Denoising auto-encoding priors in undecimated wavelet domain for MR image reconstruction

Compressive sensing is an impressive approach for fast MRI. It aims at reconstructing MR image using only a few under-sampled data in k-space, enhancing the efficiency of acquisition. In this study, we propose to learn priors based on undecimated wavelet transform and iterative reconstruction algorithm. At stage prior learning, transformed feature images obtained by are stacked as input denoising autoencoder network (DAE). The highly redundant multi-scale enables correlation different channels, which allows robust network-driven prior. reconstruction, DAE incorporated into classical procedure means proximal gradient Experimental comparisons sampling trajectories ratios validated great potential presented