33 Figure 3.4: The American College of Radiologist Mammography Accreditation Phantom (ACR MAP)............................................................................................................................................. 34 Figure 3.5: The Piranha multimeter .............................................................................................. 36 Figure 3.6: Interface of the Ocean 2014 software ........................................................................ 37 Figure 3.7: Interface of Microsoft Excel 2016 ............................................................................. 38 Figure 3.8: The interface of the ImageJ software ......................................................................... 39 Figure 3.9: Set-up for compression force ..................................................................................... 41 Figure 3.10: Set-up for compression thickness and alignment ..................................................... 42 Figure 3.11: Set-up for the kVp Accuracy and Repeatability ....................................................... 44 Figure 3.12: Image quality set-up with the ACR MAP ................................................................ 49 Figure 3.13: Set-up of the PMMA and the sandwiched aluminum filter ..................................... 51 Figure 3.14: Set – up for spatial resolution with TORMAS phantom .......................................... 52 Figure 4.1: Image of the ACR MAP ............................................................................................. 61 Figure 4.2: Graph of the percentage visibility against varying kVp and mAs ............................. 67 Figure 4.3: An image of the TORMAS phantom ......................................................................... 72 Figure 4.4: ROI's drawn in the PMMA phantom sandwiched with an Aluminum filter .............. 78 University of Ghana http://ugspace.ug.edu.gh xiii LIST OF ACRONYMS ACR-MAP - American College of Radiology Mammography Accreditation Phantom AEC - Automatic Exposure Control CC – Cranio-Caudal COV - Coefficient of Variation CR - Computed radiology CRDM - Computed Radiology Digital Mammography CT - Computed Tomography DCIS - Ductal Carcinoma In-Situ DICOM - Digital Imaging and Communication in Medicine DQE - Detective Quantum Efficiency FFDM - Full-Field Digital Mammography FITS – Flexible Image Transport System GIF – Graphics Interchange Format HHS – Human Health Series HVL - Half Value Layer IAEA – International Atomic Energy Agency ICRP – International Commission on Radiological Protection JPEG – Joint Photographic Expects Group University of Ghana http://ugspace.ug.edu.gh xiv kVp - X-ray Tube Voltage mAs - X-ray Tube Output MLO - Medial-Lateral Oblique Mo/Mo – Molybdenum/Molybdenum Mo/Rh - Molybdenum/Rhodium MRI - Magnetic Resonance Imaging MTF - Modulation Transfer Function NEMA - National Electrical Manufacturers Association PACS - Picture Archiving and Communication System PMMA – Polymethyl Methacrylate PSP - Photostimulable Phosphor QC - Quality control Rh/Al - Rhodium/Aluminum Rh/Rh - Rhodium/Rhodium ROI - Region of Interest SNR – Signal-to-Noise Ratio TIFF – Tag Image File Format USB - Universal Serial Bus W/Rh - Tungsten/Rhodium University of Ghana http://ugspace.ug.edu.gh 1 CHAPTER ONE INTRODUCTION 1.1 Background Breast cancer is a tumor often originating from the glandular tissues in the female breast. The disease is fatal, and the most frequently diagnosed disease or cancer prevalent in women, and the commonest cancer that is associated with mortality among women (DeSantis et al., 2017). In Ghanaian women, the disease is a very familiar malignant disease resulting in the bulk of deaths related to cancer (Wiredu & Armah, 2006). Unfortunately, reports or data on the disease in Ghana are inadequate (Biritwum & Amaning, 2000). In Ghana and in the sub-Saharan African region, the generality of the disease among black African women is often acute with critical features (Wiredu & Armah, 2006) due to the late presentation to the health care professionals and inadequate screening and therapeutic interventions. The disease has over the years been of concern to stakeholders and a public health challenge whose prevention remains a major problem because the cause is still unclear to researchers (Laprovitera et al., 2021). At the early stage of breast cancer, there are usually no symptoms and is most often noticed or diagnosed through breast self- examination, clinical breast examination and mammography screening and other imaging modalities. When the disease gets to a more severe stage, then the symptoms begin to manifest. The commonest symptom is a pain-free lump in the breast. Other indicators of the disease includes but are not limited to the following: continuous variations in the breast, such as stiffening, enlargement, and nipple deformity like intermittent discharges or retractions (ACR, 2019). In the beginning stages of breast cancer development, the infected cells separate themselves from each other extending to other regions of the breast implicating dire complications if not detected and University of Ghana http://ugspace.ug.edu.gh 2 managed at these early stages. However, the accurate diagnosis of breast cancer at an early stage can increase the possibility of treatment and total recovery and reduce the probability of recurrence and mortality rate. In recent times, there is evidence of decrease in breast cancer death rate in some developed countries and this milestone is attributed to early detection through screening schemes and timely and potent diagnosis and treatment protocols (Weir et al., 2003). This same narrative may not apply to Ghana and the sub-Saharan African region due to the inadequacy of these screening schemes and quality cancer care (Mensah et al., 2016). Mammography is a radiographic (X-ray) procedure optimized for the examination of the breast; it has a very high ability to detect early-stage breast cancer (lesions). Mammography has become an extremely accepted and useful non-invasive imaging modality with unmatched capabilities of lesion and microcalcification detection in the breast (ACR, 2019). A predominant target of screening mammography is its ability to detect the radiological masses (lesions) in the breast at a fairly early-stage such that ensuring treatment will reduce the possibility of death (Yaffe et al., 2013). However, just like any measuring or screening tool, mammography is not 100% perfect (ACS, 2020), it is hence highly expected that the mammography system exhibits excellent responsiveness to lesions in order are not to overlook lesions (false negative). But to achieve this feat, equipment performance and exposure parameters (technique factors) must be selected with the quality of the image and the expected detection task as priorities and ensuring that the patient’s absorbed dose is kept as low as reasonably achievable (ACR, 2019). The screening of breast cancer using X-ray mammography is highly beneficial if the modality can detect breast cancers (lesions) at an early stage. This can be made possible with high-quality breast images (mammograms) to ensure the possible detection of the smallest lesions in the breast (Yaffe & Mainprize, 2011). Ultimately, the University of Ghana http://ugspace.ug.edu.gh 3 most relevant aim regarding mammography systems is the unwavering ability of providing contrast amidst a lesion possibly occupying a region in the breast and the uninfected tissues in the same region in the breast (Chevalier et al., 2012). The intentional screening mammography programs continuously contribute immensely to guarding women within certain ages and who are uncertain due to their respective family history to the disease. Thus, the quality of mammograms is given equal importance if not more since the ability to detect lesion is directly proportional to the quality of mammograms (Yaffe & Mainprize, 2011). Mammography image quality emphasizes the system’s ability in visualizing the anatomy of the breast sufficiently, mainly depicting the internal structures of the breast (Thevi et al., 2012). A primary goal of quality control and optimization in mammography is to better the detection of microcalcifications and lesions originating from and residing in breasts. In a quantitative approach, the image quality can be measured using parameters such as spatial resolution, artefacts, noise and signal-to-noise ratio. The criteria used in diagnostic radiography for image quality in mammography addresses the standard required for image quality (Samei et al., 2005). Image quality analysis is done with reference to only a particular imaging mandate. Notwithstanding, the main image quality requirements most important in radiography are image contrast, noise, sharpness; these features are equally relevant in mammography. Considering the evolutions in technology and imaging of the breast, a matching level of image quality is required with respect