Fast, motion-robust MR elastography with distributed, generalized encoding.

Abstract

Purpose: MR elastography (MRE) data are susceptible to poor quality often caused by long scan times and subject motion, due in part to the specific data sampling requirements for determining motion fields to estimate mechanical properties. By reformulating how motion is encoded and estimated, a more efficient and flexible MRE method that allows for accelerated acquisition and motion robustness is established.

Theory and methods: A novel motion-encoding technique was implemented that uses fully distributed sampling directions in an optimized encoding matrix to collect data efficiently. These data are used in an optimization algorithm to estimate harmonic displacement fields. Simulations and in vivo brain MRE data demonstrate the performance of distributed encoding compared with traditional encoding. Estimation of motion from partial data sets after retrospective volume rejection demonstrates new capabilities for robustness to subject motion.

Results: The proposed method achieved significant acceleration over standard methods, allowing for whole-brain 3D MRE in under 1 min, while maintaining an average 2% difference from traditionally sampled images. If scan time is not prospectively shortened, retrospective removal of images from the data set, such as those corrupted by motion, maintains less than 10% voxel-wise error after removing up to half of a complete data set.

Conclusion: Through prospective reduction in sampling, reducing acquisition time, and retrospective volume rejection, this distributed encoding technique adds significant capability and flexibility to MRE acquisitions.