This scoping review analyzed advances in AI-driven coronary artery calcium (CAC) scoring from non-ECG-gated chest CT. Reviewing 13 studies, AI-based CAC scoring demonstrated strong agreement with manual methods (ICC 0.714-0.955, weighted Kappa 0.5-0.84) and supported effective risk stratification while markedly reducing processing time (average 3.5s vs 261s). AVIEW software (Coreline Soft Co., Ltd.) was utilized in multiple studies as a commercial deep learning-based fully automatic CAC scoring tool, demonstrating high reliability on both non-ECG-gated LDCT and ECG-gated cardiac CT. AI-generated CAC scores were valid for predicting cardiovascular disease mortality risk, though limitations included false negative rates and underestimation tendency due to motion artifacts.
목록으로
