New MRI method jointly estimates B0 field and distortion-free image from k-space

Standard Echo Planar Imaging (EPI) in diffusion MRI suffers from geometric distortions due to B0 field inhomogeneities. Existing correction methods rely on a sequential process: first reconstruct distorted images using parallel imaging, then estimate the B0 field and correct distortion in the image domain. This sequential approach degrades at high acceleration factors and low SNR, limiting correction quality. Researchers from institutions including Imperial College London and Stanford University have developed a physics-informed framework that jointly estimates both the B0 field and the distortion-free image directly from raw k-space data, without any intermediate parallel-imaging reconstruction.

The method represents both the image and the B0 field as superpositions of Gaussian primitives embedded within an MRI physics forward model. This explicit, continuous parameterization captures smooth regions and tissue boundaries while supporting rotated-view EPI acquisitions without interpolation. The diffusion-weighted image is modeled as real and non-negative, with the image phase absorbed into a per-shot phase factor. By using rotated views, distortions are distributed across multiple phase-encoding orientations, improving point spread function isotropy and providing stronger constraints for B0 estimation. On in vivo brain diffusion EPI, the proposed method attains the closest brain-boundary agreement with a distortion-free structural reference, with the largest improvement over sequential methods at high b-value and high acceleration. Extensive visual comparisons also show improved detail fidelity and noise suppression.