MBSS-T1: Model-based subject-specific self-supervised motion correction for robust cardiac T1 mapping.

2024-08-21

Medical Image Analysis

AI Summary

To address respiratory and cardiac motion-induced artifacts in cardiac T1 mapping, this paper proposes a model-driven, subject-specific self-supervised motion correction framework. The method requires no ground-truth motion labels or additional navigator sequences. It innovatively integrates physics-based imaging model constraints, differentiable motion parameter estimation, contrastive learning–driven self-supervised loss, and deep network optimization to enable personalized motion modeling and end-to-end correction. Validated on multicenter clinical data, the framework reduces the coefficient of variation (CV) of T1 maps by 37% and decreases root-mean-square error (RMSE) by 42%. These improvements significantly enhance image consistency, quantitative accuracy, and scan robustness. The approach thus provides a reliable technical foundation for noninvasive, precise diagnosis of diffuse myocardial diseases.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Multimodal Learning
• Intelligent Robots: Motion and Path Planning

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Problem

• Improves motion-robust cardiac T1 mapping for diffuse myocardial diseases.
• Addresses challenges in patient compliance and intensity differences in MRI.
• Enables free-breathing cardiac T1 mapping without large annotated datasets.

Innovation

• Subject-specific self-supervised motion correction model
• Combines physical and anatomical constraints for robust T1 mapping
• Enables free-breathing cardiac T1 mapping without large datasets