Advancing high-resolution 7 T diffusion MRI: Evaluating phase-encoding correction strategies for distortion...

High-resolution 7 T diffusion MRI (dMRI) is limited by image artifacts that compromise anatomical accuracy. The purpose of this study was to systematically evaluate phase-encoding (PE) acquisition and correction strategies to determine which methods best mitigate geometric distortions and improve data reproducibility.

Five healthy adults were each scanned twice on the NexGen 7 T MRI scanner with ultra-high performance gradients at 0.9 mm isotropic resolution, using a highly oversampled dMRI protocol with four PE directions (AP, PA, RL, LR). From this dataset, we created and processed eleven time-equivalent, 10-min acquisitions, ranging from uncorrected single-PE data to comprehensive 4-way PE schemes. These strategies were quantitatively compared on their geometric alignment with T1-weighted images and on the scan-rescan reproducibility of DTI-derived metrics.

(1) All distortion-corrected schemes significantly improved geometric accuracy over uncorrected data; (2) Strategies correcting with a full set of reversed-PE (2-way) diffusion weighted images (DWIs) outperformed the common approach of using only a single reversed b = 0 image; and (3) a 4-way PE acquisition consistently provided the highest image fidelity and reproducibility. The optimized acquisition enabled high-quality reconstruction of both long-range and fine-scale superficial white matter pathways.

For high-resolution 7 T dMRI, multi-PE acquisition is essential to achieve accurate geometry and stable microstructural estimates (i.e., less residual EPI distortion and better scan-rescan agreement). A 4-way PE scheme provides the most accurate and reproducible results for microstructural and connectivity modeling.