Coronary Angiography

Coronary angiography research develops ways to interpret X-ray images and videos of the heart’s arteries, where motion, overlapping vessels, and variable image quality make assessment difficult. Common approaches include U-Net and transformer models for vessel segmentation, video models for stenosis and disease scoring, and multi-view or physics-informed methods for reconstructing 3D anatomy and estimating blood flow. Researchers also use synthetic data, self-supervised and cross-modal learning, and vision-language models to reduce annotation needs and support structured reporting or catheter guidance. A central challenge is ensuring reliable performance across hospitals and acquisition conditions, with external validation and clinical testing needed to establish whether automated measurements improve care.