An Adaptive Proximal Point Algorithm for Sparse-View CT Image Reconstruction

This paper presents an adaptive proximal point algorithm (APPA) for sparse-view computed tomography (CT) image reconstruction based on total variation regularization. The proposed algorithm chooses an adaptive proximal parameter matrix which is neither necessary symmetric nor constant in each iteration. By using the framework of contraction method, the global convergence result could be established for the proposed algorithm under suitable conditions. Numerical results for 2-D CT reconstruction from simulated digital Shepp-Logan phantom data show that APPA method is effective and practical.