Development and validation of an AI system for AAS CTA diagnosis and mapping

Acute aortic syndrome (AAS), encompassing aortic dissection (AD), intramural hematoma (IMH), and penetrating aortic ulcer (PAU), demands urgent computed tomography angiography (CTA) diagnosis, while manual diagnosis efficiency and accuracy remain limited. This study develops and externally validates an AI-powered AAS decision support system (AAS-DSS) for automated 17-zone aortic segmentation (per the extended Society for Vascular Surgery/Society of Thoracic Surgeons [SVS/STS] classification) and slice-level AAS subtype classification. Trained on 586 CTA scans with nnUNet version 2 (nnUNet v.2), TotalSegmentor, and ResNet-18, and validated on 198 multi-institutional cases, AAS-DSS achieves excellent segmentation and high classification accuracy, outperforming junior radiologists and showing performance non-inferior to those of senior radiologists, with reduced interpretation time. The findings confirm AAS-DSS's strong cross-institutional generalizability, accelerating time-critical AAS diagnosis and supporting standardized management, especially in resource-limited settings.