Pre- to Post-Contrast Synthesis of Breast DCE-MRI using Latent Bridge Matching

2026-08-07

AI Summary

This work proposes a Latent-space Bridge Matching (LBM) framework to alleviate the scanning burden associated with gadolinium-based contrast agents in breast DCE-MRI by synthesizing contrast-enhanced images as substitutes for real post-contrast acquisitions. Departing from conventional diffusion models that start from Gaussian noise, LBM anchors synthesis on patient-specific anatomical structures and learns a conditional bridging trajectory between pre-contrast and peak-enhancement images within a VAE latent space. Tumor masks are incorporated as conditional inputs, and a latent-space UNet iteratively refines intermediate states to generate anatomically consistent images with high lesion fidelity. Evaluated on 91 cases from the DUKE dataset, tumor-conditioned LBM significantly outperforms baseline methods, achieving an MSE of 0.940, a FRD of 4.716, and a tumor-region SSIM of 0.429.

Abstract

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is central to breast cancer imaging, but gadolinium administration increases scan burden and motivates contrast-reduced alternatives, including synthetic contrast generation. We propose a latent bridge matching (LBM) framework for synthesizing peak-enhanced breast DCE-MRI from pre-contrast images in the MAMA-SYNTH challenge setting. Instead of starting from Gaussian noise as in conventional latent diffusion models (LDMs), the proposed model learns a conditional bridge between paired pre-contrast and peak-enhanced VAE latents. A latent UNet predicts the remaining correction from intermediate bridge states to the peak-enhanced latent, enabling iterative refinement while keeping the trajectory anchored to patient-specific anatomy. We evaluated two LBM conditioning variants on 91 DUKE validation cases. For the tumor-conditioned variant, tumor masks were used as conditioning inputs. Tumor-conditioning improved performance compared with pre-contrast conditioning, reducing MSE from 1.023 to 0.940 and FRD from 7.523 to 4.716, while increasing tumor SSIM from 0.355 to 0.429. The tumor-conditioned LBM also outperformed the evaluated LDM baseline on this validation cohort. These results suggest that latent bridge matching is a promising pre-contrast-anchored formulation for virtual contrast enhancement, while further work is needed to validate generalization and remove dependence on ground-truth tumor masks at inference.

Problem

• DCE-MRI
• contrast reduction
• synthetic contrast
• breast cancer imaging
• virtual enhancement

Innovation

• Latent Bridge Matching
• Synthetic DCE-MRI
• Conditional Latent Diffusion
• Tumor-conditioned Synthesis
• Breast MRI Enhancement