X-ray CT Metal Artifact Reduction Using Segmentation and TV Regularisation

Abstract

Metal artifacts pose a significant challenge in computed tomography (CT) image reconstruction. In this work, we present an approach based on sinogram inpainting and segmentation of both trace and metal objects for metal artifact reduction (MAR). We employ region growing segmentation to extract the metal trace from the sinogram as well as the metal objects. A first-order method is utilized in the sinogram inpainting step. The artifacts are substantially reduced when we apply the segmentation on the metal objects image obtained from the metal trace. To demonstrate the effectiveness of our approach, we evaluate it on both simulated and real images. Our MAR technique yields visually acceptable results with a reduced impact of metallic artifacts on the reconstructed tomographic images.