Abstract / Summary
Computed tomography angiography (CTA) is the reference imaging modality for diagnosis, endovascular planning, and surveillance of abdominal aortic aneurysms (AAAs), but conventional assessment remains time-consuming, operator-dependent, and limited by complex anatomy. This narrative review examines the role of artificial intelligence (AI) in three-dimensional aortic CTA segmentation for AAA management. The narrative synthesis drew on literature identified through PubMed/MEDLINE, Scopus and the Cochrane Library, with the original search covering January 2000 to July 2026. Ten primary studies from the existing bibliography were characterized in a study-level table, and reported Dice coefficients were displayed for nine studies. Current evidence indicates that AI-based segmentation, particularly using convolutional neural networks, U-Net-derived models, and nnU-Net frameworks, can improve the speed and reproducibility of aortic lumen, wall, thrombus, branch vessel, endograft, and aneurysm sac delineation. These techniques may support more accurate morphometric measurements, device sizing, procedural planning, quantitative surveillance, and the development of advanced imaging biomarkers. However, available studies remain heterogeneous in datasets, anatomical evaluation, imaging protocols, validation strategies, and reported performance metrics, limiting direct comparison and clinical generalizability. AI-assisted 3D segmentation represents a promising tool for transforming CTA from descriptive imaging into quantitative decision support, but robust multicenter validation, workflow integration, and expert supervision remain essential before widespread clinical adoption.