Abstract

Abstract Aortic angiopathy is a common manifestation of cardiovascular disease (CVD) and may serve as a surrogate marker of CVD burden. While the maximum aortic diameter is the primary prognostic measure, the potential of other features to improve risk prediction remains uncertain. This study developed a deep learning framework to automatically quantify thoracic aortic disease features and assessed their prognostic value in predicting CVD mortality among heavy smokers. Using non-contrast chest CTs from the National Lung Screening Trial (NLST), aortic features quantified included maximum diameter, volume, and calcification burden. Among 24,770 participants, 440 CVD deaths occurred over a mean 6.3-year follow-up. Aortic calcifications and volume were independently associated with CVD mortality, even after adjusting for traditional risk factors and coronary artery calcifications. These findings suggest that deep learning-derived aortic features could improve CVD risk prediction in high-risk populations, enabling more personalized prevention strategies.