Evaluation and Treatment Strategy Development of Coronary Heart Disease Guided by OCT Based on Multimodal Deep Learning

In this study, we planned to retrospectively collect 2,000 subjects who underwent optical coherence tomography-guided percutaneous coronary stent implantation with an optical coherence tomography system constructed by a deep learning algorithm from 3 centers to assess stent apposition for coronary stent implantation, and to classify subjects into a group with poor stent apposition (axial distance >400 μm or length >1 mm) and a group with good stent apposition. All subjects were followed up within 12 months after the procedure.

Inclusion criteria

• Age ≥20 years old;
• Angiography was performed, and OCT imaging of criminal blood vessels was performed before intervention;
• Type of coronary heart disease: Unstable angina pectoris (UA), ST elevation myocardial infarction (STEMI) And non-ST elevation myocardial infarction (NSTEMI);

Exclusion criteria

• Lack of medical records;
• Failure to complete follow-up;
• Previous coronary artery bypass grafting;
• Severe liver or kidney insufficiency;
• Infectious diseases, malignancies and bleeding diseases;
• OCT image quality was caused by large thrombus volume or residual blood in lumen and percutaneous coronary angiography Poor and further excluded.