Multi-dimensional flow-adapted compressed sensing (MDFCS) for time-resolved velocity-encoded Phase Contrast MRA

Abstract

4D time-resolved velocity-encoded Phase-Contrast MRI (t-v PC MRI) is a fully non-invasive technique to assess hemodynamics in-vivo and can thus be extremely helpful in diagnosis and treatment planning. A common drawback of 4D acquisitions with high spatial and temporal resolution is the long acquisition time, which still inhibits the wider clinical use of such exams. Since a certain degree of data redundancy or sparsity can be expected in temporal as well as in velocity encoding direction, the combination of recently proposed Compressed Sensing techniques with t-v PC MRI might be promising. In this work, a Multi-Dimensional Flow-Adapted compressed Sensing (MDFCS) method is presented, which aims at exploiting these properties by using a dedicated incoherent data acquisition pattern as well as a temporal regularization term. Acceleration factors of up to 7.3 were reached while image quality was preserved. While non-regularized iterative reconstruction led to a drop of 42.8% ± 7.5% in Contrast-to-Noise compared to the fully sampled reference, the proposed MDFCS method was able to achieve values of 90.5% ± 12.2% and to reproduce the flow curve behavior with high accuracy.