A Data-Driven Subspace Reconstruction for Distortion-Free Diffusion-Relaxometry Echo Planar Time-Resolved Imaging

Motivation: The subspace-based reconstruction is an SNR-efficient approach for distortion-free diffusion-relaxometry MRI with highly under-sampled echo-planar time-resolved acquisition (EPTI), in which the needed bases can be estimated from simulations. However, the simulations may not be able to fully capture the signal evolution in complex human tissue. Goal(s): To improve the subspace-based EPTI reconstruction by estimating the bases from acquired calibration data. Approach: The efficacy of the new data-driven subspace reconstruction was evaluated with in vivo EPTI experiments. Results: High-resolution, under-sampled EPTI images are reliably reconstructed using the data-driven subspace reconstruction. Impact: Our study presents a new data-driven approach for estimating the bases for the subspace-based echo-planar time-resolved imaging (EPTI) reconstruction, which may better reflect the underlying microstructure than the numerical simulation and further facilitate studies with diffusion-relaxometry MRI.