An iterative reconstruction technique for metal artifact reduction

Metal artifact reduction (MAR) is a major problem with X-ray computed tomography. In this article we propose a hybrid iterative reconstruction technique for metal artifact reduction. This method is mainly composed of three steps: a pre-iteration step in which an approximate solution is obtained; a simple thresholding sinogram segmentation performed to get an estimation of the metal projection region (MPR); a post-iteration step initialized with the formal solution in which data in the MPR is ignored. In both iteration steps total variation (TV) regularized iterative construction is adopted, taking advantage of TV, which effectively selects an ideal solution from all feasible solutions. This method is easy for implementation and successfully suppresses metal artifacts. (orig.)