Bayesian Motion Recovery Framework for Myocardial Phase-Contrast Velocity MRI

摘要

Detailed assessment of myocardial motion provides a key indicator of ventricular function, enabling the early detection and assessment of a range of cardiac abnormalities. Existing techniques for myocardial contractility analysis are complicated by a combination of factors including resolution, acquisition time, and consistency of quantification results. Phase-contrast velocity MRI is a technique that provides instantaneous, in vivo measurement of tissue velocity on a per-voxel basis. It allows for the direct derivation of contractile indices with minimal post-processing. For this method to be clinically useful, SNR and image artifacts need to be addressed. The purpose of this paper is to present a Maximum a posteriori (MAP) restoration technique for high quality myocardial motion recovery. It employs an accurate noise modeling scheme and a generalized Gaussian Markov random field prior tailored for the myocardial morphology. The quality of the proposed method is evaluated with both simulated myocardial velocity data with known ground truth and in vivo phase-contrast MR velocity acquisitions from a group of normal subjects.