🤖 AI Summary
Multi-shot diffusion-weighted imaging (DWI) is severely degraded by motion artifacts, and the absence of artifact-free ground-truth labels critically hinders deep learning–based reconstruction.
Method: We propose a physics-informed synthetic paired-data generation framework that jointly models diffusion physics and motion-induced phase evolution—enabling high-fidelity training data synthesis without requiring real artifact-free labels. Our amplitude-phase co-synthesis architecture embeds multimodal physical priors and employs a U-Net backbone. Trained once on 100,000 synthetic samples, it generalizes robustly across varying spatial resolutions, b-values, undersampling rates, and multi-site/multi-vendor scanners.
Contribution/Results: On in vivo data, our method significantly outperforms conventional approaches (p < 0.001) and achieves clinically validated improvements in motion artifact suppression, reconstruction stability, and robustness—as confirmed by seven radiologists.
📝 Abstract
Diffusion magnetic resonance imaging (MRI) is the only imaging modality for non-invasive movement detection of in vivo water molecules, with significant clinical and research applications. Diffusion weighted imaging (DWI) MRI acquired by multi-shot techniques can achieve higher resolution, better signal-to-noise ratio, and lower geometric distortion than single-shot, but suffers from inter-shot motion-induced artifacts. These artifacts cannot be removed prospectively, leading to the absence of artifact-free training labels. Thus, the potential of deep learning in multi-shot DWI reconstruction remains largely untapped. To break the training data bottleneck, here, we propose a Physics-Informed Deep DWI reconstruction method (PIDD) to synthesize high-quality paired training data by leveraging the physical diffusion model (magnitude synthesis) and inter-shot motion-induced phase model (motion phase synthesis). The network is trained only once with 100,000 synthetic samples, achieving encouraging results on multiple realistic in vivo data reconstructions. Advantages over conventional methods include: (a) Better motion artifact suppression and reconstruction stability; (b) Outstanding generalization to multi-scenario reconstructions, including multi-resolution, multi-b-value, multi-under-sampling, multi-vendor, and multi-center; (c) Excellent clinical adaptability to patients with verifications by seven experienced doctors (p<0.001). In conclusion, PIDD presents a novel deep learning framework by exploiting the power of MRI physics, providing a cost-effective and explainable way to break the data bottleneck in deep learning medical imaging.
