Metal artifacts reduction in x-ray CT based on segmentation and forward-projection

Abstract

X-ray computed tomography (CT) is a powerful clinical diagnosis tool and has been used widely in many clinical and biological settings. Metal artifacts, caused by high density implants, are commonly encountered in clinical CT applications, thereby affecting the detection of abnormal structures and undermining CT's diagnostic value. In this paper, we developed a metal artifact reduction approach based on image segmentation and forward-projection. We further demonstrate the usefulness of our approach by using a biomedical specimen consisting of muscles, bones and metals. Our aim is to remove the inaccurate metal artifact pixels in the original CT slices and exactly reconstruct the soft-tissue using the forward projections with no metal information. During the reconstruction, artifacts are reduced by replacing the metal projection using the forward projection. The presented work is of interest for CT biomedical applications.