# Evaluation of Geometric Priors for MRI Phase Reconstruction

Martin Benning and Tuomo Valkonen^{*}

The velocity of a fluid along a gradient can be encoded in the phase difference of two complex magnetic resonance images. In order to speed up the measurements, the acquired k-space data is only subsampled. Previously reconstruction of the velocity image has been attempted in a wavelet basis, with further prior on the distribution of static objects in the fluid. Aiming at improved accuracy of the reconstruction of the velocity data from the sparse measurements, we study variational regularisation approaches including bregmanised Total Generalised Variation.

Mathematics Subject Classification: 49N45

Keywords: MRI, compressed sensing, TGV2, TV, Bregman iterations

Minisymposion: High Resolution Imaging with Geometric Priors