Accelerated MR image reconstruction using non convex regularized non-subsampled Contourlet transform and total variation based prior

In the domain of compressive sensing MRI (CSMRI), MRI image is reconstructed using partially acquired k-space data, in order to shorten the acquisition time of MRI. Various priors and their combinations have been examined for compressive sensing MRI application. In this paper, we propose a CS-MRI method comprised of the combination of non convex regularized non-subsampled contourlet transform (NSCT) based sparsity and hybrid total variation as a prior. The NSCT captures multi-directional and multi-scale image features effectively, while hybrid total variation preserves edges better while suppressing artifacts. The non conex regularization provides better measure of sparseness and preserve structural details better compared to convex regularization. Alternating direction method of multipliers is used to solve the proposed method. The experimental results of the proposed method are analyzed for different under-sampling patterns.