DMD-augmented Unpaired Neural Schr\"odinger Bridge for Ultra-Low Field MRI Enhancement

TL;DR

This paper introduces a novel unpaired MRI translation framework that enhances ultra-low field brain MRI images by combining neural Schr"odinger bridges, diffusion-guided distribution matching, and anatomical structure preservation, improving realism and structural fidelity.

Contribution

It proposes a new unpaired MRI translation method integrating diffusion guidance and structure preservation, addressing the scarcity of paired data and improving image quality.

Findings

Enhanced realism on unpaired benchmarks

Increased structural fidelity on paired data

Outperformed baseline methods in quality metrics

Abstract

Ultra Low Field (64 mT) brain MRI improves accessibility but suffers from reduced image quality compared to 3 T. As paired 64 mT - 3 T scans are scarce, we propose an unpaired 64 mT $\rightarrow$ 3 T translation framework that enhances realism while preserving anatomy. Our method builds upon the Unpaired Neural Schr\"odinge Bridge (UNSB) with multi-step refinement. To strengthen target distribution alignment, we augment the adversarial objective with DMD2-style diffusion-guided distribution matching using a frozen 3T diffusion teacher. To explicitly constrain global structure beyond patch-level correspondence, we combine PatchNCE with an Anatomical Structure Preservation (ASP) regularizer that enforces soft foreground background consistency and boundary aware constraints. Evaluated on two disjoint cohorts, the proposed framework achieves an improved realism structure trade-off, enhancing distribution level realism on unpaired benchmarks while increasing structural fidelity on the paired cohort compared to unpaired baselines.