Study Examines CT-Based AI Detection of Incidental Abdominal Aortic Aneurysms

New research presented at the American Roentgen Ray Society (ARRS) conference suggests that artificial intelligence may have a significant impact in facilitating automated detection of incidental abdominal aortic aneurysms (AAAs) on contrast-enhanced computed tomography (CECT) exams.

For the retrospective study, researchers assessed the use of the Viz AAA software (Viz.ai) in 440 patients who had CECT exams. Comprised of 52 percent men and 48 percent women, the cohort included 123 patients (28 percent) with a known AAA, according to the study.

The study authors found that the AI software provided an overall sensitivity of 87.4 percent and a specificity of 94.1 percent for detection of AAAs. For fusiform infrarenal AAAs, the AI software demonstrated a 90.9 percent sensitivity and a negative predictive value (NPV) of 96.5 percent, according to the researchers.¹

In new retrospective research involving 440 patients who had contrast-enhanced CT exams, study authors found that the AI software Viz AAA provided an overall sensitivity of 87.4 percent and a specificity of 94.1 percent for detection of abdominal aortic aneurysms (AAAs).

“The findings validate high specificity and sensitivity for detection of AAA on contrast-enhanced CT by an AI algorithm,” noted lead study author Mark Gedrich, DO, a fourth-year radiology resident affiliated with Cooper University Hospital in Camden, N.J.

The researchers also noted the AI software showed no statistically significant difference between men and women with respect to sensitivity for AAA detection (89.7 percent vs. 88.9 percent).¹

Overall, the study authors maintained that automated detection with the AI software may facilitate timely diagnosis and appropriate intervention for patients with AAA.

“Implementation of an automated detection program can help capture and generate a population of patients eligible for work-up and management of AAA at a vascular clinic,” added Gedrich and colleagues.

The researchers acknowledged that the AI software failed to analyze 10 cases, including CT scans for four patients that had AAA.¹

(Editor’s note: For related content, see “Improving Adherence to Best Practices for Incidental Abdominal Aortic Aneurysms on CT and MRI,” “New AI Algorithm for Abdominal Aortic Aneurysm Detection on CTA Gets FDA Nod” and “Pertinent Insights into the Imaging of Patients with Marfan Syndrome.”)

Reference

1. Gedrich M, Ch J, Clark A, Tolaymat B, Batista P, Gefen R. Validation of artificial intelligence-based detection of infrarenal abdominal aortic aneurysm. Presented at the American Roentgen Ray Society (ARRS) conference April 27-May 1, 2025, San Diego. Available at: https://www2.arrs.org/am25/ . Accessed April 29, 2025.