Three-dimensional image quality test phantom for AuthorS: planar X-ray imaging John B. Noonoo1 Edem Sosu1,2 Francis Hasford1,2 We aimed to produce a simple, inexpensive 3D printed phantom as a prototype for image quality AFFILIAtIoNS: assessment of contrast, contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR) and resolution in planar 1School of Nuclear and Allied Sciences, University of Ghana, Accra, X-ray imaging systems. The test phantom was designed using SOLIDWORKS software, printed with a Ghana polylactic acid material and filled with paraffin wax. Circular aluminium sheets were used as inserts for 2Radiological and Medical Sciences Research Institute, Ghana Atomic contrast evaluation. A planar X-ray system was used for imaging and DICOM images were evaluated using Energy Commission, Accra, Ghana ImageJ software. We evaluated spatial resolution, contrast, CNR and SNR. For resolution, full width at half maximum (FWHM) was measured on different grid sizes. For contrast, intensity of grey values and CorrESPoNDENCE to: standard deviation were calculated on the different grid sizes. For CNR and SNR, difference in greyscale John Noonoo of investigated tissue and background per standard deviation of noise in the background was calculated. EMAIL: Resolution of the system was evaluated to be 1.57 and 1.80 lp/mm on grids A and B respectively. Contrast jbnoonoo@gmail.com showed differential attenuation per variation in thickness. CNR increased from −13.7 for a thickness of 0.00 mm to 24.90 for a thickness of 28 mm. CNR did not change for a thickness greater than 16.0 mm. DAtES: The SNR of the system fell in the acceptable range of ≥ 5 . The results from the analyses performed received: 05 July 2022 revised: 14 Dec. 2022 indicate that the test phantom has great potential to be a good substitute for the commercially available Accepted: 12 Jan. 2023 phantoms on the market, especially for low-resource settings. Published: 08 Aug. 2023 Significance: hoW to CItE: This study highlights the emergence of 3D printing technology and its suitability within radiology and medical Noonoo JB, Sosu E, Hasford F. physics for the production of cost-effective phantoms which can serve as substitutes for commercial Three-dimensional image quality test phantoms in low-resourced medical facilities in low- and middle-income countries. phantom for planar X-ray imaging. S Afr J Sci. 2023;119(7/8), Art. #14269. https://doi.org/10.17159/ sajs.2023/14269 Introduction ArtICLE INCLuDES: Image quality assessment as a component of quality control in radiology departments is one of the many job ☒ Peer review descriptions of the clinical physicist. A variety of modality-specific phantoms are used in quality assurance ☐ Supplementary material examinations. However, these phantoms are expensive and sometimes delicate, and expert technicians are needed for their use and to evaluate their data.1,2 Typically, commercially available phantoms are in the price range of DAtA AVAILABILItY: USD4000 to USD10 000, depending upon the specifications and the applicable imaging modality. ☐ Open data set ☐ All data included Since its introduction in the 1980s, three-dimensional (3D) printing technology has progressed from its use in ☒ On request from author(s) research facilities to being a widely used method for the construction of phantoms for use in diagnostics and ☐ Not available radiotherapy.3-5 Besides the printer technology, attempts have been made to synthesise materials that can be ☐ Not applicable utilised to create 3D models such as resins. The printing process and material determine the quality of the printed product. 