to the necessary parameters. This study will specifically focus on the ability of the mammography system in detecting masses in the breast with similar optical densities with reference to high signal- to-noise ratio which are commonly defined for mammography equipment with variation, depending on the manufacturer and the system specifications (Thevi et al., 2012). University of Ghana http://ugspace.ug.edu.gh 4 1.2 Statement of the Research Problem Lesions and microcalcifications are the most important finding in asymptomatic patients with early breast cancer. The early detection of cancer using mammography highly reduces the risk of death and increases treatment options. Breast cancer detection in mammography is often challenged by the difficulty to differentiate between breast tissues and other pathological findings which appear to have almost the same or close linear attenuation in the energy range used in mammography. Moreover, the seeming imperfections in mammography may result in missing the detection of cancer (false negative). The rise in false-positive and high false-negative detection rates in mammography can largely be linked to the poor quality of mammograms leading to the difficulty in the differentiation of lesions and the anatomical background. In Ghana, an estimated 10% of positive breast cancers cases missed in mammography examination (Dzidzornu et al., 2021) basically due to poor image quality. The widely accepted confidence in digital systems to produce maximum quality images is unmatched, yet there exists the potential of non-detection of breast cancer even by the virtue of digital imaging modalities. Important parameters that determine image quality and lesion detectability in mammography are noise, (spatial fluctuation), artefacts and signal-to-noise. These parameters help improve diagnostic accuracy while reducing accumulative radiation exposure to unsuspecting women. This work will assess the major parameters used in mammography and their respective effects on image quality and lesion detection abilities. 1.3 Objectives The primary aim of this study was to examine the image quality and lesion detection in the Full- Field Digital Mammography (FFDM) and Computed Radiology Digital Mammography (CRDM). University of Ghana http://ugspace.ug.edu.gh 5 The specific objectives of the study were to: 1. compare the pre-exposure and exposure performance between the two modalities; 2. assess the qualitative image quality at varying exposures between the two modalities; 3. assess the quantitative image quality at varying exposures between the two modalities. 1.4 Relevance and Justification A major aim of mammography is the ability to commence the treatment as early as the lesions are detected hence reducing the rate of breast cancer deaths that may occur as a result late detection or late start of treatment of the disease at far-advanced stages (Yaffe et al., 2013). The need to optimize the exposure factors on the mammography system is highly necessary for the enhancement of image quality, noting the relevance of higher contrast techniques and their effect on the detection of a shallow contrast lesions. The need to determine how well modern mammography machines can distinguish between possible tiny lesions and normal tissues requires maximum confidence in the quality of images produced by digital systems. A small number of studies also acknowledges the impact of high noise on the detection of infections residing in soft tissue with reference to minimized dose to the patient hence keeping exposures as low as reasonably achievable. The ICRP report number 103 (ACR, 2019) recommendation considers the breast as a radiosensitive and vulnerable organ with the breast glandular tissue having a tissue- weighting factor of 0.12, making it important for quality control routines to be consistent in enhancing the adequate performance of the digital systems. It is also important to adhere to other conditions/parameters necessary for maximum performance in digital mammography. The relative decrease in the number of death of women linked to breast cancer due to reliable screening protocols is closely merited to the ability to detect microcalcifications or lesions early. It is however crucial to quantify the detection rate of small masses and lesions in full-field digital University of Ghana http://ugspace.ug.edu.gh 6 mammography system and compare its abilities to that of computed radiology digital mammography, hence substantiating the superior imaging modality or technique for scheduled screening programs for breast cancer. The enthusiasm to reduce the occurrence of high false negative and high false positive mammograms is an important justification for the comparative study between the traditional computed radiology digital mammography and the full-field digital mammography systems. 1.5 Scope and Limitation This study was phantom-based, conducted in five mammography centers within the Greater Accra region, Ghana. 1.6 Organization of the study This investigative work is reported in chronological order of five chapters. Chapter one gives a broad overview to the study topic, the chapter also introduces the statement of the research problem and description of the main problems addressed in the study. Chapter two reviews the existing literature and relevant research associated with the problem of this study. Chapter three presents and focuses on the materials and methods employed for the collection of relevant data and their respective analysis to ensure successful image quality and lesion detection assessment. The data analysis and its results presentation are indicated in chapter four, whereas chapter five offers the needed summary and related discussion of findings, their implications to everyday practice or clinical implementation, the conclusion, recommendations and suggestions for future research. University of Ghana http://ugspace.ug.edu.gh 7 CHAPTER TWO LITERATURE REVIEW 2.1 Introduction Mammography is an examination of the breast using X-rays of relatively low energies. Using mammography for screening aids in the ease of detecting microcalcifications and lesions. In this section, the relevant literature on the subject is reviewed. These include the general knowledge of the female breast, breast cancer, mammography, Computed Radiology Digital Mammography (CRDM), Full-Field Digital Mammography (FFDM), and the general comparisons of image quality and lesion detection abilities between CRDM and FFDM. 2.2 The Female Breast The breast (Figure 2.1), is such a crucial organ in the human body and contributes immensely to the growth and development of an infant (Mother and Child, 2019). It is seated on an area of the chest wall of the human body. Women as well as men have breasts, but the breast is most often well developed and more useful in women/females. The breast is a gland mostly made up of lobes positioned all over the nipple. The breast produces milk for providing food and nutrition to the baby (Boquien, 2018). Understanding the healthy anatomy and function of the female breasts will help the ease of detecting abnormalities. The tissues of the human/female breast start to develop as early as the fifth to the sixth week of expected infant stages (Bland & Copeland, 1993; Hopkin’s, 2020). Often, the aggregate of fatty tissues in the breast determines its size (Mayor Clinic 2020; WMMC 2018). Hence, the size and shape of women’s breasts may vary as a result of genetics, diet and ethnicity, and the age of the woman (Pierce et al., 2002). The majority of the volume in the breast is confined to the upper surface part which is often compromised during breast cancer University of Ghana http://ugspace.ug.edu.gh 8 (Adesunkanmi et al., 2006; Lambe et al., 1994), making that region the common region of tumors in the breast. Though the structure of the male breast seems identical to the female’s, the male breast does not possess the needed or special structures for milk production requirements. The breast is made up of special tissues for a specific physiological purpose as milk production, combination, and the discharge of milk (Hassiotou & Geddes, 2013). These functions are performed by the network of hormones and other relevant factors that control the production of milk. Within the four segments of the breast are 15 to 20 lobes specifically for the production of milk, these lobes are opened to the nipple (Pandya & Moore, 2011; Zucca-Matthes et al., 2016). The smaller structures within the lobes are the lobules. Figure 2.1. Anatomy of the breast (Johns Hopkins Medicine 2019) Just like other organs in the various systems of humans, the breast and its connecting systems change gradually from birth to adolescence, the menstrual cycle, through pregnancy and lactation. University of Ghana http://ugspace.ug.edu.gh 9 These changes in the breast are so distinctive since no other organ in the body changes as much in function, shape and size. It is however very crucial to understand the anatomy of the breast and its conduct and response to radiation and imaging modalities. 