Fast Cross-Strength Multi-Contrast Brain MRI Translation using Latent Bridge Matching
Researchers present a single conditional latent bridge matching model for fast MRI cross-field-strength and modality translation, scoring competitively in the MRIxFields2026 challenge.
A unified conditional latent bridge matching model synthesizes MRI contrasts across field strengths without task-specific architectures, performing competitively on all three MRIxFields2026 validation tasks. It generates all modality and field-strength combinations for 30 axial slices in under 90 seconds and full-volume cross-modality-strength translation in under 70 seconds on a single NVIDIA A5000. Code is publicly released alongside extensive component ablations.
- Single conditional latent bridge matching model handles all field-strength and modality translation tasks
- Single inference step yields 30 slices in under 90 seconds on an NVIDIA A5000
- Competitive on all three MRIxFields2026 challenge validation tasks without task-specific training
- Code and extensive ablation studies publicly released
- Vendors
- NVIDIA
- Products
- A5000
- Organizations
- MRIxFields2026
From the original article
1 min read at arxiv.orgMagnetic Resonance Imaging (MRI) acquired at different field strengths exhibits pronounced variation in noise, resolution, homogeneity, and contrast, which limits comparability across acquisition settings and complicates downstream analysis. We address this with a unified conditional model for controllable field-to-field synthesis, built on the framework of conditional latent bridge matching.
By Siddharth Srivastava · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.