Advances in Artificial Intelligence for Non-ECG-Gated Coronary Artery Calcium Scoring: A Scoping Review

Artificial intelligence (AI) is rapidly reshaping cardiology, with notable progress in calcium scoring for cardiovascular risk stratification. Emerging approaches now enable AI-driven scoring on non-ECG-gated chest CT, yet questions remain about comparative accuracy and real-world utility. This review synthesizes current work on AI applications in non-ECG-gated coronary artery calcium (CAC) scoring and subsequent cardiac risk assessment. We outline typical model inputs, development strategies, performance characteristics, and use cases across diverse care settings. Across the literature, AI-generated CAC scores generally show strong agreement with manual methods and support effective risk stratification, while markedly reducing processing time and workflow burden. Reported classifications align with clinically relevant outcomes, suggesting the potential to enhance the early identification of at-risk patients and streamline population-level screening. Despite promising results, heterogeneity in datasets, evaluation metrics, and deployment environments limits head-to-head comparisons and broad generalization. Implementation challenges, including image quality variation, integration with reporting pipelines, and oversight of edge cases, also warrant attention. Overall, AI for non-ECG-gated CAC scoring appears accurate, efficient, and clinically useful, with meaningful potential to improve cardiovascular risk assessment when thoughtfully validated and integrated into routine practice.