Accelerated Parallel Magnetic Resonance Imaging with Combined Gradient and Wavelet Sparsity

Parallel Magnetic Resonance Imaging (pMRI) is a fast developing technique to reduce MR scanning time. In pMRI, multi-channel coils simultaneously receive a fraction of k-space data and the field of view (FOV) is then reconstructed with the coil profiles. The techniques for pMRI can be mainly divided in two groups: image domain techniques such as PILS, SENSE and Fourier domain techniques like SMASH and GRAPPA. In this paper, we propose a new method based on SENSE framework to reconstruct MR image from multi-coil data. The proposed method combines compressive sensing (CS) to further improve the acceleration rate and utilizes total variation and wavelet sparsity regularization to remove artifacts. Both reconstruction problems can be solved by a recent fastest algorithm. Experiments show that the proposed method outperforms all other previous methods under SENSE framework.