3D printing has recently been utilised to create phantoms for multimodality and modality-specific imaging.6 EDItor: Michael Inggs According to Huda and Abrahams7, image quality in radiological imaging is determined by factors such as contrast, spatial resolution, signal-to-noise ratio and noise. Contrast5 is defined as the difference between the mean greyscale KEYWorDS: in a region of interest in a study material (S t ) and the mean greyscale in a background area of interest (S b ). This 3D printing, phantom, image quality, is commonly referred to as the contrast-to-noise ratio (CNR) in digital imaging. In the presence of noise, it is an contrast, resolution object-size-independent estimate of the signal strength in the study tissue. This is represented as8: FuNDING: None C NR = _ St − S b σ Equation 1b where σ b is defined as the greyscale standard deviation of the noise in the background. Spatial resolution is described9 as the capacity of an imaging modality to distinguish two neighbouring structures as distinct from one another, i.e. image detail visibility. The resolution is mostly estimated using the full-width-at- half-maximum (FWHM) measure in units of line pairs per millimetre (lp/mm). Signal-to-noise ratio is a measure of true signal (real anatomy) to noise. A lower SNR normally results in images with a gritty appearance. In radiology, SNR is proportional to the amount of contrast in the square root of transmitted photons.10 Noise is an undesirable feature in images as it obstructs visualisation and comprehension of an anomaly of interest.11 The two most prevalent sources of noise in medical images are anatomical noise and radiographic noise. Anatomical noise is the term for undesirable anatomical anomalies in an image. As a result, anatomical noise characterisation is task dependent and independent of the inherent performance of a detector. Radiographic noise describes unwanted image variations that are not produced by the image subject. © 2023. The Author(s). Published under a Creative Commons Attribution Licence. Research Article 1 Volume 119| Number 7/8https://doi.org/10.17159/sajs.2023/14269 July/August 2023 3D Image quality test phantom for planar X-ray imaging Page 2 of 7 A good quality control program can be used to evaluate the clinical The PLA material is composed of hydrogen (0.058), carbon (0.541), performance of imaging systems. The outcomes of routine image nitrogen (0.018) and oxygen (0.383). It has a specific density of 1.43 quality control are compared to those acquired during equipment g/cm3, tensile strength of 28.8 MPa, bending strength of 58.6 MPa and acceptance testing or to predefined baseline values at regular intervals. a Hounsfield unit of −530 ±2 5. It has linear attenuation coefficients Differences in image quality are indicated by deviations from the of 0.439, 0.286 and 0.244 at kiloelectron volts of 30, 45 and 60 keV, acceptance test or baseline values. Periodic quality control helps to respectively.13,14 discover departures from optimal efficiency and lays the groundwork for continual development by providing frequent feedback. This might The 3D-printed phantom was filled with paraffin wax, and aluminium be beneficial to patient diagnosis and therapy.12 sheets of 99% purity were inserted into the wax. Image quality assessment, using the phantom, was done on the Philips DuraDiagnost Three-dimensional (3D) printing has gained prominence in recent times Release 4 X-ray machine (Koninklijke Philips N.V, Netherlands). for building volumetric objects with the help of a computer-aided design application and the use of a wide range of materials such as ceramics, ImageJ software (Version 1.51, US National Institutes of Health and resins, metals and thermoplastics (e.g. acrylonitrile butadiene styrene the Laboratory for Optical and Computational Instrumentation (LOCI, (ABS), polyethylene terephthalate glycol-modified (PETG) and polylactic University of Wisconsin), USA) was used for image analysis. A tape acid (PLA)).13 3D printing in radiology has been used predominantly in measure, the Ocean software, Piranha quality control meter, the the construction of phantoms for diagnostic radiology, nuclear medicine collimator and beam alignment quality control test tool and a beam and radiotherapy. It has also been used in the printing of breast phantoms alignment phantom were used for quality control procedures. using materials as tissue substitutes for their attenuation coefficients.14,15 Simple symmetrical phantoms for use in computerised tomography Phantom design and modelling (CT) have been manufactured through 3D printing using tissue equivalent A circular-shaped phantom