A commercial AI software application is highly accurate for detecting intracranial aneurysms on CT angiography studies, according to research presented April 26 at the American Association of Neurological Surgeons (AANS) 2025 meeting.
The analysis also found that the Rapid Aneurysm software (by RapidAI) spotted a “substantial number” of aneurysms that were missed by radiologists, highlighting its value as a complementary tool in the early detection of these high-risk patients.
“These findings underscore AI’s potential role in enhancement of diagnostic accuracy, and proper management of [intracranial aneurysms],” wrote presenting author Reza Dashti, MD, PhD, of Stony Brook University in New York.
The study included 11,694 consecutive CT angiography (CTA) studies completed between September 2020 and June 2024. Of these, 1,206 were flagged by the radiologists and/or the Rapid Aneurysm software as showing a potential aneurysm. On clinical review, 524 of these studies had one or more confirmed aneurysms (4.5%).
Rapid Aneurysm correctly detected 460 of the 524 aneurysms (sensitivity 87.8%); the radiologists detected 372 aneurysms. The software identified 150 aneurysms that had been missed in the original radiology report; 63 aneurysms detected by the radiologist were missed by the software. The software correctly identified 10,863 of the 11,109 aneurysm-negative studies, with a specificity of 97.8%.
Furthermore, Rapid Aneurysm was especially effective at identifying aneurysms greater than or equal to 3 mm in size, with a sensitivity of 92.5% and a specificity of 96.4%.
“This study shows how AI can augment clinical expertise and meaningfully advance diagnostic accuracy in real-world settings,” Dashti said in an association statement.
Leave a Reply