Relation Between Artificial Intelligence (AI)-Assisted Quantitative Coronary Angiography and Positron Emission Tomography-Derived Myocardial Blood Flow

Percutaneous coronary angiography (CAG) is a standard method for evaluating coronary artery disease. Traditionally, a reduction in the luminal diameter of the coronary arteries by 50% or more during angiography has been considered a significant stenotic lesion. However, the assessment of coronary artery stenosis is usually based on visual estimation by the operator in daily routine clinical practice, which interferes with the objective evaluation.

Quantitative coronary angiography (QCA) has been developed to overcome this limitation. This technique involves the software-based analysis of coronary images obtained through CAG. The previous study showed that there was low concordance between the QCA and visual estimation of coronary artery stenosis (Kappa=0.63) and a reclassification rate of approximately 20%. Furthermore, visual assessments tended to overestimate the degree of coronary artery stenosis, particularly in complex lesions such as bifurcation lesions.

However, there are some limitations to adopting QCA in our daily routine practice. The QCA cannot analyze coronary images on-site and is not fully automated, requiring manual adjustments by humans. Recent advancements have led to the development of artificial intelligence (AI)-based QCA software, which achieves complete automation in the analysis process and provides real-time objective evaluations of coronary artery stenosis.

This study aims to examine the clinical significance of AI-QCA by assessing the correlation between the degree of coronary stenosis detected by AI-QCA and myocardial blood flow abnormalities observed in 13NH3-Ammonia PET scans in patients with coronary artery disease.