2.3 Breast Cancer Breast cancer is a heterogeneous disease caused by the amassing of genetic anomalies, such as regional alterations, displacements/dislocations, and reprinting (Sabel, 2009). Simply put, when cells that make up the breast grow abnormally forming lumps. Breast cancer originates from such abnormal breast cells. Usually, these defects are formed in the ducts which transport produced milk to the nipple. Breast cancer occurs not only in women but also in men, although breast cancer in men is sparse (Anderson et al., 2010; Gethins, 2012; Kreiter et al., 2014). Its impact on the health of women on the other hand is dire, making the disease a key public health challenge for women all over the world (Ghoncheh et al., 2016), resulting in frequent/high deaths among women (Ferlay et al., 2015; Fidler et al., 2017; Wiredu & Armah, 2006). At the pre-malignant lesion or non-invasive stage of the disease, it is contained in the ducts without spreading to other breast tissues that are healthy whiles invasive breast cancer has already outspread beyond or outside the ducts to other good breast tissues, alternatively beyond the breast to lymph nodes or other organs. Breast cancers, just like any other malignancy, are known to spread through the continuous pilling up of several changes in the genetic track controlling enlargement, death, gene repair, and rapid responses to therapy in the neural region (Campbell & Polyak, 2007; Pelengaris et al., 2002). Since the mortality reached its peak value in 1989, the breast cancer death rate in women had declined by some 41% in 2018 due to earlier detection through screening and increased awareness and education on symptoms and the advancement in treatment, bringing the pace of decline to as low as 1% per year (2013 to 2018) compared to the nearly 2% during the 2000s (ACS, 2022). University of Ghana http://ugspace.ug.edu.gh 10 Worldwide cancer statistics in recent times indicate a rising incidence of breast cancer and the increase among African developing countries including Ghana is at a faster rate than the low incidence rate recorded some years back (Ohene-Yeboah & Adjei, 2012). Some studies have shown the late state at which the patients they seek healthcare whiles cancer may already be at an advanced stage not belittling the menace by poor or lack of diagnostic expertise in the high incidence rate in these developing countries like Ghana. In Ghana, data on the disease is scanty but it is the particular disease with the most incidence in Ghana (Wiredu & Armah, 2006). It is however worrying that the mean age at the stage of breast cancer diagnosis amongst Ghanaian women is 46 years ranging from 26 to 80 years (Ohene-Yeboah & Adjei, 2012) relative to that of over 62 years in other parts of the world (ACS, 2022). The early diagnosis of the disease can increase the rate or chance of early detection and improve survival rates (WHO, 2004). Mammography is however the known most efficient screening tool for detecting breast cancer as early as possible. 2.4 Symptoms of Breast Cancer Though there is a slight decline in breast cancer incidence worldwide as a result of major factors including early detection (Parkin & Fernández, 2006), it is highly advisable to visit your healthcare provider as early as possible if you experience any changes in your breast or on condition of any of the below-listed symptoms. It is worth noting that these symptoms could reflect other health states though they may be that of breast cancer. Signs such as swelling under the armpit or around the collarbone may be a symptom of a lymph node lump. However, you should discuss any concerns with your doctor. Breast cancer symptoms may largely involve but are not limited to the following: a lump or mass in the breast, changing breast dimensions, rash on the nipple, hollowing of the skin, swelling in the armpit, discharge from the nipple, consistent aches in the breast (CDC University of Ghana http://ugspace.ug.edu.gh 11 Breast Cancer, 2022) and so on. Nonetheless, the symptoms may sometimes differ and some patients may not even exhibit any signs, hence the need for regular screening. 2.5 Risk Factors/Causes of Breast Cancer To date, scientists and researchers have not proven a particular cause of breast cancer (Delen et al., 2005). However, multiple factors seem to be linked with a higher risk of developing the disease. It is important to note that as common to almost every form of the disease, exhibiting risk factors does not necessarily mean you have or would have the disease and on the other hand, not also showing these risk factors does not mean one will not get the disease. Most of such factors bear a considerable surge in risk for the woman (Trickey et al., 2012). Old age and sex are the only risk factors for women who make up about 50% of those with the disease, according to literature (Madigan et al., 1995; Renganathan et al., 2014). Over the years, the control and reduction in many of the most delegated risk factors of the disease have not and cannot be controlled by any human intervention (Sosu, 2018). These risk factors include but are not limited to the following. 2.5.1 Gender Since breast cancer is predominantly woman-related cancer, being born a woman is a major risk factor for breast cancer (ACS, 2022) since the woman's breast undergoes a series of changes and contains more hormones (Armstrong et al., 2000). Men also do get the disease as a result of increased levels of estrogen and androgen, but the disease is more common in women relative to men (Giordano et al., 2002). University of Ghana http://ugspace.ug.edu.gh 12 2.5.2 Age The risk of most cancers increases with the increase in age, thus beyond sex, age is a crucial risk factor for breast cancer. This is due to the relationship between the increasing rate of breast cancer amongst the slightly aged (ACS, 2022) since breast cancer is extremely uncommon in women below 30 years. Over 80% of breast cancer-related deaths in other parts of the world were recorded in females above the ages of 40 and 60 years (Coughlin, 2019). 2.5.3 Breast Cancer History A woman who is diagnosed with breast cancer in one breast has a higher risk and an increased likelihood of developing an all-new infection of the disease in the supplementary breast or other regions of the same breast (Breastcancer.org, 2022). This may differ from regenerated cancer, the risk associated with this is so low in general but maybe so high in younger women with or treated with cancer. 2.5.4 Race/Ethnicity Though the difference is quickly closing up in recent times, there is however still existing disparities indicating that light skin women develop breast cancer at a higher rate compared to black women. Yet black women are also more prone to die from the disease at random ages due to the prevalence of the triple-negative form of the disease in black women, a more severe form relative to the other forms (Breastcancer.org 2022). It is long confirmed that women except Africans and Asians are seemingly more vulnerable to developing cancer of the breast yet African/black women are more exposed to developing other aggressive and slightly advanced forms of breast cancer likely diagnosed at young ages (Hirko et al., 2022). University of Ghana http://ugspace.ug.edu.gh 13 2.5.5 Height Various existing studies show that women who are highly endowed with height (taller) are at higher risk of breast cancer than short women (Kabat et al., 2015; Willett et al., 2014), basically as a result of nutrition in the early stages of life, hormonal or genetic factors. This risk factor can also be related to the growth shoot tall women may have had in their childhood through adolescence from higher levels of hormones and other growth indicators (van den Brandt et al., 2021). 2.5.6 Denser Breast The breast is composed of fat tissues, fibrous tissues, and glandular tissues. If the breast appears denser on a mammogram, this means the breast has more glandular and fibrous tissue but fewer fat tissues (CDC Breast Cancer, 2022). It is normal to have dense breasts, but in dense breast tissues, it can very difficult to detect or see lesions/cancers on a mammogram. Women who have dense breasts on a mammogram are at high risk of breast cancer compared to women with average breast density though it is not clear if lowering the density of the breast decreases the risk of the disease (Pandya & Moore, 2011; WMMC, 2018). 2.5.7 Early menstrual periods and later menopause Starting menstrual periods so early increases the woman’s risk of developing the disease. In effect, these women turn to have more menstrual cycles compared to other women because they started menstruating so early and they are slightly at higher risk of breast cancer than the others (Eaton, 2002). This increase in the risk can be attributed to a longer period/lasting exposure to the estrogen and progesterone hormones. When the onset of a woman’s menstrual cycle starts earlier than it is University of Ghana http://ugspace.ug.edu.gh 14 supposed to, this results also in the extension of menopause thus cancer risk of those women who go through menopause at older ages increases (Breastcancer.org, 2022). 