Spatio-Temporal Mixture-of-Modality-Experts Diffusion for Quantitative DCE-MRI Synthesis from Incomplete MR Sequences

2026-06-24

AI Summary

This work addresses the challenge of missing quantitative parameter maps in dynamic contrast-enhanced MRI (DCE-MRI), which arises from risks associated with contrast agents and variability in scanning protocols, causing conventional methods to fail under incomplete input modalities. To overcome this, the authors propose ST-MoME, a conditional 3D diffusion model operating directly in image space. ST-MoME integrates a Swin Transformer backbone with a novel spatiotemporal mixture-of-experts mechanism, enabling voxel-wise and time-step-adaptive multimodal feature fusion through a spatiotemporal gating network. The framework adheres to an interpretable strategy that prioritizes early fusion of structural priors and late fusion of physiological information. Evaluated across 16 missing-modality scenarios in 386 brain tumor patients, ST-MoME consistently outperforms existing methods in composite NMSE, demonstrating particularly superior accuracy in reconstructing vp and ve parameters and critical tumor regions.

Abstract

Quantitative maps from dynamic contrast-enhanced MRI (DCE-MRI) are essential for tumor assessment but are often unavailable due to contrast-agent risks and protocol variability. Prior methods predict these maps from other MRI modalities, yet most assume fixed, fully observed inputs and fail under realistic missingness. We present Spatio-Temporal Mixture-of-Modality-Experts (ST-MoME), a conditional diffusion framework that synthesizes 3D DCE parameter maps from diverse subsets of multimodal MRI. ST-MoME fuses modality-specific expert features through a spatio-temporal gating network that produces voxel-wise, timestep-dependent weights, forming a conditioning tensor that guides denoising. To preserve quantitative fidelity, ST-MoME performs diffusion directly in image space with 3D patch-based training and a Swin-based backbone. On a clinical brain-tumor cohort of 386 patients, we evaluate ST-MoME across 16 controlled modality-availability scenarios. It achieves the lowest mean Normalized Mean Square Error (NMSE) aggregated across all three DCE parameters, with leading performance on $v_p$ and $v_e$, competitive results on $K^{\mathrm{trans}}$, and the lowest reconstruction error within the clinically critical tumor region. A post-hoc analysis of the learned gating dynamics shows a structural-early, physiological-late fusion schedule consistent with clinical intuition.

Problem

• DCE-MRI
• quantitative synthesis
• missing modalities
• multimodal MRI
• tumor assessment

Innovation

• Spatio-Temporal Mixture-of-Experts
• Conditional Diffusion Model
• DCE-MRI Synthesis
• Modality Missingness
• Quantitative MRI