Assessment of CT number accuracy, image uniformity and noise of different computed tomography imaging machines

Abstract

Regular and recommended quality control process is essential for quality images to avoid repeated examinations. Since the dose reduction technique is important for radiation protection of patients and repeated examination should be minimized in computed tomography scan. For this reason, quality control tests for image quality based on CT number accuracy, image uniformity and image noise were done regularly. The study was to determine whether Hounsfield of water varies across the images and whether the standard deviation of the HU for a large region of interest at the centre of a reconstructed image is within tolerance as part of the QC process of varried CT machines. Additionally, the QC test is also to determine whether the CT number or the HU varies across different materials or tissues other than water.The design methodology used for this study was based on IAEA harmonized diagnostic radiology quality control programme for diagnostic radiology. These were done with two Toshiba 128 and 640 slice CT machine, two 64 slice Philip CT equipment and two 16 slice General Electric CT machine in Accra, which were all installed between 2016 and 2022. Comprehensive cross-sectional quality control procedures were performed using two phantoms, including, manufacturers CT water phantom and ACR CATPHAN. A software analysis was used to estimate CT number accuracy, image uniformity and image noise, CT numbers of different materials and the recommended levels. The ROI was drawn in both the centre of the image and four ROIs at the periphery of the image. The mean and standard deviation were recorded, the mean values of the CT number and standard deviation were then used to estimate CT number uniformity and image noise. Additionally, CT numbers of different materials using the CATPHAN which contain different materials were analysed using the ROI technique. The study results shows that the measured HU for water were between 0.24-2.21, -0.59-1.85 and -2.31-0.84 for Center A, B and C respectively. Additionally, the results of the various measured uniformity tests were 0.07, 0.9 and 0.76 at center A, B and C respectively. This is within the accepted recommendations by both the IAEA harmonized CT protocol for diagnostic radiology and the manufacturers recommendation of +/-5. The study also shows that the maximum standard deviation between the center ROI and the peripheral ROI were greater than +/-5 HU. Therefore, both the two 64 slice Philips scanners and the 16 Slice GE scanner passed the noise test. These were done with two Toshiba 128 and 640 slice CT machine, two 64 slice Philip CT equipment and two 16 slice General Electric CT machine passed CT number of water and `image uniformity. However, they failed image noise tests because the values were outside the manufacturer’s recommendation and acceptable limit by the IAEA harmonized CT protocol for diagnostic radiology for Africa.