2.5.8 Overweight Women who are overweight and obese are at higher risk of breast cancer compared to women who maintain a healthy weight, most importantly after menopause (Picon‐Ruiz et al., 2017). Overweight/obese expressed in terms of the individual's body mass index concerning many studies show that being overweight increases the risk of breast cancer redeveloping in women who have been diagnosed with the disease some time passed due to the continuous production of estrogen. Extra fatty cells replicate more estrogen in the body, and estrogen can make hormone receptor- positive breast cancers develop and grow (Breastcancer.org, 2022). Also, insulin levels in obese women are higher, another hormone which has been associated with some cancers, especially breast cancer (van den Brandt et al., 2021; CDCBreastCancer 2022). 2.5.9 Chest Exposure/radiation Women who encounter radiations to the chest area either for diagnosis or treatment for other diseases or cancers at younger ages have a notably higher risk for breast cancer (Moskowitz et al., 2014). This is highly dependent on the woman’s age at exposure to the radiation, riskier at teen ages or young adult ages where the breast is yet still developing lowers after 40 years (Yaffe & Mainprize, 2011). University of Ghana http://ugspace.ug.edu.gh 15 2.5.10 Childbirth Women who have not been able to keep or complete a pregnancy term or whose first child comes after their 30th birthday tends to be at higher risk of breast cancer (Zhang et al., 2012) compared to women who have had a full pregnancy term before age 30. Significantly, the birth of a child lowers a woman’s risk of breast cancer (Eaton, 2002). It is however proven in the literature that the risk of breast cancer increases exponentially in the years just after childhood and this trend, as a result, induces a growth enhancement effect on cells by hormonal variations during pregnancy (Wohlfahrt, 2001). 2.5.11 Breastfeeding Breastfeeding is largely beneficial to the infant, by reducing many childhood infections (Mother and Child, 2019). Breastfeeding is not just beneficial to the baby but also to the mother by reducing breast cancer risk and other health benefits like reducing the risk of ovarian cancers (Cramer, 2012). Literature linking breastfeeding to breast cancer is expanding. Breastfeeding for at least 12 months is advisable for all women since it is a modifiable risk factor (Breastcancer.org, 2022). 2.5.12 Physical Activities Being physically active is of great importance to the body. The evidence is increasingly clear that regular physical activity helps to reduce breast cancer risk, mostly after menopause (Breastcancer.org, 2022). This has a massive effect on the body’s weight, and hormone levels. 2.5.13 Alcohol People who are very heavy drinkers, i.e. tend to have improper meals without fruits and vegetables (Breastcancer.org, 2022). Evidence from the literature shows that alcohol increases the risk for University of Ghana http://ugspace.ug.edu.gh 16 breast cancer (Coronado et al., 2011) and a woman’s risk of the disease slightly increases with the increased intake of alcohol since alcohol contains a crucial compound with dire effects. The level of risk is directly proportional to the amount of alcohol the woman drinks relative to women who do not drink any alcohol at all (Breastcancer.org, 2022). Alcohol drinking and tobacco smoking show an interaction in the aetiology of several cancers (ACS, 2022). 2.5.14 Family Related Factors Amongst all cancers, breast cancer is highly likened to the prevailing family trends, thus breast cancer risk in a particular family is directly proportional to its prevalence in the family (Dumitrescu & Cotarla, 2005). Notwithstanding, the risk of breast cancer for a woman whose close relative such as the mother or sister has had the disease at an old age and without any other family member developing the disease may be comparatively and largely low, whiles a woman is at a higher risk if she has multiple of her family members diagnosed of breast cancer at younger ages (Liu et al., 2021). It is however evident that near one-quarter of the incidence of the disease is related to its existence in the family (Bistoni & Farhadi, 2015). 2.5.15 Environmental Factors The unavoidable exposure to ionizing radiation, whether through nuclear accidents/explosions or medical, be it for diagnostic/therapeutic purposes, highly increases the risk of breast cancer (Golubicic et al., 2008). The breast is radiosensitive (Mehnati et al., 2019) and the lengthy delay for breast cancers caused by exposure to radiation, in addition to the high response to mutation impairment in developing breasts, after forty years, radiation exposure produces a negligible increase in cancer risk, but on the other hand, early life exposures is a detrimental risk (Ronckers et al., 2004). Supplementary environmental factors, including the vulnerability to electromagnetic University of Ghana http://ugspace.ug.edu.gh 17 fields and some organochlorine pesticides, are seen to contribute to the increased risk of breast cancer, but more data is needed before drawing firm conclusions. 2.5.16 Genetic Factors The probability of inheriting breast cancer is partially associated with the gene defect of genes of the breast, like BRCA1 and BRCA2 (Fackenthal & Olopade, 2007). The early identification of these genes (BRCA1 and BRCA2) provides new insights and helps in the prevention of the disease. When the disease is diagnosed at a young age or when multiple relatives are diagnosed with the disease or when there exists enough history of some other cancers in the family, then the woman is exposed to a large risk of breast cancer. Also, when a woman experiences a gene defect in any of the genes (BRCA1 and BRCA2) or when a man experiences gene defects in BRCA2 then they may develop breast cancer with a lifetime risk (Gethins, 2012; Giordano et al., 2002). 2.5.17 Hormonal Factors Many studies have identified that in most women, the abnormal growth of cells in the breast is related to the reproductive hormones women possess and their continuous exposure to several estrogens (Breastcancer.org, 2022). During pregnancy, the breast cells experience diverse differentiation resulting in a longer period for DNA repair (Albrektsen et al., 1995). Though available data suggest that prolonged lactation reduces the risk of breast cancer, abortion by the use of contraceptives on the other hand has a dire risk of breast cancer, whether spontaneous or induced, escalates the risk of breast cancer (Mørch et al., 2017). The termination of a pregnancy may also terminate/stop several growth changes in the breast and hence increasing the breast cancer possibilities. At this stage, very saturated estrogen levels of early pregnancy are released to and by the breast. University of Ghana http://ugspace.ug.edu.gh 18 2.5.18 Ductal Carcinoma in Situ (DCIS) The widespread screening with mammography has improved the detection rate of DCIS significantly (Groen et al., 2017), whiles the breast-conserving therapy’s use for the treatment of invasive carcinoma has led to the enhancement of the management of women with DCIS. DCIS is an entity distinct in both its clinical presentation and its biological potential from the other lesion classified as noninvasive carcinoma (Pinder, 2010). 2.5.19 Benign Breast Conditions Women who have some existing breast conditions are likely to have an increased probability of breast cancer risk (Breastcancer.org, 2022). Often, several of these conditions appear to be more connected to the disease at risk than others. If a woman has a trace of breast cancer and hyperplasia or atypical hyperplasia in her family, she is highly prone to breast cancer (Page et al., 2003). 2.6 Breast Cancer Screening Though diagnostic mammography and screening mammography differ, they are both used for breast cancer detection. Patients exhibiting symptoms of breast cancer are recommended for diagnostic mammography whiles women without symptoms undergo breast screening (Jatoi, 1999). Screening mammography is purposefully done to find cancers at a so much early stage without it not spreading to other parts to reduce breast cancer mortality by using better treatment options just as the same applies to all kinds of diseases. Mammography has over the years proven efficient amongst other modalities such as clinical breast examination, breast self-examination, screening ultrasonography, magnetic resonance imaging and breast tomosynthesis (ACS, 2022). Mammography screening options are available to all women, especially for women between the ages of 40 and 55, such women can schedule an individual screening plan or may have themselves University of Ghana http://ugspace.ug.edu.gh 19 screened each year or every other year until they are convinced of a healthy breast (Monticciolo et al., 2017; Oeffinger et al., 2015). 