CTSL: Codebook-based Temporal-Spatial Learning for Accurate Non-Contrast Cardiac Risk Prediction Using Cine MRIs

2025-07-22

AI Summary

This study addresses the challenge of predicting major adverse cardiac events (MACE) from non-contrast cine MRI sequences. We propose a codebook-driven spatiotemporal mediation learning framework that integrates self-supervised pretraining, multi-view knowledge distillation, and motion-cue-guided lesion self-detection—enabling, for the first time, end-to-end MACE risk prediction without myocardial segmentation masks or gadolinium-based contrast agents. Our method decouples spatiotemporal features, models dimensionality-reduced cine short-axis (cine-SA) sequences, and leverages a learnable codebook for compact, robust representation learning. Evaluated on real-world clinical data, the model significantly outperforms conventional contrast-dependent approaches (AUC improvement: +8.2%), while offering rapid, non-invasive, generalizable, and deployable predictions. This work establishes an accessible, efficient paradigm for cardiac risk assessment in resource-limited settings.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

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: On-Device user modeling, personalization, and recommendation

Abstract

Accurate and contrast-free Major Adverse Cardiac Events (MACE) prediction from Cine MRI sequences remains a critical challenge. Existing methods typically necessitate supervised learning based on human-refined masks in the ventricular myocardium, which become impractical without contrast agents. We introduce a self-supervised framework, namely Codebook-based Temporal-Spatial Learning (CTSL), that learns dynamic, spatiotemporal representations from raw Cine data without requiring segmentation masks. CTSL decouples temporal and spatial features through a multi-view distillation strategy, where the teacher model processes multiple Cine views, and the student model learns from reduced-dimensional Cine-SA sequences. By leveraging codebook-based feature representations and dynamic lesion self-detection through motion cues, CTSL captures intricate temporal dependencies and motion patterns. High-confidence MACE risk predictions are achieved through our model, providing a rapid, non-invasive solution for cardiac risk assessment that outperforms traditional contrast-dependent methods, thereby enabling timely and accessible heart disease diagnosis in clinical settings.

Problem

• Accurate contrast-free MACE prediction from Cine MRI
• Self-supervised learning without segmentation masks
• Decoupling temporal-spatial features for cardiac risk assessment

Innovation

• Self-supervised learning from raw Cine MRI
• Multi-view distillation for spatiotemporal feature decoupling
• Codebook-based dynamic lesion self-detection