Constrained TV-minimization image reconstruction from sparse-view diagnostic CT data

Advanced diagnostic CT scanners acquire data at a large number of projections. Conventional image reconstruction algorithms, such as FDK algorithm, are analytic-based. Recently, optimization-based algorithms have been under investigation because they may reconstruct images with improved quality, and has the potential and flexibility for image reconstruction from data in non-conventional configurations, such as data collected at sparse views. In this work, using the adaptive-steepest-descent-projection-onto-convex-sets (ASD-POCS) algorithm, we investigate image reconstruction from swine data sets collected by using Toshiba 320-slice CT scanner. We performed ASD-POCS image reconstruction from 1200-, and 600-view data sets and compared them with those obtained with currently used analytic-based algorithm. The results demonstrate that images reconstructed from 1200- and 600-view data sets by use of ASD-POCS algorithm can be comparable or improved over images obtained with conventional algorithms.