Comprehensive Evaluation of Artifact Reduction and Tissue Recovery Effects of Metal Artifact Reduction Function Based on Full-Reference Metric

For the comprehensive evaluation of metal artifact reduction (MAR) function, which is a post-processing technique that artifacts reduction caused by metal implants, not only the removal of metal artifacts but also the evaluation of the area restored by MAR is required. In this study, we have conducted the computed tomography (CT) scan both the evaluation image and the reference image at the same location of phantom for the full-reference based evaluation. The evaluation image and reference image were reconstructed into 24 image sets according to the tube potentials, image reconstruction method, and use of the MAR function. Images of two different positions were selected according to the distance from metal and material (bone, tissue) distribution, and bone and tissue were automatically segmented in all images. After Gaussian modeling of each segmented area, FWHM and centroid metrics were extracted. These two metrics were used to compare non-MAR image and MAR image compared to non-metal image. In addition, FWHM NM , FWHM M , CENT NM , and CENT M were calculated using the ratio of the evaluation image and the reference image, then MAR image and non-MAR image were compared. The occurrence of metal artifacts differed according to the distance from the metal and the tube potential. The effect of artifact reduction by the MAR function was confirmed in most regions except for some regions very close to the metal (p < .05). The tissue recovery effect through the application of the MAR function was different depending on the distance from the metal and the material, but the distribution of pixel values was maintained well even after the MAR function was applied (p < .05). Overall, the MAR function restored the original image information while reduction metal artifacts. However, in the regions of severe metal artifacts that are very close to the metal, the MAR function was limited.