Cross-Modal Contrastive Learning of ECG and Angiography Representations for Severe Stenosis Classification

In three linesStenCE, a cross-modal contrastive learning framework, detects severe coronary artery stenosis from non-invasive ECGs. Evaluated across stenosis severity thresholds, the model outperforms prior work and enables early risk stratification in asymptomatic patients.

Summary generated by Claude — human-verified