MSPR-GRE for accelerated multi-echo MRI reconstruction
November 11, 2026
MSPR-GRE is a multi-stage phase-aware self-supervised framework for high-resolution mGRE MRI. It uses sequential unrolled networks and a relaxation-informed temporal subspace to reconstruct complex images and preserve anatomical detail without requiring fully sampled targets for training.
HOW THIS AFFECTS YOU
●
researcherYou can use self-supervised reconstruction to handle undersampled k-space data in multi-echo imaging.
●
healthThis could lead to faster, higher-resolution MRI scans that better preserve quantitative mapping data.