was designed, with a whole-body diameter materials like resins and thermoplastics.16 Anthropomorphic phantoms of 150 mm, radius of 75 mm, and thickness of 45 mm and with eight depicting the whole body17 as well as body parts such as the spine18 and circular holes on the surface (Figure 1). The holes were arranged in a head19 have all been manufactured and are playing a key role in radiology coordinated orientation with equal tolerance in between them for accuracy departments worldwide. in alignment. This arrangement was relevant for the measurement of contrast. Opposite the circular holes was a group of four squares of Studies have been done on the manufacture of 3D-printed image equal dimensions (27 mm × 27 mm) with gridlines of varied spacing quality assessment phantoms suitable for conventional X-ray imaging, in decreasing order of visibility, which is relevant for resolution. The mammography and fluoroscopy. One study6 used PLA material to print design was saved in the STL format, a universally accepted format a low-contrast phantom with air holes of different radii ranging from for computer-aided designs. ISO/ASTM 52900:2015 was used for the 0.5 mm to 4.5 mm and irradiated with a fluoroscopy machine of 40 kV design and fabrication process. – 70 kV and the results were feasible. The manufacture of a universal image quality phantom for use in general X-ray, mammography, CT and 3D printing fluoroscopy has been explored.2 The fused deposition modelling method was used for printing the Three relevant issues are prevalent in image quality assessment in thermoplastic material. Printing of the phantom was done using different resource-limited facilities: the high cost of commercial phantoms, parameters. The axial setup was in the order of x-50, y-130 and z-9.99. lack of human resources, and time constraints, with cost being chief The temperature of the printer’s nozzle remained constant at 240 °C. among them. In this study, we therefore aimed to use 3D printing Printing time ranged from 1 h to 18 h for the different components of technology to develop an in-house image quality assessment phantom the phantom. A constant bed temperature of 60 °C was maintained for resolution, contrast, contrast-to-noise ratio and signal-to-noise ratio throughout the printing process. The fan embedded in the printer, which for general X-ray imaging systems of a low-resourced centre in a low- to primarily is used for cooling, was maintained at 50% of its capacity middle-income country (LMIC). throughout the printing process. Materials and methods Extruder filling Materials Paraffin wax was used as the filling material for the printed ‘image SOLIDWORKS® software (Version 2019, Dassault Systemes, France) quality’ phantom (Figure 2). Paraffin wax was chosen for its high density was used for the design of the test phantom. The Creality Slicer software and similarity to human tissue by properties. Paraffin wax candles were (V.4.8.2 build 177 win 64, Shenzhen Creality 3D Technology Co., Limited, heated to 80 °C, allowed to cool and poured into the printer extruder China) was used for G-code conversion of the SOLIDWORKS design to (phantom) while in a semi-liquid state to fill in the empty spaces evenly STL files and a Creality CR-20 Pro 3D Printer (Shenzhen Creality 3D without air gaps. The phantom was left in the open for up to 2 h to Technology Co., Limited, China) was used to print the PLA material solidify and evenly fill every space. The finished test phantom had the (density = 1.250 g/cm3) into the 3D test phantom. specifications shown in Table 1. Figure 1: The eight-hole design concept and animation done with SOLIDWORKS. Research Article Volume 119| Number 7/8 https://doi.org/10.17159/sajs.2023/14269 2 July/August 2023 3D Image quality test phantom for planar X-ray imaging Page 3 of 7 DICOM (Digital Imaging and Communications in Medicine) is the primary file format for storing and transferring medical images.The DICOM image acquired (Figure 3) was uploaded to the ImageJ software and analysed using the various image quality assessment tools.4 The acquired X-ray showed fine and clear details with minimal or no