BAYESIAN MYOPIC PARALLEL MRI RECONSTRUCTION

Parallel magnetic resonance imaging (pMRI) is a technique of accelerating the acquisition of MRI data with high spatial and temporal resolutions. It aims to reconstruct the reduced Field-of-View (FOV) images from under-sampled data in the Fourier space (k-space) by using multiple receiver coils. Therefore, several reconstruction techniques of full FOV image have been proposed. Currently, the most used technique is SENSitivity Encoding (SENSE). However, reconstructed images by SENSE are tainted by artifacts, mainly caused by the noise and inaccurate sensitivity maps. The objective of this paper is to propose a new regularization technique to improve the reconstruction of pMRI images by taking into account the sensitivity maps errors. Our technique is developed in a Bayesian framework using a Markov Chain Monte Carlo (MCMC) sampling scheme and accounts for complex-valued signals. The proposed approach is validated on both simulated and real data. The obtained results show the performance of the proposed technique for the restoration of complex-valued images with sensitivity maps errors.