Meta-analysis of deep learning approaches for automated coronary artery calcium scoring: Performance and clinical utility AI in CAC scoring: A meta-analysis: AI in CAC scoring: A meta-analysis

Ting-Wei Wang and colleagues conducted a meta-analysis to evaluate the performance and clinical utility of deep learning models, particularly Convolutional Neural Networks (CNNs), for automated coronary artery calcium (CAC) scoring. Following PRISMA guidelines, 25 studies involving 19,092 patients were analyzed, focusing on the agreement between deep learning and manual scoring using Cohen’s kappa statistics. The pooled kappa statistic was 0.83, indicating strong concordance. Performance varied across imaging modalities, with coronary CT angiography and standard CT outperforming low-dose CT. Commercially available software like AVIEW CAC was utilized, underscoring the integration of AI in clinical practice. This study highlights the potential of deep learning to enhance CAC scoring efficiency and consistency, reducing interrater variability. However, heterogeneity in data and imaging protocols remains a challenge. Future research should address standardization, multicenter trials, and real-world validation to ensure robust and generalizable performance. This research advances AI-driven risk prediction in cardiovascular care.