noise or artefacts. Resolution Analysis of resolution was done by drawing regions of interest across the gridlines. A polynomial fit of distance versus greyscale values was performed for each gridline. For each grid, the FWHM representing the resolution of the X-ray system was determined. The average resolution of each grid was calculated using Equation 2: Grid X, Line Y = _H_ig_h_e_s_t_ c_u_r_v_eResolution peak _ _p_e_ak_ 2 Equation 2 Resolution = Highest value on xaxis − Lowest value on x − axis…. Contrast Contrast was analysed by drawing regions of interest in the phantom image. Each of the holes in the image represented a different thickness and contrast due to the difference in the number of aluminium discs. Each thickness corresponded to the position of the circular hole in increasing order of contrast. A polynomial graph of thickness against mean greyscale value was plotted to show the curvature of contrast. Contrast-to-noise ratio Noise was determined by using the contrast-to-noise ratio (CNR). The standard deviation of the mean greyscale values was calculated. For each Figure 2: The 3D-printed phantom fully filled with paraffin wax and evenly hole and thickness, the average greyscale in the region of interest (ROI) set up for exposure with a conventional X-ray machine. in the hole was found as well as the average greyscale in the ROI in the surrounding background. These two parameters were used to find the CNR table 1: Specifications and related dimensions of the test phantom together with the standard deviation of the noise in the background. Specification Dimensions (mm) Signal-to-noise ratio Phantom radius (R) 75 The SNR was estimated using Equation 3. For each thickness, the mean grey value in the ROI was found. The standard deviation of the mean Phantom thickness (t) 45 Radius of circular holes (r) 5 Tolerance between circular holes ( T r ) 8 Radius of aluminium discs 3.5 Thickness of aluminium discs 0.8 Spacing of resolution lines/bars 0.8/1.0/1.2/1.4 Radius of aluminium rod 4 Thickness of aluminium rod 45–10 (factor of 5) Contrast squares 27 × 27 Exposure and acquisition of DICOM image Images for the image quality assessment were acquired using the setup depicted in Figure 2. The circular holes on the phantom were filled with aluminium discs of diameter 7 mm. The discs, of 0.8 mm thickness, were placed on top of each other to form varying thicknesses in seven of the eight holes, with the first hole unfilled. The phantom was set up on the couch of the X-ray machine just beneath the source. The distance between the source and the phantom was 700 mm. The X-ray source was set to 52 kVp, 63.0 ms and 32 mAs with 1.35 μGym2. Figure 3: DICOM image of the phantom for image quality assessment. Research Article Volume 119| Number 7/8 https://doi.org/10.17159/sajs.2023/14269 3 July/August 2023 3D Image quality test phantom for planar X-ray imaging Page 4 of 7 grey values within the ROI was calculated. These two parameters were Table 2 shows the accumulation of the distance and initial and final grey used to find the SNR. values of Grids A and B. The distance and grey values for each line of the grids were generated from the ImageJ software. The final grey value is calculated by subtracting the initial grey value from the minimum value. Mean grey value within ROI S NR = ________________________________ Equation 3 This was done for all lines of Grids A and B. Standard deviation of grey value within ROI Spatial resolution of the planar X-ray scanner was determined by calculating the full width at half maximum (FWHM) from each of the Estimation of covariance (CoV) of the measured grey values gives a graphs in Figure 4. The FWHM was calculated using Equation 5. good insight into what is happening with the increase in thickness and was calculated as: F WHM = |z 1 − z 2| Equation 5 = _s_ta_n_d_a_r_d_ d_e_v_ia_t_io_n_ o_f_ _g_rey values CoV _ ______ × 100%.... Equation 4 where z1 is the minimum distance (mm) value corresponding to the mean g rey values minimum grey value at 1/2hmax and z2 is the maximum distance (mm) value corresponding to the maximum grey value at 1/2hmax. Subtracting the