Calibrationless Parallel MRI with Joint Total Variation Regularization

In this paper, a calibrationless method is proposed for parallel magnetic resonance imaging (pMRI). It is motivated by the observation that the gradients of the aliased images are jointly sparse. Therefore, the pMRI problem can be formulated as a joint total variation regularization task. The field of view is finally obtained via a sum of square approach. We develop an iterative algorithm to efficiently solve this problem. Experiments on pMRI datasets demonstrate that our method outperforms the state-of-the-art pMRI methods even when they can achieve sufficient calibration data, and far better than existing calibrationless pMRI algorithms. Clinic MR applications could benefit from this method even when accurate calibration is limited or not possible at all.