Cyclic Self-Supervised Diffusion for Ultra Low-field to High-field MRI Synthesis
Authors:
- Zhenxuan ZhangZZZhenxuan Zhang
- Peiyuan JingPJPeiyuan Jing
- Zi WangZWZi Wang
- Ula BriskiUBUla Briski
- Coraline BeitoneCBCoraline Beitone
- Yue YangYYYue Yang
- Yinzhe WuYWYinzhe Wu
- Fanwen WangFWFanwen Wang
- Liutao YangLYLiutao Yang
- Jiahao HuangJHJiahao Huang
- Zhifan GaoZGZhifan Gao
- Zhaolin ChenZCZhaolin Chen
Preprint
Abstract
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 \emph{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 $\pm$ 2.70 dB in PSNR, 0.943 $\pm$ 0.102 in SSIM, and 0.0864 $\pm$ 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.
Keywords
All Related Versions
This publication has 24 references indexed in Scilit:
- UNITY: A low-field magnetic resonance neuroimaging initiative to characterize neurodevelopment in low and middle-income settingsDevelopmental Cognitive Neuroscience, 2024
- Synthesized 7T MPRAGE From 3T MPRAGE Using Generative Adversarial Network and Validation in Clinical Brain Imaging: A Feasibility StudyJournal of Magnetic Resonance Imaging, 2023
- Diffusion models in medical imaging: A comprehensive surveyMedical Image Analysis, 2023
- Low-field magnetic resonance image enhancement via stochastic image quality transferMedical Image Analysis, 2023
- On the usability of synthetic data for improving the robustness of deep learning-based segmentation of cardiac magnetic resonance imagesMedical Image Analysis, 2022
- Low‐field MRI: Clinical promise and challengesJournal of Magnetic Resonance Imaging, 2022
- Magnetic field strength dependent SNR gain at the center of a spherical phantom and up to 11.7TMagnetic Resonance in Medicine, 2022
- Score-based diffusion models for accelerated MRIMedical Image Analysis, 2022
- Low‐field MRI: An MR physics perspectiveJournal of Magnetic Resonance Imaging, 2019
- Fast free-form deformation using graphics processing unitsComputer Methods and Programs in Biomedicine, 2010