Autocalibrated regularized parallel mri reconstruction in the wavelet domain

To reduce the scanning time in some MRI applications, parallel acquisition techniques with multiple coils have been developed. Then, the full Field of View (FOV) image is reconstructed from the resulting registered subsampled k-space data. To this end, several reconstruction techniques have been proposed such as the widely-used SENSE method. However, the reconstructed image generally presents artifacts especially when perturbations occur in both the measured data and in the estimated coil sensitivity maps. In order to alleviate such shortcomings by limiting the distortions, Tikhonov regularization in the image domain has also been investigated. In this paper, we present a novel algorithm for SENSE reconstruction which proceeds with regularization in the wavelet domain, the hyperparameters being estimated from the data. Experiments carried out on real T1-weighted MRI data at 1.5 T indicate that the proposed algorithm generates reconstructed images with reduced artifacts in comparison with conventional reconstruction techniques.