Anguinus Sculpturae: Compositional Synthesis of Peak-Enhancement Breast DCE-MRI Scans

2026-09-27

AI Summary

This study addresses the safety and cost concerns associated with contrast agents in dynamic contrast-enhanced MRI (DCE-MRI) by proposing an efficient method to synthesize peak-enhancement images directly from non-contrast scans. Abandoning complex diffusion models, the approach simplifies the generative process into single-step residual prediction along a straight path. A complete synthesis pipeline is constructed by integrating an nnU-Net segmentation-guided dual-generator architecture, asymmetric Tversky loss, and Gaussian-weighted regional blending, effectively restoring global fidelity while precisely preserving lesion structures. Evaluated on the MAMA-MIA dataset, the proposed method achieves state-of-the-art performance in Fréchet Radiomics Distance (FRD) and Dice scores, demonstrating the effectiveness of this lightweight framework for contrast-free DCE-MRI synthesis.

Abstract

Dynamic contrast-enhanced breast MRI (DCE-MRI) is rich in anatomical and perfusion information, but its reliance on gadolinium-based contrast agents raises safety concerns and adds cost. Virtual contrast enhancement, synthesizing post-contrast from pre-contrast images, is a promising alternative. We address the MAMA-SYNTH challenge task of predicting peak-enhancement breast MRI. Rather than adopting the full machinery of diffusion or flow matching, we observe that under a rectified, straight-line path the generative process collapses to a single difference prediction: the synthetic peak image is the pre-contrast image plus a predicted enhancement map, recovered in one forward pass. Around this we build Anguinus Sculpturae, a compositional pipeline in which nnU-Net segmentations of lesion, foreground and breast region guide two generators - one optimized for global fidelity, one for lesion structure through an asymmetric Tversky term routed via a frozen segmenter - composited region-wise with Gaussian-weighted blending. On the held-out Duke subset of MAMA-MIA our model achieves the best FRD and Dice among all evaluated variants, showing that single-step difference prediction with segmentation guidance suffices to recover both global fidelity and lesion structure. Code is available at https://github.com/MIC-DKFZ/AnguinusSculpturae.

Problem

• DCE-MRI
• virtual contrast enhancement
• peak-enhancement prediction
• gadolinium-free MRI

Innovation

• virtual contrast enhancement
• single-step difference prediction
• segmentation-guided generation
• compositional synthesis
• asymmetric Tversky loss