Image reconstruction of sparse fan-beam projections using a hybrid algorithm

Abstract

Image reconstruction from sparse fan-beam projection data would result in image error. In this paper, a hybrid imaging algorithm from sparse fan-beam projections is proposed. The first, high exact sparse spectrum data is extracted from image reconstruction from sparse fan-beam projections by filtered back-projection (FBP), and then image is reconstructed from the data using the iterative next neighbor regridding (INNG) algorithm combined with total variation (TV) gradient descent method. The INNG step can restrict image distortions around the model and the TV gradient descent step can remove small oscillations in the model while preserving edges. The combined method is compared with the original INNG algorithm and TV gradient descent method. Computer simulation results demonstrate that the hybrid algorithm is effective for sparse fan-beam projection reconstruction.