2.7 Mammography Mammography is a radiographic (X-ray) procedure optimized (low dose) solely for the examination of the breast with the abilities of early breast cancer detection and improving survival rate (Gardner, 2006; Schleede et al., 2014). Whether screen-film or digital, mammography is an effective and widely used examination in medical imaging for the breast with emphasis on low patient motion, high contrast, low noise images, and other important image assessment conditions. Mammograms or mammography is not perfect whether being used as a screening or diagnostic tool, some breast cancers or lesions (micro-classifications) may go unidentified (Dzidzornu et al., 2021). If lesions are detected, further examinations may be conducted to understand their extent. A normal mammogram may consist of either medial-lateral oblique (MLO) or craniocaudal (CC) views. For adequate image quality and maximum clarity in viewing mammograms, correct breast positioning and breast compression are highly important (Meaney et al., 2000). The quality of mammographic images is extremely crucial and essential for the maximum assurance that the entire imaging system of various techniques is working optimally hence the need for regular mammography quality control tests. During a mammography examination, the breast is adequately compressed to evenly spread out the thickness of the breast hence increasing image quality. Compression also, among other things, reduces the thickness of the breast tissue for X-rays to penetrate unperturbed and reduces scatter radiation which hitherto may increase the amount of required radiation dose and reduce image quality (Wikipedia, 2022). Until recently, screen-film systems were largely used in mammography. But for the demands for higher spatial resolution and University of Ghana http://ugspace.ug.edu.gh 20 other factors, digital mammography or Full-Field Digital Mammography (FFDM) is gradually taking over (Lewin et al., 2002). 2.8 Computed Radiology Digital Mammography Computed radiology (CR) digital mammography is similar in technology to Full-field digital mammography (FFDM) except for the image receptor. CR employs the photostimulable phosphor (PSP) plates which must be sent to a digital scanner (or reader) and read using a laser beam to extract the stored image to a digital display (Yaffe, 2010). This imaging system also relies on picture archiving and communication systems (PACS) for the transfer and storage of digital images. CR systems also use a PSP plate to store images emerging from the interaction of x-rays with the phosphor. This stored image is then stimulated by a laser beam with a very small focus (Carter & Veale, 2022). This stimulation results in the emission of light which is captured and digitized to produce the digital array. CR is unreliable in terms of image quality for patient dose ranges due to the initial storage, and the stimulation and readout of the image information but highly efficient for high spatial resolution by employing the use of thinner phosphor materials with a finer sampling pitch (Pongnapang, 2005). 2.9 Full-Field Digital Mammography Over the past twenty years and so, digital mammography has become more relevant than just a simple innovation to becoming the most useful imaging tool for early breast cancer detection (Lewin et al., 2002). Full-Field Digital Radiography (FFDM) is a type of digital mammography that uses x-ray very sensitive plates to rightly gather data during a breast examination and quickly and easily transfers same to a programmed system without using an intermediate cassette (Marchiori, 2004; van den Brandt et al., 2021). The advantages of FFDM University of Ghana http://ugspace.ug.edu.gh https://en.wikipedia.org/wiki/Digital_mammography 21 include but are not limited to its time efficiency in processing and the ability to digitally store, transfer and enhance images quickly and easily (Seeram, 2019). Also, less radiation can be used to produce the image's desired quality. Unlike the use of X-ray film, digital radiography employs the use of a digital image capture device. Thus, digital mammography distinguishes the image acquisition process from the substantive display of the image. This is more advantageous by producing immediate image preview and availability, doing away with costly films and processing steps, a very wide dynamic range, with other processing techniques that enhance the overall display of quality images. Flat panel detectors (FPDs) are the most recommended for digital detectors (Neitzel, 2005), these detectors are grouped as direct and indirect FPDs. The indirect FPDs are made up of amorphous silicon (a-Si) is the most readily available material for commercial FPDs. Being combined with a scintillator in the detector’s outer layer, also made from Caesium Iodide (CsI). This converts X-rays to light hence the a-Si detector is referred to as an indirect imaging device. The light is channeled through the a-Si photodiode layer where it is converted to a digital output signal. The Direct FPDs on the other hand is made up of amorphous selenium (a-Se) FPDs, and they are also called direct detectors since the X-ray photons are converted directly into charges. The outer layer of the flat panel in this type is a high- voltage electrode. X-ray generates radical electron pairs in a-Se, and the transit of these electrons and holes depends on the potential of the bias voltage charge (Wikipedia, 2022). 2.10 Image Quality and Lesion detectability The necessity of image quality for accurate and early detection of breast lesions is of utmost importance to every mammography system making it the goal of each piece of equipment used in mammography. Widely, the dependence on high-quality images in mammography makes the modality the most reliable in the accuracy of breast lesion detection. The term ‘image quality’ University of Ghana http://ugspace.ug.edu.gh https://en.wikipedia.org/wiki/Radiation https://en.wikipedia.org/wiki/Amorphous_silicon https://en.wikipedia.org/wiki/Scintillator https://en.wikipedia.org/wiki/Selenium https://en.wikipedia.org/wiki/Photon 22 refers to the clarity of images and the ability to present radiologically relevant details in an image. In the context of breast imaging, the ability to detect breast lesions must correspond to the high quality of mammographic images.” (Haus & Yaffe, 2000). Screening mammography is an important tool for breast cancer diagnosis and it is frequently used for its early detection (Siu, 2016). Digital mammography provides the potential for a high detection of breast lesions (Pisano et al., 2000), but may be less effective for the detection of small lesions or microcalcifications within dense breasts (Wang, 2017). But for the mammography image to detect early any lesion or microcalcification, the system must exhibit a comparative high contrast since the breast is made up of tissues with almost equal radiation attenuation. Several studies have shown that the cancerous tissues are often denser than the non-infected tissues, hence the amount of X-ray attenuation is expected to be relatively higher, resulting in the contrast variation in the image produced. Digital mammography enables the ability for improved breast lesions detection (Schueller et al., 2008). The higher lesion detectability is directly proportional to higher diagnostic/image quality, and since this is also related directly to an increase in the radiation dose, the radiation delivered must be as low as reasonably achievable (IAEA, 2016). Optimization in mammography systems is so important to image quality and diagnostic accuracy, the best performance of image quality parameters with the radiological dose must be kept as low as possibly achievable, without gaining the best diagnostic outcomes (IAEA, 2016). The Image quality evaluation practice in mammography is an effective support system designed for maintaining high standards, and effective cancer detection and diagnosis (Perry et al., 2008). The Commission further defined the mammographic image quality criteria to include. The image quality criteria defined by the Commission refer to the accurate representation of the internal structures of the breast, and other physical attributes such as contrast, sharpness, artefacts and visualization of microcalcifications. In the image quality assessment, the use of physical quantities such as the measure of contrast, University of Ghana http://ugspace.ug.edu.gh 23 spatial resolution, noise, modulation transfer function (MTF), detective quantum efficiency (DQE) (Perry et al., 2008) and others. Also, the signal difference to noise ratio and the spatial resolution in the system must show lesion and microcalcifications in the breast, its number and size/shape for appropriate interventions. The lesion detectability increases with a decrease in resolutions (Gagne et al., 2001), thus the lesion detection improves within a certain resolution range whiles microcalcification detection decreases at lower doses with roughly a square root dependence on dose (Gagne et al., 2005). 