Iterative Navigator-Based Retrospective Motion Correction for Multi-Shot Brain MR Elastography and Evaluation of Single-Shot Post-Reconstruction Correction.

Purpose

Brain MR elastography (MRE) is uniquely sensitive to subject motion, as even small phase deviations can produce large errors in the mechanical properties estimated. Existing motion correction strategies have not been validated for MRE and are often incompatible with complex multi-shot or joint temporal reconstructions.

Methods

We propose COMET (Correction Of Motion in Elastography via Transformation), a framework that estimates rigid-body motion from navigator images and embeds transformation matrices into an iterative reconstruction and displacement-vector reorientation for multi-shot MRE. We additionally validate post-reconstruction correction and displacement-vector reorientation for single-shot MRE. The framework was evaluated on MRE from 9 pairs of EPI single-shot and 9 pairs of spiral multi-shot acquisitions, each containing a with-motion scan and a reference. Additionally, four representative motion patterns with increasing amplitude were simulated to indicate how different motions affect MRE quality and results, and a brain tumor MRE dataset with simulated motion was used to assess clinical effect.

Results

Motion correction improved stiffness accuracy across all tested datasets, reducing average stiffness errors by up to 240.86% in multi-shot spiral MRE, and by up to 44.28% in single-shot EPI, across in vivo motion ranging from 4.81 to 67.58 mm translation and 1.85°-23.63° rotation. In a pediatric glioma dataset, simulated motion completely obscured tumor stiffness contrast, while correction restored tumor stiffness visibility. Robustness of estimated stiffness maps depended on motion amplitude, temporal motion patterns, and encoding strategies.

Conclusion

This approach enables robust motion correction for single- and multi-shot brain MRE, and expands reliable MRE to clinical and challenging populations.