Photon‐counting normalized metal artifact reduction (NMAR) in diagnostic CT

Purpose Metal artifacts can drastically reduce the diagnostic value of computed tomography (CT) images. Even state-of-the-art algorithms cannot remove them completely. Photon-counting CT inherently provides spectral information, similar to dual-energy CT. Many applications, such as material decomposition, are not possible when metal present. Our aim is develop a prior-based artifact reduction specifically for photon-counting that correct each bin image individually or in their combinations. Methods sorts incoming photons into several energy bins, producing and threshold images containing information. We use this information obtain better prior algorithm FSNMAR. First, we apply non-linear transformation bone-emphasized Subsequently, forward-project multiply resulting sinograms with other element-wise mimic beam hardening effects. These reconstructed linearly combined produce an artifact-reduced image. The coefficients linear combination automatically determined by minimizing threshold-based cost function domain. After thresholding, FSNMAR, which applied individual lowest test our normalized (PCNMAR) on forensic data compare it conventional where generated via sinogram inpainting. For numerical analysis, compute both standard deviation ROI CNR soft tissue fat. Results PCNMAR effectively without sacrificing overall quality. Compared method produces fewer secondary more consistent measurements. Areas contain metal, air, accurate PCNMAR. In some cases, reduced than 50% relative while values similar. If extreme present, unable outperform Using either two, four, only highest yielded comparable results. Conclusions effective reducing available highly beneficial reduction, especially high-energy bin, contains artifacts. While scanning four instead two bins does provide allows freedom selection thresholds.