14470
>
clinical study

Performance Assessment of an AI Algorithm for Opportunistic Screening of Abdominal Aortic Aneurysms

Materials & Methods
This retrospective study evaluated an FDA-cleared AI algorithm (Aidoc, Tel Aviv, Israel) designed for automated detection and measurement of abdominal aortic aneurysms (AAA) on CT abdomen/pelvis scans performed in the emergency setting between July 2020 and May 2021. A total of 4,023 studies were analyzed using both AI image review and natural language processing (NLP) of radiology reports to identify unreported aneurysms. Discrepant cases were independently reviewed by a board-certified radiologist.

Results
Among 3,955 NLP-negative studies, 16 were flagged by AI as potentially positive for AAA; 5 of these were confirmed as true, previously unreported aneurysms, representing an enhanced detection rate of 7.4%. This equated to approximately one missed AAA per 800 CT scans. Extrapolated nationally, AI-assisted opportunistic screening could identify up to 18,600 additional AAAs annually across the United State’s population.

Conclusions
AI-based opportunistic screening significantly improved detection of incidental abdominal aortic aneurysms in high-throughput emergency workflows. Implementation of such FDA-cleared tools may enable earlier diagnosis, improved follow-up and enhanced patient outcomes through automated identification and management integration.

Explore the Latest AI Insights, Trends and Research