September 15, 2026
AI Helps Radiologists Detect 39% More Brain Aneurysm Cases in Real-World Study
A new study published in the Journal of the American College of Radiology (JACR) found that an artificial intelligence (AI) algorithm can meaningfully enhance brain aneurysm detection when used alongside radiologists. The findings provide important real-world evidence supporting AI as a clinical assistive tool for brain aneurysm detection.
The prospective study, conducted by researchers at Northwell Health, evaluated an FDA-cleared AI algorithm from Aidoc (Tel Aviv, Israel) for detecting intracranial aneurysms on 3,856 CT angiography (CTA) examinations performed across a large integrated health system.
Researchers found that radiologists and the AI tool agreed in more than 96% of cases. AI identified 55 true-positive aneurysm cases that were not reported by radiologists, corresponding to a 39% relative increase in detection compared with radiologist-only performance. Although AI was more sensitive than radiologists alone (84.6% versus 71.8%), meaning it found more aneurysms that were truly present, radiologists were more likely to be correct when they identified an aneurysm (92.7% versus 78.2%). Both radiologists and AI were similarly strong at ruling out aneurysms when none were present and at avoiding false alarms. Radiologists interpreted all exams in routine clinical practice without access to the AI results, allowing investigators to assess the algorithm’s real-world performance and determine the added value of combining AI with physician interpretation.
“Intracranial aneurysms can remain clinically silent until rupture, which can result in devastating consequences for patients,” said senior author Pina C. Sanelli, MD, MPH, FACR, Professor of radiology and Vice Chair of research at the Zucker School of Medicine at Hofstra/Northwell, and director of the Harvey L. Neiman Health Policy Institute Policy Research and IMaging Effectiveness (PRIME) Center. “Many of the additional aneurysms identified by AI in our study were among the smallest lesions. Early detection of small aneurysms provides an opportunity for risk assessment, surveillance, and, when appropriate, treatment before a life-threatening hemorrhage occurs.”
The study was conducted as a prospective “shadow-mode” evaluation, meaning AI processed examinations in parallel but did not influence patient care decisions during the study period. When discrepancies arose, independent expert neuroradiologists reviewed the scans to establish the correct finding. Overall, both radiologists and the AI missed some true aneurysms, but the AI identified 55 additional true-positive aneurysm cases not reported by radiologists, compared with 30 true-positive aneurysms identified by radiologists that were missed by the AI. While 46 of the 101 AI-only findings proved to be false positives, the number of incremental true-positive detections exceeded the false-positive alerts, resulting in a favorable benefit-to-burden ratio.
“The study illustrates the complementary strengths of radiologists and AI,” said lead author Shlomit Goldberg-Stein, MD, FACR, Professor of Radiology at the Zucker School of Medicine at Hofstra/Northwell and Director of Artificial Intelligence in the Department of Radiology at Northwell Health. “The algorithm found additional true-positive aneurysms that enhanced overall detection performance, while radiologists identified important aneurysms that the algorithm missed. Together, they achieved better results than either could alone.”
AI performance varied substantially across care settings. The inpatient setting demonstrated the most favorable AI performance across all metrics, where it identified 18 additional aneurysms while generating only 7 false-positive alerts. Performance was also favorable in the emergency department, while benefits were more modest in the outpatient setting, where AI contributed only four additional detections and generated more false-positive than true-positive findings.
“A likely explanation is that higher-acuity inpatient and emergency settings involve more clinically complex examinations, creating additional opportunities for AI to provide value by serving as a complementary detection tool alongside radiologist interpretation,” said Matthew Barish, MD, FACR, FSAR, Professor of Radiology at the Zucker School of Medicine at Hofstra/Northwell and CMIO of Clinical Shared Services for Northwell Health.
“The findings demonstrate why healthcare organizations should evaluate AI based on how it improves physician performance and patient care in real-world use, not solely on results achieved in its original testing environment,” said Elizabeth Rula, PhD, Executive Director of the Harvey L. Neiman Health Policy Institute and study co-author. “This study shows that AI can deliver meaningful clinical value by helping radiologists find additional aneurysms while also revealing important differences in performance across care settings. These findings underscore the importance of ongoing monitoring and evaluation after implementation.”
To arrange an interview with a spokesperson, contact Nichole Gonzalez at ngonzalez@neimanhpi.org.
About the Harvey L. Neiman Health Policy Institute
The Harvey L. Neiman Health Policy Institute is one of the nation’s leading medical imaging socioeconomic research organizations. The Neiman Institute studies the role and value of radiology and radiologists in evolving health care delivery and payment systems and the impact of medical imaging on the cost, quality, safety and efficiency of health care. Visit us at www.neimanhpi.org and follow us on X, BlueSky or LinkedIn.