BibTex format
@article{Zhang:2026:10.1109/TMI.2026.3685559,
author = {Zhang, Z and Jing, P and Wang, Z and Briski, U and Beitone, C and Yang, Y and Wu, Y and Wang, F and Yang, L and Huang, J and Gao, Z and Chen, Z and Islam, KT and Yang, G and Lally, PJ},
doi = {10.1109/TMI.2026.3685559},
journal = {IEEE Trans Med Imaging},
title = {Cyclic Self-Supervised Diffusion for Ultra Low-field to High-field MRI Synthesis.},
url = {http://dx.doi.org/10.1109/TMI.2026.3685559},
volume = {PP},
year = {2026}
}
RIS format (EndNote, RefMan)
TY - JOUR
AB - Synthesizing high-quality images from low-field MRI holds significant potential. Low-field MRI is cheaper, more accessible, and safer, but suffers from low resolution and poor signal-to-noise ratio. This synthesis process can reduce reliance on costly acquisitions and expand data availability. However, synthesizing high-field MRI still suffers from a clinical fidelity gap. There is a need to preserve anatomical fidelity, enhance fine-grained structural details, and bridge domain gaps in image contrast. To address these issues, we propose a cyclic self-supervised diffusion (CSS-Diff) framework for high-field MRI synthesis from real low-field MRI data. Our core idea is to reformulate diffusion-based synthesis under a cycle-consistent constraint. It enforces anatomical preservation throughout the generative process rather than just relying on paired pixel-level supervision. The CSS-Diff framework further incorporates two novel processes. The slice-wise gap perception network aligns inter-slice inconsistencies via contrastive learning. The local structure correction network enhances local feature restoration through self-reconstruction of masked and perturbed patches. Extensive experiments on cross-field synthesis tasks demonstrate the effectiveness of our method, achieving state-of-the-art performance (e.g., 31.80 ± 2.70 dB in PSNR, 0.943± 0.102 in SSIM, and 0.0864 ± 0.0689 in LPIPS). Beyond pixel-wise fidelity, our method also preserves fine-grained anatomical structures compared with the original low-field MRI (e.g., left cerebral white matter error drops from 12.1% to 2.1%, cortex from 4.2% to 3.7%). To conclude, our CSS-Diff can synthesize images that are both quantitatively reliable and anatomically consistent. The code is available at: https://github.com/ayanglab/CSS-Diff.
AU - Zhang,Z
AU - Jing,P
AU - Wang,Z
AU - Briski,U
AU - Beitone,C
AU - Yang,Y
AU - Wu,Y
AU - Wang,F
AU - Yang,L
AU - Huang,J
AU - Gao,Z
AU - Chen,Z
AU - Islam,KT
AU - Yang,G
AU - Lally,PJ
DO - 10.1109/TMI.2026.3685559
PY - 2026///
TI - Cyclic Self-Supervised Diffusion for Ultra Low-field to High-field MRI Synthesis.
T2 - IEEE Trans Med Imaging
UR - http://dx.doi.org/10.1109/TMI.2026.3685559
UR - https://www.ncbi.nlm.nih.gov/pubmed/42009337
VL - PP
ER -