Cross-Modal Contrastive Learning of ECG and Angiography Representations for Severe Stenosis Classification
Researchers have developed StenCE, a novel pretraining framework designed to identify coronary artery stenosis from electrocardiogram (ECG) data. This method aims to enable early diagnosis by detecting stenosis-specific signals within ECGs, which are non-invasive and routinely acquired. The framework has demonstrated improved performance in classifying severe stenosis and other ECG-related conditions, outperforming previous approaches and offering a new tool for risk stratification. AI
IMPACT Enables early detection of cardiovascular disease using non-invasive ECG data, potentially improving patient outcomes.