minimum value of distance from the maximum value gives the results and discussion resolution value for the grid and line under consideration. Resolution Table 3 shows the spatial resolution measured from the three lines for Grids A and B. It can be observed that the grey value increases The spatial resolution of the planar X-ray system was determined from constantly with a corresponding increase in distance until it peaks the gridlines of the phantom. The phantom consisted of four grids: A, and decreases steadily from the point of the highest peak as it B, C and D. The width of the lines/bars and their spacing for each grid approaches zero. were 0.8 mm, 1.0 mm, 1.2 mm and 1.4 mm, respectively. The ImageJ program evaluated the resolution by means of a data set collected in From Table 3, it can be seen that the average spatial resolution decreases the parameters of distance and greyscale values. A polynomial graph of with decreasing spatial frequency. The spatial frequency of Grid A (Table distance versus greyscale value was plotted for each line of every grid. 2) produced an average FWHM of 1.57 mm, while the spatial frequency Image resolution was evaluated quantitatively. of Grid B (i.e. 1.0 mm width and 1.0 mm spacing) produced an average table 2: Cumulative table of distance and initial and final pixel grey values for Grids A and B Distance Grey value (initial) Grey value (final) (mm) GAL1 GAL1 GAL2 GAL3 GBL1 GBL2 GBL3 GAL1 GAL2 GAL3 GBL1 GBL2 GBL3 – GBL3 0.0 1304.3 1033.2 1429.8 901.0 1028.0 2061.0 360.1 88.2 106.1 18.0 0.0 86.0 0.1 1107.5 945.0 1486.1 1117.3 1616.9 2138.6 163.3 0.0 162.4 234.3 588.9 163.6 0.3 1115.4 988.5 1436.8 2016.8 2431.4 2009.2 171.2 43.5 1131.0 1133.8 1403.4 34.2 0.4 1268.9 1146.1 1551.0 2496.5 2682.5 2147.8 324.7 201.1 227.3 1613.5 1654.5 172.8 0.6 1529.2 1498.3 1792.7 2491.9 2660.0 2470.3 585.0 553.3 469.0 1608.9 1632.0 495.3 0.7 2056.8 1959.4 2129.8 2418.5 2690.2 2677.1 1112.6 1014.4 806.1 1535.5 1662.2 702.1 0.8 2425.4 2174.7 2592.1 2543.3 2848.3 2795.0 1481.2 1229.7 1268.4 1660.3 1820.3 820.0 1.0 2604.2 2538.4 2665.9 2742.1 2881.9 2775.2 1660.0 1593.4 1342.2 1859.1 1853.9 800.2 1.1 2652.4 2534.0 2648.1 2813.6 2761.9 2832.4 1708.2 1589.0 1324.4 1930.6 1733.9 857.4 1.3 2701.8 2623.3 2660.5 2850.0 2670.5 2858.0 1757.6 1678.3 1336.8 1967.0 1642.5 883.0 1.4 2509.6 2612.4 2523.2 2802.0 2770.2 2770.7 1565.4 1667.4 1199.5 1919.0 1742.2 795.7 1.5 2552.0 2553.5 2112.5 2755.4 2694.1 2910.4 1607.8 1608.5 788.8 1872.4 1666.1 935.4 1.7 2362.0 2078.9 1731.5 2375.3 2528.0 2804.0 1417.8 1133.9 407.8 1492.3 1500.0 829.0 1.8 1940.0 1577.3 1388.4 2247.7 2386.3 2455.8 995.8 632.3 64.7 1364.7 1358.3 480.8 2.0 1376.4 1248.5 1323.7 2288.6 2265.1 2182.7 432.2 303.5 0.0 1405.6 1237.1 207.7 2.1 1024.6 1102.9 1387.6 2155.0 1977.5 1978.8 80.4 157.9 63.9 1272.0 949.5 3.8 2.3 944.2 1016.4 1449.6 1585.6 1396.9 1975.3 0.0 71.4 125.9 702.6 368.9 0.3 2.4 1044.5 1068.9 1328.5 1033.9 1035.3 2175.8 100.3 123.9 4.8 150.9 7.3 200.8 2.5 1249.4 1149.4 1570.1 883.0 1030.0 2427.0 305.2 204.4 246.4 0.0 2.0 452.0 Min value 944.2 945.0 1323.7 883.0 1028.0 1975.3 Research Article 4 Volume 119| Number 7/8https://doi.org/10.17159/sajs.2023/14269 July/August 2023 3D Image quality test phantom for planar X-ray imaging Page 5 of 7 Figure 4: Resolution graphs for lines 1, 2 and 3 for Grids A and B. table 3: Resolution for lines of Grids A and B Line 1 Line 2 Line 3 GrID Average of differences Max Min Diff Max Min Diff Max Min Diff A 2.06 0.40 1.66 2.02 0.42 1.60 1.90 0.45 1.45 1.57 B 2.12 0.32 1.80 2.08 0.22 1.86 2.14 0.40 1.74 1.80 FWHM of 1.80 mm. This means that Grid A could be used to resolve lines/ with an increase in thickness saturating from ≥ 20.0 mm. A CoV of ≤5% bars that are 1.57 mm wide with 1.57 mm spacing, while Grid B can be is normally acceptable and ≤10% is within a good range. Data from this used to resolve lines/bars that are 1.80 mm wide with 1.80 mm spacing. analysis are presented in Table 4 and Figure 5. The smaller the FWHM, the better the spatial resolution. Due to the difficulty in the 3D printing of Grid C (i.e. 1.2 mm width and 1.2 mm spacing) and The X-ray intensity attenuation across