2.11 Breast Compression Breast tissue is randomly flexible, comprising more of the fibro-glandular tissues which decrease exponentially with increasing pressure, unlike fat. Breast compression is useful in the reduction of the mobility of the breast, reduction of scattered radiation, spreading of superimposed tissues and also reducing radiation absorbed by glandular tissue in the breast. The only limitation is the worry of some women about the compression pain and the uncomfortable nature of the process scaring many away from undertaking regular screening programs (Dustler et al., 2012). The European Commission outlines several recommendations including the limiting threshold of the compression force not exceeding 200 N. The appropriate compression is a major contributing factor to the quality of mammographic images, enhancing visualization of lesions whiles reducing radiation scatter and breast dose. The aggregate of breast compression during mammography largely contributes to the image quality achieved (Haus et al., 1977). 2.12 Spatial Resolution Some pertinent details in the breast may be of low-medium contrast and detecting such may not be restricted to high contrast spatial resolution but by contrast resolution. In mammography, the University of Ghana http://ugspace.ug.edu.gh 24 appropriate description of spatial resolution is the potential of the modality in capturing and recording outstanding spatial details and this is crucial to image quality and breast lesion detection. The need to detect microcalcifications dictates that the spatial resolution should be significantly higher than it is for normal radiography. If the appropriate resolution is not achieved as a result of blurring or breast positioning in the image acquisition and breast motion, image blurring is inevitable where details of the image are spread across structural boundaries (Haus & Yaffe, 2000). The modulated transfer function is considered a more complex mode of measuring resolution through the exhibition of the thin relationship that exists between contrast and resolution in imaging. In the same trajectory, an increase in the spatial resolution increases the visibility of breast details enhancing breast lesion detectability and this knowledge inspired the current design of digital mammography systems with a high spatial resolution (Yaffe, 2010). 2.13 Noise The major source of distraction for mammographic imaging is the discrepancy in the anatomical background which can confound diagnostic tasks. Noise is a collection of all disordered signals overspread on the relevant signal in the measuring chain or the transmission system (Haus et al., 1977). The useful signal represents the desired details whiles noise is a perturbation in understanding the usable signal, often noise hides or disturbs the clarity of breast details of interest. Anatomical noise which occurs as a result of normal tissue composition in a radiological image has similar energy dependence as the contrast (Izdihar et al., 2015). When anatomical noise dominates, there is a loss in image quality thus preventing the observer/radiologist from seeing the pathological details they are looking for though the anatomical background only partly acts as noise, and the observer is partly able to ignore the anatomical background in the detection task. Breast lesion detection on a mammographic image does not have equal interpretation if it appears University of Ghana http://ugspace.ug.edu.gh 25 on an adipose or glandular background. Bochud et al (1997) proved that anatomical noise contributes tremendously to the system’s noise when detecting small spherical objects in mammography (Bochud et al., 1997). Noise levels may be different based on the manufacturer of the system under consideration knowing that the noise in an image increases as the detector pixel size decreases. In digital mammography, there is no usefulness for the film’s receptor granularity however, there may be differences in the sensitivity of the receptor, resulting in images unrelated to the detailed tissues in the breast (Fredenberg et al., 2011). As long as the system design ensures that these variations are unperturbed, the noise can largely be eliminated by imaging with uniform X-ray fields and the use of the recorded image as a correction mask to make the image uniform. 2.14 Artefacts Lesion detection in mammography can be a very difficult task because it must be done in the presence of complex breast structure that is often highly variable over the mammogram and differs from patient to patient. Artefacts or undesired contrasts appearing in the image with unrelated anatomical structures in the breast go a long way to lower the quality of the image or increase noise whiles reducing the detection of lesions or difficulty in the interpretation of images (Li et al., 2010). In digital systems, a component of the image receptor and miscalibration or nonuniformity in the detector feedbacks within the same image perimeter. It is worth noting that can easily be eradicated or reduced when the mammography unit is well designed, maintained and properly calibrated (Yaffe, 2010). Artefacts easily mimic lesions and may distract observers hence the specific artefacts per the system in use must be identified, studied/understood and most importantly avoided to enhance the quality of the image. In digital mammography systems, artefacts are categorized into, hardware artefacts, patient-based artefacts, and technologist and software-based artefacts. University of Ghana http://ugspace.ug.edu.gh 26 2.15 The Automatic Exposure Control The automatic exposure control (AEC) system plays an important and basic role in establishing the equilibrium in image quality and the breast absorbed dose in mammography thus achieving a recommendable dose and maximum image quality without the limitations of the composition or thickness of the breast (Salvagnini et al., 2011). In full-field digital mammography (FFDM) systems the AEC form is built to hold pixel value constant even when the thickness or the size of the breast changes and its dependency on contrast does not apply to images captured by the digital detectors but signal-to-noise ratio (SNR). Therefore, most digital mammography AEC modes are built to keep the signal generated by the X-ray detector constant as a function of beam quality (Salvagnini et al., 2015). This however is not a guarantee for improved lesion detectability at different thicknesses. The algorithms behind AEC are keen on the optimization of the X-ray technique (kVp and mAs), the exposure time, and the target/filter combination for a given breast thickness and density. An automated AEC system picks all the necessary exposure parameters, with no operator intervention. 2.16 X-ray Spectrum Understanding the X-ray energy spectrum from a diagnostic X-ray tube is crucial for effectively ranking its image quality and absorbed dose to the patient. The need to improve mammographic image quality has been ascertained. Although mammography is a widely used modality for the early detection of breast cancer, the system’s sensitivity/image quality is largely lowered with breast thickness and density (Izdihar et al., 2015). Factors such as the x-ray spectrum, characterized by the anode/filter combination and tube potential, play a fundamental role in affecting both image quality and absorbed dose whiles image contrast was been shown to be independent of selected mAs values/radiation intensity (Huda et al., 2003). When both the X-ray tube output mAs and X- University of Ghana http://ugspace.ug.edu.gh 27 ray tube voltage kVp are selectively diversified, there appears corresponding variations in image quality and dose with references to the existing tradeoffs between dose and image quality in digital mammography for the detection of lesions within an average breast. Changes in the mAs value affect just the beam quantity and without an effect on beam quality whiles the doses increase with a rise in kVp since both beam quality (more penetrating radiation) and quantity (higher tube output) increase when the kilovoltage setting is increasing (Dance et al., 2000). The use of digital mammography imaging systems strives to keep patient doses as low as reasonably achievable hence a relatively low photon energies spectrum is used to produce the high contrast sensitivity, this energy spectrum can be altered to meet prevailing breast sizes whiles appropriating contrast and radiation dose to the breast. The merger of a molybdenum anode and a molybdenum filter (Mo/Mo) has over the period been the preferred choice for mammographic X-ray tubes. However, there are systems with different anode/filter combinations such as molybdenum/rhodium (Mo/Rh), rhodium/rhodium (Rh/Rh), rhodium/aluminum (Rh/Al) and tungsten/rhodium (W/Rh) are now in use largely (Huda et al., 2003). These new X-ray spectra produce different and improved imaging and dosimetry performance. 