Free-Breathing, 3D Cardiac Magnetic Resonance Elastography for Myocardial Stiffness Mapping

Magnetic resonance elastography (MRE) of the heart has predominantly utilized breath-held acquisitions with limited anatomic coverage. This work investigates the feasibility of 3D, free-breathing cardiac MRE.

A 3D hybrid radial and EPI acquisition is utilized and combined with retrospective binning of k-space via physiologic monitoring, iterative reconstruction, MRE processing and inversions, and post-processing specialized for cardiac imaging.

Feasibility of free-breathing stiffness mapping of the left ventricular myocardium was demonstrated in 11 participants. Myocardial stiffness estimates were obtained throughout the left ventricle for seven temporal states of the cardiac cycle. Stiffness estimates were presented in participant-specific images, 17-segment bullseye plots, and stiffness plots throughout the cardiac cycle. Results indicate increased stiffness in systolic states, with localized stiffness heterogeneity observed between participants, illustrating a key benefit of 3D cardiac MRE acquisition. Mean stiffness estimates obtained through 11 subjects were 4.44 ± 0.68 kPa (begin-systole), 5.74 ± 0.94, 5.42 ± 0.87 (end-systole), 3.84 ± 0.71 (begin-diastole), 3.51 ± 0.74, 3.54 ± 0.52, and 3.55 ± 0.59 kPa (end-diastole). Statistical testing indicated significant differences in stiffness across cardiac states with one-way, repeated-measures ANOVA (p < 0.001) and Friedman's test (p < 0.001). Experiments in scanning three participants with and without applied motion indicated that stiffness changes were more robustly detected in systole than in diastole, although patterns were subject-specific. Comparison of free-breathing stiffness measurements against reference breath-held measurements yielded similar systolic stiffness estimates.

This study establishes the feasibility of free-breathing, 3D cardiac MRE in healthy volunteers. With the current framework, stiffness estimation appears more robust in systole, and the reliability of diastolic estimates is limited and subject-specific.