Self-Alignment Learning to Improve Myocardial Infarction Detection from Single-Lead ECG

2025-09-22

AI Summary

Single-lead electrocardiogram (ECG) signals lack spatial information, limiting myocardial infarction (MI) detection performance. Method: We propose SelfMIS, a self-alignment learning framework that bypasses conventional data augmentation and generative modeling. Instead, it directly aligns single-lead and multi-lead ECG representations in the latent space via a novel self-cropping strategy—shifting the learning objective from transformation invariance to representation enhancement—and thereby improves local signal modeling within global cardiac pathological context. The framework relies solely on an encoder network and self-supervised learning, ensuring architectural simplicity and low computational overhead. Contribution/Results: SelfMIS achieves state-of-the-art performance across all nine MI subtypes, consistently outperforming baseline models. This demonstrates the effectiveness and generalizability of latent-space cross-lead alignment for MI detection from single-lead ECGs.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Myocardial infarction is a critical manifestation of coronary artery disease, yet detecting it from single-lead electrocardiogram (ECG) remains challenging due to limited spatial information. An intuitive idea is to convert single-lead into multiple-lead ECG for classification by pre-trained models, but generative methods optimized at the signal level in most cases leave a large latent space gap, ultimately degrading diagnostic performance. This naturally raises the question of whether latent space alignment could help. However, most prior ECG alignment methods focus on learning transformation invariance, which mismatches the goal of single-lead detection. To address this issue, we propose SelfMIS, a simple yet effective alignment learning framework to improve myocardial infarction detection from single-lead ECG. Discarding manual data augmentations, SelfMIS employs a self-cutting strategy to pair multiple-lead ECG with their corresponding single-lead segments and directly align them in the latent space. This design shifts the learning objective from pursuing transformation invariance to enriching the single-lead representation, explicitly driving the single-lead ECG encoder to learn a representation capable of inferring global cardiac context from the local signal. Experimentally, SelfMIS achieves superior performance over baseline models across nine myocardial infarction types while maintaining a simpler architecture and lower computational overhead, thereby substantiating the efficacy of direct latent space alignment. Our code and checkpoint will be publicly available after acceptance.

Problem

• Detecting myocardial infarction from single-lead ECG with limited spatial information
• Converting single-lead to multi-lead ECG leaves large latent space gap
• Existing alignment methods focus on transformation invariance mismatch single-lead detection

Innovation

• Self-cutting strategy pairs multi-lead with single-lead segments
• Direct latent space alignment enriches single-lead representation
• Learns global cardiac context from local ECG signal