2.17 Signal-to-Noise Ratio (SNR) To measure the detection rate of an object such as microcalcifications/lesions in an image, the signal-to-noise ratio (SNR) is largely applicable. The signal-to-noise ratio is a generic measure of the actual signal that reflects the true anatomy of noise or a measure of the image signal in a given region (ROI) to the background (Bochud et al., 1997). A lower signal-to-noise ratio generally results in a grainy appearance of images, hence a dependent parameter to measure the ability to visualize objects in a noisy background. Using the Albert Rose’s model, which is a measure of the image signal in a given region is compared to the background of the mammography image (Erwin University of Ghana http://ugspace.ug.edu.gh 28 et al., 1996). Albert Rose’s model is a mathematical model that describes the perception of contrast in an image. It is used to evaluate the quality of mammography images. Since photon noise is considered to be the dominant source of noise, any increase in the number of photons, such as increasing the mAs (tube current-exposure time product), increases the signal-to-noise ratio (Lemacks et al., 2002). Also, as shown by Nishikawa & Yaffe (1985), the signal-to-noise ratio (SNR) is associated and similar for systems at low frequencies, while at higher frequencies, SNR is determined by the modulation transfer function (MTF). The systems having higher MTF have higher SNR. The MTF is a measure of the ability of an imaging system to transfer contrast from the object to the image. A higher MTF indicates that an imaging system can better distinguish between different parts of an object. The diagnosis of breast cancer from a mammographic image requires the visualization of normal breast anatomy and soft tissue pathology as well as lesions which can be almost negligible in diameter. The ultimate limit of detection of subtle lesions in a radiological image depends not only on signal and contrast but also on SNR. SNR-dependent parameters are noise-equivalent quanta (NEQ) and the detective quantum efficiency (DQE) (Lemacks et al., 2002). University of Ghana http://ugspace.ug.edu.gh https://radiopaedia.org/articles/milliampere-seconds-mas?lang=us 29 CHAPTER THREE MATERIALS AND METHODS 3.1 Overview This chapter gives detailed description of materials and methods used in gathering and analyzing collected data during this study. This chapter also provides the step-by-step procedures used to achieve the expected results. 3.2 Materials Materials utilized during the study involved the mammography machine, the American College of Radiology mammography accreditation phantom, (ACR MAP), semi-circle polymethylmethacrylate (PMMA) slabs, Digital Imaging and Communication in Medicine (DICOM) software, Piranha Quality Control kit, Ocean 2014 software, ImageJ software, bathroom scale, towel, lawn tennis ball, a meter rule, aluminum filters, and Microsoft Excel 2010 software. 3.2.1 Mammography Machine Five digital mammography machines installed at five different facilities (3 hospitals and 2 diagnostic centers) were used for this study. The facilities were the University of Ghana Medical Centre (UGMC), Korle - Bu Teaching Hospital (KBTH), 37 Military Hospital, the Diagnostic Centre and the Paradise Diagnostic Centre. All the three hospitals are government owned whiles the two diagnostic centers are private owned. As at the time of this study, the mammography equipment at KBTH and 37 military hospital were FFDM whiles the other three (3) facilities were CRDM. Specifications of the mammography machines are summarized in Table 3.1 and Figure 3.1 shows the image of a typical mammography machine. University of Ghana http://ugspace.ug.edu.gh 30 Table 3.1: Specifications of the mammography machines used for this study SPECIFICATIONS SYSTEM A SYSTEM B SYSTEM C SYSTEM D SYSTEM E Equipment Type FFDM FFDM CRDM CRDM CRDM Manufacturer Helianthus Siemens Philips Varian Medical systems Planmed OY Equipment Model Helianthus C Mammomat Inspiration MammoDiagnost AR Alpha RT Nuance Manufacture Year 2019 2018 2015 2013 2009 Serial Number M845W/67 W189 (422)276/(21)11076 MAR/0036/CSO 16370 SMH 40967 Function Mode AEC/Manua l AEC/Manual AEC/Manual AEC only AEC/Manual Anode/Filter Type W/Rh W/Rh Mo/Rh Mo/Rh Mo/Mo SID (mm) 65cm 65cm 65cm 60cm 65cm kVp Range 20-35 20-35 20-35 22-30 20-35 mAs Range 1-640 1-600 1-640 40-120 10-400 University of Ghana http://ugspace.ug.edu.gh 31 Figure 3.1 Image of a Mammography machine (Field Work) 3.2.2 Polymethylmethacrylate (PMMA) Slabs Poly (methyl methacrylate) is the scientifically approved name for the synthetic (resin) polymer commonly referred to as acrylic glass. The PMMA shown in Figure 3.2 is a thermoplastic which appears transparent in nature, it is highly compatible with the human tissue hence its use in the production of lenses which are implanted in the human eye for the correction of cataracts. Phantom-based studies and measurements in radiology especially mammography uses the PMMA for Quality Control (QC) test including the various step by step approaches of evaluating optimum functioning of the X-ray machines. The Polymethyl Methacrylate (PMMA) is one of the most appropriate media for breast modeling. The PMMA slabs used in this study were in 10 mm and 5 mm respectively but paired as 20 mm, 45 mm and 70 mm. University of Ghana http://ugspace.ug.edu.gh 32 Figure 3.2. Image of the PMMA slab (Field Work) 3.2.3 The TORMAS Phantom The TORMAS phantom is a Leeds testing tool primarily utilized for the mammography image quality test under quality control. It is thought to be the UK's most sensitive test object and is made to be used quickly and easily on a regular basis to provide equipment effectiveness checks of imaging performance. The TORMAS phantom's internal features allow for the verification of a ten-step grey-scale wedge, a high-contrast resolution limit (1.0 to 20.0 lp/mm), low-contrast large- detail detectability (12 details, 5.6 mm diameter), high-contrast small-detail detectability (11 details, 0.5 and 0.25 mm diameter), irregularly shaped particles on a step-wedge background (median sizes 125, 234). Figure 3.3 shows the TORMAS phantom used in the study with serial number 443. University of Ghana http://ugspace.ug.edu.gh 33 Figure 3.3: The TORMAS Phantom (Field Work) 3.2.4 American College of Radiology Mammography Accreditation Phantom (ACR MAP) The American College of Radiology Mammography Accreditation Phantom as shown in Figure 3.4 is an acrylic phantom purposefully used for verifying the performance of a mammography unit using the visualization abilities to qualitatively evaluate a unit’s capability to capture tiny masses comparable to relevant microcalcification for early detection breast cancer detection. The phantom's objects resemble tumor masses, fibrous microcalcifications and calcifications. The ACR MAP consists of a wax block embedded with six nylon fibers varying in thickness through 0.40 mm towards 1.56 mm, five (5) groups of six high contrast simulation microcalcifications/specks, each with an individual diameter of 0.16 mm to 0.54 mm, and five groups of five low contrast simulation masses, each with an individual diameter of 0.25 mm to 2.00 mm, for a total of sixteen different test objects (ACR, 2019). The ACR MAP simulates a flatten breast tissue like 4.2 cm average glandular/adipose constitution, 50 % of adipose tissue and 50 % of glandular tissue. It also provides physical standard baseline which assures quality images and tests systems sensitivity and quickly detects objects from 0.16 mm to 2.0 mm. The ACR MAP phantom used in this study is manufactured by Gammex with serial number 800004-1410115. University of Ghana http://ugspace.ug.edu.gh 34 Figure 3.4: The American College of Radiologist Mammography Accreditation Phantom (ACR MAP) - (Field Work) 3.2.5 Digital Imaging and Communications in Medicine (DICOM) The National Electrical Manufacturers Association (NEMA) created the Digital Imaging and Communications in Medicine (DICOM) protocol in the early 1990s. It is now a widely used transmission tool for medical imaging and the file format of images obtained from different imaging methods in medicine, including computed tomography (CT), mammography, magnetic resonance imaging (MRI), ultrasound imaging, and radiography, among others. (DICOM, 2012). It has been widely adopted in medical facilities that use medical imaging, and as more hospitals, clinics, and practices start to digitize their medical imaging, archiving, and distribution processes, its application is growing to include many other medical specialties besides radiology. The DICOM software was used in this study to efficiently integrate imaging devices, servers, workstations, printers, and PACS (Picture Archiving and Communication System) from various University of Ghana http://ugspace.ug.edu.gh 35 manufacturers, enabling the archiving, retrieval, and distribution of phantom images and related information as well as ensuring interoperability between medical imaging computer systems. For additional analysis, the DICOM files were exported or saved in other formats like the JPEG format. 