the material is approximately Grid D (i.e. 1.4 mm width and 1.4 mm spacing), images of Grid C and D the same for thicknesses greater or equal to 16 mm. This implies that were not included for calculating the spatial resolution. As the resolution of the maximum contrast that can be measured by the phantom using Grid A is lower than those of Grid B, Grid C and Grid D, it was more efficient exposure parameters of 52 kVp, 32 mA, 63 ms and a source to phantom in resolving structures with sizes less than 1.80 mm. distance of 700 mm is 4095 at a phantom thickness greater or equal to 16 mm. This indicates the attainment of a saturation point for the mean Resolution of the phantom could be attributed to the type of material grey values. used for measuring the spatial resolution or exposure parameters, such as the source to image (phantom) distance, output voltage, tube current, Contrast-to-noise ratio and exposure time used in the acquisition of images. When the source The CNRs for each of the eight circular targets (holes) were also to image distance increases, the X-ray beam diverges, forming a cone calculated using Equation 1.8 shape and thereby affecting the intensity of the X-ray beam and quantity of X-rays. Also, attenuation due to low kilovoltage peak may lead to the Table 5 shows the average greyscale in the ROI in the investigated desired image not generating. tissues, backgrounds and standard deviations. In this study, the FWHM was calculated based on the slit (i.e. grids) Negative CNR values indicate less signal than noise, and positive CNR 22 method on the digital detector. The resultant penumbral image provided values indicate more activation signal than noise. a line spread function or Gaussian curve, from which the FWHM was From Figure 5b, there is a steady increase in the CNR per aluminium estimated, due to a partial block of the radiation from the source by the thickness. This is because an increase in the depth of aluminium grids.20,21 discs increases the relative X-ray transmittance in the phantom. The Contrast CNR increased from a negative value of −13.7 for a thickness of0.00 mm to 24.90 for a thickness of 28.00 mm. In the exposure of Contrast was measured from eight holes of the same diameter and the phantom for assessment, the first hole was left empty (without any radius (Figure 3), filled with different thicknesses of aluminium inserts. aluminium insert) and this accounted for the 0.00 cm thickness. This is Using the elliptical measurement tool in the Radiant DICOM viewer, an because without any aluminium insert present, there is no attenuation of area of 0.1289 cm2 was drawn in the centre of each circular image X-rays in the medium, indicating less signal than noise. The CNR steadily (i.e. as region of interest) to obtain the intensity of grey values and their increased per increase in thickness because attenuation increased as standard deviation. The thickness of the target increased with greyscale aluminium thickness increased. However, the CNR did not change values, but greyscale values remained fairly the same beyond a target considerably with an increase in thickness from 16 mm to 28 mm. thickness of 16 mm. There is an attainment of saturation in grey value The standard deviation within this range also showed a steady change, for an aluminium thickness beyond 16 mm. Also, the CoV became better indicative of the attainment of a saturation point. Research Article Volume 119| Number 7/8 https://doi.org/10.17159/sajs.2023/14269 5 July/August 2023 3D Image quality test phantom for planar X-ray imaging Page 6 of 7 table 4: Position of holes and their corresponding thicknesses, pixel grey values and signal-to-noise (SNR) ratios Standard Position thickness (mm) Min grey value Max grey value Mean grey value SNr Covariance (%) deviation 1 0.0 618 976 815.44 61.93 13.17 7.6 2 4.0 1393 2019 1744.7 102.65 16.99 5.9 3 8.0 2638 3229 2892.29 99.01 100.26 3.4 4 12.0 3104 3835 3485.29 114.48 30.44 3.3 5 16.0 3431 4095 3793.71 113.39 33.46 3.0 6 20.0 3713 4095 4018.94 85.80 46.84 2.1 7 24.0 3719 4095 4009.25 86.95 46.11 2.1 8 28.0 4039 4095 4064.00 84.97 47.83 2.1 a b Figure 5: (a) Mean pixel grey value per aluminium thickness and (b) contrast-to-noise ratio per aluminium thickness. table 5: Thickness, mean grey pixel values of the region of interest (ROI) in investigated tissue and background, with associated standard deviations and contrast-to-noise ratios (CNR) Mean grey value in roI in Mean grey value in roI in Standard deviation of Position thickness (mm) CNr investigated tissue background background 1 0.0 815.44 2084.17 92.88 −13.66 2 4.0 1744.7 2153.48 94.22 −4.34 3 8.0 2893.29 2199.22 98.48 7.05 4 12.0 3485.39 2276.26 107.63 11.23 5 16.0 3793.71 2240.47 97.51 15.93 6 20.0 4018.94 2145.87 96.05 19.50 7 24.0 4009.25 1994.34 86.96 23.17 8 28.0 4064.00 1849.41 88.91 24.91 Signal-to-noise ratio Conclusion The signal-to-noise ratio (SNR) for each of the eight circular targets This study has shown that 3D printing techniques can be used for (holes) was calculated using Equation 322 as shown in Table 4. the manufacture of test phantoms for image quality assessment in planar The SNR as a measure compares a desired signal to the level of X-ray imaging. We successfully designed and printed a test phantom background noise. The higher the CNR is between structures, the for in-house use in a low-resourced medical imaging facility in a LMIC lower the SNR needed. From Figure 5b, there is a steady increase that assessed image quality successfully. The phantom demonstrated in the CNR per aluminium thickness, hence a higher SNR is required the capability of being used for analysing image quality parameters, for differentiation. The SNR changed steadily from a thickness of including resolution, contrast and CNR on general X-ray imaging systems. 20 mm to 28 mm. The standard deviation within this range also Subsequent plans include acceptance testing and commissioning tests showed a steady change, indicative of the attainment of a saturation for clinical use. This in-house quality control equipment, at a unit point. According to the Rose model, the image quality of a system is price of USD150, could be a good substitute for relatively expensive acceptable if the SNR is greater or equal to 5.23 commercially available phantoms. Research Article Volume 119| Number 7/8 https://doi.org/10.17159/sajs.2023/14269 6 July/August 2023 3D Image quality test phantom for planar X-ray imaging Page 7 of 7 Acknowledgements 10. Cohen-Adad J, Wald LL. Chapter 2.1 – Array coils. In: Cohen-Adad J, Wheeler- Kingshott CAM, editors. Quantitative MRI of the spinal cord. Amsterdam: This research was carried out at the Medical Physics Department of Academic Press; 2014. p. 59–67. https://doi.org/10.1016/B978-0-12-396 the School of Nuclear and Allied Sciences in the University of Ghana 973-6.00005-8 as part of the master’s degree of J.B.N. The authors are grateful to the department for its support. 11. Samei E. Performance of digital radiographic detectors: Quantification andassessment methods. Advances in digital radiography: RSNA Categorical Competing interests Course in Diagnostic Radiology Physics. Durham, NC: Duke UniversityMedical Center; 2003. p. 37–47. We have no competing interests to declare. 12. Reis C, Pascoal A, Sakellaris T, Koutalonis M. Quality assurance and quality Authors’ contributions control in mammography: A review of available guidance worldwide. Insights Imaging. 2013;4(5):539–553. https://doi.org/10.1007/s13244-013-0269-1 J.B.N.: Conceptualisation, methodology, data collection, sample analysis, data analysis, validation, data curation, writing – the initial 13. ISO/PRF 17296-1 – Additive manufacturing – General principles [Internet]. [cited 2021 Dec 03]. Available from: https://standards.iteh.ai/catalog/standar draft, writing – revisions, project management. F.H.: Conceptualisation, ds/iso/a31c7a1c-bb3c-4b56-acb9-33e9602e9a2b/iso-prf-17296-1 methodology, sample analysis; data analysis, validation, data curation, writing – revisions, student supervision, project leadership, project 14. He Y, Liu Y, Dyer BA, Boone JM, Liu S, Chen T, et al. 3D-printed breast management. E.S.: Conceptualisation, methodology, sample analysis, phantom for multi-purpose and multi-modality imaging. Quant Imaging Med data analysis, validation, data curation, writing – revisions, student Surg. 2019;9(1):63–74. https://doi.org/10.21037/qims.2019.01.05 supervision, project leadership, project management. 