Tag
This paper introduces StenCE, a pretraining framework that uses cross-modal contrastive learning between ECG and X-ray angiography representations to detect severe coronary stenosis from ECGs, achieving high performance and enabling early diagnosis even in asymptomatic patients.