Sparse Tensor Constrained for Low Dose CT Reconstruction

The broad application of X-ray computed tomography (CT) has led to a compelling case for the relevance of low dose CT (LDCT) reconstruction. In this paper, we developed a sparse tensor constrained reconstruction (STCR) algorithm for LDCT imaging. In the proposed algorithm, we try to divide the CT image into patch group and utilize the nonlocal similarity prior information relies on a sparse tensor representation to suppress noise and artifacts, and the objective function is optimized by alternating updated between a CT image reconstruction step, and a patch grouping and sparse tensor coding step in the iterative process. By comparing the STCR with previously published work based on the simulated and real clinical dataset, we demonstrated that the proposed method can lead to a promising improvement of LDCT image quality.