3.2.6 Piranha Multimeter Piranha is a Radiation to Information (RTI) package used for an instant X – ray quality control and quality assurance solutions. It is an all-in-one multimeter that can be connected to a computer remotely or via Universal Serial Bus (USB). It works with the diagnostic RTI software ocean 2014, which is used to display, record and report all the measurements, waveforms and facilitate the reading on a screen. For radiography, fluoroscopy, CT, dental, and mammography, use a Piranha QC multimeter. It instantly and simultaneously measures and displays all parameters. The Piranha meter is used in the company with the Ocean 2014 software to display parameters such the tube voltage (kVp), exposure time, half value layer (HVL), total filtration, Dose, Dose rate and presents waveforms. The Piranha system used in this study is the Piranha 657, version 5.5 with serial number, CB2-15020088. Figure 3.5 shows the Piranha QC meter used for this study. University of Ghana http://ugspace.ug.edu.gh 36 Figure 3.5: The Piranha multimeter (Field Work) 3.2.7 Ocean 2014 Software Ocean 2014 is a powerful tool for everybody working with Quality Assurance of X-ray systems (Ocean 2014, 2015), whose interface is shown in Figure 3.6. It is the preferred diagnostic software used with radiation-to-information (RTI) instrument. On a simple-to-read screen, it shows all measurements and waveforms gathered. The Piranha QC meter if plucked to a computer which has the ocean 2014 software already installed on it makes the special quick check runs automatically upon detecting the Piranha QC meter. Gathered data can easily be migrated to Microsoft Excel, printed or saved manually. The software enables a carefully thought-out measurement protocol which can eliminate any setup uncertainties. As a result, there will be fewer University of Ghana http://ugspace.ug.edu.gh 37 errors, more efficiency, and better workflow coordination. Users within the same facility can easily share their measurements and templates with coworkers. Figure 3.6: Interface of the Ocean 2014 software (Field Work) 3.2.8 Microsoft Excel 2016 The software used for the data analysis in this study was Microsoft Excel 2016 version 14.0 included in Microsoft office 2016 installed on a Windows 10 system. Microsoft Excel which is a spreadsheet developed for Windows functions by solving statistical, engineering and financial problems and it is used to display data as line graphs, histograms and charts. It supplies pivot tables and scenario manager that can be used to section data in order to view dependencies between variables (Greg, 2007). It can also be used for numerical methods to solve differential equations in mathematics and physics. Excel 2016 was used to prepare the main data collection sheet for this study and to perform some of the analyses. Figure 3.7 shows the interface of excel 2016 running on Windows 10. University of Ghana http://ugspace.ug.edu.gh 38 Figure 3.7: Interface of Microsoft Excel 2016 (Field Work) 3.2.9 ImageJ Software ImageJ is a potent image analysis program (LOCI, University of Wisconsin). It is a free Java image processing program that can be downloaded or used online on any computer with a Java virtual machine version 1.1 (ImageJ Manual, 2022). 8-bit and grayscale, 16-bit, and 32-bit images can be displayed, edited, examined, processed, saved, and printed using ImageJ. The software whose interface is shown in Figure 3.8, is able to read text-based unprocessed files, like data from spreadsheets, as well as other image formats, including TIFF, GIF, DICOM, FITS, and JPEG. It can measure distances and angles as well as compute statistics for the area and pixel values of specific regions. Measured values can be quickly and easily "cut and pasted" into a spreadsheet during analysis. The open architecture of ImageJ makes it possible for users to create plugins that solve a variety of image processing and exploration issues, including automated hematology University of Ghana http://ugspace.ug.edu.gh 39 systems, rounded live-cell imaging, radiological image processing, and large imaging unit information collations. Figure 3.8: The interface of the ImageJ software (Field Work) Additionally, ImageJ supports common image processing operations like edge detection, sharpening, smoothing, convolution, Fourier analysis, and median filtering. It also supports contrast manipulation and Fourier analysis. It performs geometric transformations like flips, rotations, and scaling. 3.3 Methodology Quality control (QC) also referred to in this study as the pre-exposure/exposure test is a very essential aspect of quality assurance (QA) which involves frequent and annual assessment of the components of an imaging system. Some QCs were performed on the mammography equipment used in this study to ensure their optimum performance. Effective quality control (QC) routines have a great impact on improving image quality and reducing dose to the patient. The quality control (QC) examinations performed on all five machines used included compression force, compression thickness, compression alignment, tube voltage (kVp) accuracy and repeatability, output linearity and repeatability, tube voltage (kVp) and time linearity, half value layer (HVL), University of Ghana http://ugspace.ug.edu.gh 40 and timer repeatability. Quantitative image quality assessments were carried out and the lesion detectability evaluated using observer detection abilities whiles qualitative assessment was done using ImageJ and signal-to-noise ratio calculations. Relevant data collection sheet was designed for recording relevant information from the specific QCs performed. Images acquired from exposures for QC and image quality purposes were exported to the DICOM for further analysis. All the tests were conducted using the IAEA protocols (IAEA, 2009, 2011). 3.3.1 Compression Test For the purposes of high image quality and maximum reduction in dose, it is highly essential for the mammography system to be adequately compressed. Hence the main rationale of this test was to check whether or not the machine gives sufficient compression, as well as to verify the precision of the compression force (IAEA, 2011). 3.3.2 Compression Force The bathroom scale was positioned on the platform above a bath towel on the breast support plate. Centrally positioned beneath the compression paddle was the bathroom scale. On the bathroom scale, a lawn tennis was placed to shield the compression plate in the manner that readings on the scale are not disrupted. The indicator reading on the bathroom scales was also adjusted to the zero mark before the activation of compression. The compression plate was released downward by using the manual compression mode and the compression paddle gradually to firmly hold the PMMA slab. As the lawn tennis deflects its reading translated to the bathroom scale is recorded in kilograms (kg) whiles the compression force indicated on the mammography machines was also recorded in Newtons (N), both on the data collection sheet for further analysis. The compression University of Ghana http://ugspace.ug.edu.gh 41 plate was released using the compression paddle. Tolerance value for the displayed and measured forces should be ±20 N. Figure 3.9 shows the setup for the compression force. Figure 3.9: Set-up for compression force (Field Work) 3.3.3 Compression Thickness To stop the compression plate from deforming and lessen measurement errors in the thickness indication caused by the compression plate's tilt angle, an 18 cm × 24 cm PMMA slabs were used. The PMMA slabs of varying thicknesses (20 mm, 45 mm and 70 mm) respectively were in line with the breast support platform's edge along the chest wall. Compression was activated and released gradually until it was firm to the PMMA slabs with regards to the compression force used in the clinical setting, special care was taken not to compress too much. The displayed and the centrally measured thicknesses were recorded on the data sheet and compared, the tolerance limit University of Ghana http://ugspace.ug.edu.gh 42 of these thicknesses is ± 5 mm of phantom thickness. Figure 3.10 shows the set-up of compression thickness. Figure 3.10: Set-up for compression thickness and alignment (Field Work) 3.3.4 Compression Alignment As in the compression thickness test, the PMMA slabs of varying thicknesses (20 mm, 45 mm and 70 mm) respectively were in line with the breast support platform's edge along the chest wall. Compression was activated and released gradually until it was firm to the PMMA slabs with regards to the compression force u