375 Use Of Artificial Intelligence Software to Detect Intracranial Aneurysms: A Comprehensive Stroke Center Experience

Abstract

INTRODUCTION: The prevalence of unruptured intracranial aneurysms (UIAs) in the United States is 1-3.2%, with an annual risk of rupture ranging from 2-10%. Hemorrhagic strokes are an enormous burden for the healthcare system in the U.S., presenting a cost of over $30,000 per patient. The Viz Aneurysm AI software (Viz-AI) uses an algorithm to screen images and indicate those suspected to have a UIA. METHODS: 770 CTAs were reviewed by neuroradiologists who reported the presence or absence of saccular aneurysms. Subsequently, the images were analyzed by VIZ-AI, if the software suspected an aneurysm, it flagged the corresponding image. In cases where there was a mismatch between the radiologist’s report and the Viz-AI findings, an expert neurosurgeon evaluated CTA images providing a definitive conclusion on the presence or absence of an aneurysm. RESULTS: Viz-AI flagged 33 cases as potential aneurysms, of these, 16 cases were positively identified as aneurysms by radiologists, while 17 were dismissed. A total of 737 cases were considered negative by Viz-AI, while in that same group, radiologists identified aneurysms in 28 CTAs. Compared to the radiologist’s report, Viz-AI performance had a sensitivity of 36%, specificity of 97.6%, and negative predictive value of 96.2%. There were 45 mismatch cases between Viz AI and radiologists, 17 cases were flagged as having aneurysm by Viz-AI but unreported by radiology, the expert neurosurgeon confirmed that 7 of 17 cases had an aneurysm. In 28 Viz-AI negative cases, radiologists indicated aneurysms, 17 of those confirmed by the neurosurgeon. CONCLUSIONS: Viz-AI has the potential to increase the diagnosis of UIAs, however, it must be used as an adjacent tool within the standard of care due to limited applicability in real-world settings.

Ähnliche Arbeiten