AI-Guided Coronary Plaque Analysis from Coronary CTA: An Emerging Paradigm for Personalized Preventive Cardiology

Abstract Coronary artery disease (CAD) remains the leading cause of death worldwide and is driven by atherosclerotic plaque formation. Due to advances in CT technology, coronary CTA (CCTA) has emerged as a leading noninvasive imaging technique to analyze the coronary artery lumen and atherosclerotic plaque. CCTA can characterize plaque types (calcified, noncalcified, and low-attenuation lipid-rich) components, which carry different risks. Total plaque burden measured on CCTA, especially the volume of noncalcified plaque, has emerged as a strong predictor of acute coronary syndrome (ACS), independent of traditional risk factors and calcium score. Contemporary CCTA reporting requires manual plaque segmentation, which can be time-intensive and show suboptimal inter- and intraobserver reproducibility. Artificial intelligence-guided quantitative plaque analysis (AI-QPA) algorithms have emerged to address these challenges and increase analytic throughput. In multiple studies over the past few years, AI-QPA has demonstrated superiority over conventional myocardial perfusion imaging and achieved excellent agreement with expert human readers and invasive imaging. Furthermore, the therapeutic basis of lipid-lowering medications was demonstrated using AI-QPA, ushering in an era of personalized preventative cardiology. This review briefly delves into the common AI-QPA workflow, the inner workings, and validation for the 3 most common commercially available AI-QPA platforms: Cleerly , HeartFlow , and PlaqueIQ (Elucid).