Classification of Left and Right Coronary Arteries in Coronary Angiographies Using Deep Learning

Multi-frame X-ray images (videos) of the coronary arteries obtained using angiography (CAG) provide detailed information about anatomy and blood flow in play a pivotal role diagnosing treating ischemic heart disease. Deep learning has potential to quickly accurately quantify narrowings blockages from CAG videos. A consists videos acquired separately for left artery right (LCA RCA, respectively). The pathology LCA RCA is typically only reported entire CAG, not individual However, training stenosis quantification models difficult when are unknown. Here, we present deep learning-based approach classifying Our enables linkage with pathological findings. We manually labeled 3545 520 (approximately seven per CAG) enable testing models, respectively. F1 scores 0.99 on test set classification set. performance was further investigated extensive experiments across different model architectures (R(2+1)D, X3D, MVIT), input sizes, data augmentations, number used training. results showed that could be curated learning, which an essential preprocessing step downstream application diagnostics