THE CT NUMBER UNIFORMITY AND HOMOGENEITY: WHICH IS BETTER FOR DETECTING THE BEAM HARDENING ARTEFACTS?

This study compares the CT number uniformity and homogeneity for detecting beam hardening artefact from images of the American College of Radiology (ACR) computed tomography (CT) phantom scanned with 20 CT scanners from four manufacturers, and images with and without beam hardening artefact. Software to automatically measure CT number uniformity and homogeneity was developed. For CT number uniformity, the coordinates of the peripheral positions of the region of interest (ROIs) at four different positions and one ROI at the middle of the phantom were automatically determined. For CT number homogeneity, the rectangular ROIs of 32 pixels across 85% of the area of the phantom were automatically arranged. CT number uniformity and homogeneity of images from 20 CT scanners were investigated using an ACR CT phantom. To find usefulness of the CT number homogeneity, images with and without beam hardening artefacts were evaluated. The developed software successfully measured CT number uniformity and homogeneity of the images. All scanners produced achievable range of CT number uniformity, i.e., within 5 HU. However, some scanners had CT number homogeneities less than 5 HU, while others had CT number homogeneities more than 5 HU. It is found that the CT number homogeneity is able to detect the beam hardening artefact, while CT number uniformity is not able to detect it. A system for automatically measure CT number uniformity and homogeneity has been developed. The developed system is easy to implement and human-observer independent. CT number homogeneity is more sensitive for detecting beam hardening artefact than CT number uniformity.