15. Schopphoven S, Cavael P, Bock K, Fiebich M, Mäder U. Breast phantoms for 2D digital mammography with realistic anatomical structures and attenuation references characteristics based on clinical images using 3D printing. Phys Med Biol. 1. Madamesila J, McGeachy P, Villarreal Barajas JE, Khan R. Characterizing 3D 2019;64(21):215005. https://doi.org/10.1088/1361-6560/ab3f6a printing in the fabrication of variable density phantoms for quality assurance 16. Solomon J, Ba A, Bochud F, Samei E. Comparison of low-contrast detectability of radiotherapy. Phys Medica. 2016;32(1):242–247. https://doi.org/10.1016 between two CT reconstruction algorithms using voxel-based 3D printed /j.ejmp.2015.09.013 textured phantoms. Med Phys. 2016;43(12):6497–6506. https://doi.org/10 2. Groenewald A, Groenewald WA. Development of a universal medical X-ray .1118/1.4967478 imaging phantom prototype. J Appl Clin Med Phys. 2016;17(6):356–365. 17. Lee MY, Han B, Jenkins C, Xing L, Suh TS. A depth-sensing technique on https://doi.org/10.1120/jacmp.v17i6.6356 3D-printed compensator for total body irradiation patient measurement and 3. Ehler ED, Barney BM, Higgins PD, Dusenbery KE. Patient specific 3D printed treatment planning. Med Phys. 2016;43(11):6137. https://doi.org/10.1118/ phantom for IMRT quality assurance. Phys Med Biol. 2014;59(19):5763– 1.4964452 5773. https://doi.org/10.1088/0031-9155/59/19/5763 18. Javan R, Bansal M, Tangestanipoor A. A prototype hybrid gypsum-based 4. Leng S, McGee K, Morris J, Alexander A, Kuhlmann J, Vrieze T, et al. Anatomic 3-dimensional printed training model for computed tomography-guided spinal modeling using 3D printing: Quality assurance and optimization. 3D Print pain management. J Comput Assist Tomogr. 2016;40(4):626–631. https://do Med. 2017;3(1), Art. #6. https://doi.org/10.1186/s41205-017-0014-3 i.org/10.1097/RCT.0000000000000415 5. Chan HK, Griffin J, Lim JJ, Zeng F, Chiu ASF. The impact of 3D printing 19. Kamomae T, Shimizu H, Nakaya T, Okudaira K, Aoyama T, Oguchi H, et al. technology on the supply chain: Manufacturing and legal perspectives. Int J Three-dimensional printer-generated patient-specific phantom for artificial in Prod Econ. 2018;205:156–162. https://doi.org/10.1016/j.ijpe.2018.09.009 vivo dosimetry in radiotherapy quality assurance. Phys Med. 2017;44:205– 211. https://doi.org/10.1016/j.ejmp.2017.10.005 6. Kapetanakis I, Fountos G, Michail C, Valais I, Kalyvas N. 3D printing X-ray quality control phantoms: A low contrast paradigm. J Phys Conf Ser. 20. Malliori A, Daskalaki A, Dermitzakis A, Pallikarakis N. Development of physical 2017;931(1), Art. #012026. https://doi.org/10.1088/1742-6596/931/1/01 breast Phantoms for X-ray imaging employing 3D printing techniques. Open Med 2026 Imaging J. 2020;12(1):1–10. https://doi.org/10.2174/1874347102012010001 7. Huda W, Abrahams RB. X-ray-based medical imaging and resolution. Am 21. Wang S, Pavlicek W, Roberts CC, Langer SG, Zhang M, Hu M, et al. An J Roentgenol. 2015;204(4):W393–W397. https://doi.org/10.2214/AJR.14. automated DICOM database capable of arbitrary data mining (including 13126 radiation dose indicators) for quality monitoring. J Digit Imaging. 2011;24(2):223–233. https://doi.org/10.1007/s10278-010-9329-y 8. De Crop A. Image quality evaluation in X-ray medical imaging based on Thiel embalmed human cadavers [PhD thesis]. Ghent: Ghent University; 2015. htt 22. Welvaert M, Rosseel Y. On the definition of signal-to-noise ratio and contrast- p://hdl.handle.net/1854/LU-8514957 to-noise ratio for fMRI data. PLoS One. 2013;8(11):e77089. https://doi.org/1 0.1371/journal.pone.0077089 9. Vajuhudeen Z, Jones J. Spatial resolution. Radiopaedia.org. 2009 Jun 05. https://doi.org/10.53347/rID-6318 23. Bath M. Evaluating imaging systems: Practical applications. Radiat Prot Dosim. 2010;139:26–36. https://doi.org/10.1093/rpd/ncq007 Research Article 7 Volume 119| Number 7/8https://doi.org/10.17159/sajs.2023/14269 July/August 2023