Dual-domain Multi-path Self-supervised Diffusion Model for Accelerated MRI Reconstruction

2025-03-24

AI Summary

MRI accelerated reconstruction faces three clinical bottlenecks: reliance on fully-sampled ground-truth data, high computational cost, and absence of uncertainty quantification. To address these, we propose the first self-supervised dual-domain (k-space and image-space) diffusion framework that eliminates the need for fully-sampled labels. Our method introduces a self-supervised contrastive regularization strategy for unsupervised training, a lightweight hybrid attention network, and a multi-path denoising sampling scheme to jointly optimize efficiency and reconstruction fidelity. Crucially, it simultaneously generates high-fidelity reconstructions and pixel-wise uncertainty maps. Experiments demonstrate superior artifact suppression and fine anatomical preservation at high acceleration factors (×8), outperforming state-of-the-art supervised and self-supervised baselines. Moreover, the estimated uncertainty maps exhibit strong correlation with true reconstruction errors (Pearson r > 0.92), enabling clinically actionable confidence quantification.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Adversarial Attacks & Robustness
• Reasoning under Uncertainty: Uncertainty Representations

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Magnetic resonance imaging (MRI) is a vital diagnostic tool, but its inherently long acquisition times reduce clinical efficiency and patient comfort. Recent advancements in deep learning, particularly diffusion models, have improved accelerated MRI reconstruction. However, existing diffusion models' training often relies on fully sampled data, models incur high computational costs, and often lack uncertainty estimation, limiting their clinical applicability. To overcome these challenges, we propose a novel framework, called Dual-domain Multi-path Self-supervised Diffusion Model (DMSM), that integrates a self-supervised dual-domain diffusion model training scheme, a lightweight hybrid attention network for the reconstruction diffusion model, and a multi-path inference strategy, to enhance reconstruction accuracy, efficiency, and explainability. Unlike traditional diffusion-based models, DMSM eliminates the dependency on training from fully sampled data, making it more practical for real-world clinical settings. We evaluated DMSM on two human MRI datasets, demonstrating that it achieves favorable performance over several supervised and self-supervised baselines, particularly in preserving fine anatomical structures and suppressing artifacts under high acceleration factors. Additionally, our model generates uncertainty maps that correlate reasonably well with reconstruction errors, offering valuable clinically interpretable guidance and potentially enhancing diagnostic confidence.

Problem

• Reducing MRI acquisition time while maintaining diagnostic quality
• Lowering computational costs of diffusion models for MRI reconstruction
• Providing uncertainty estimation to enhance clinical interpretability

Innovation

• Self-supervised dual-domain diffusion model training
• Lightweight hybrid attention network design
• Multi-path inference strategy for accuracy