Viz.ai has announced new data supporting advancements in neurology and vascular care, with two studies—both set to be presented at the Radiological Society of North America (RSNA) annual meeting (1–5 December 2024, Chicago, USA)—showing positive outcomes and the real-world impact of the company’s artificial intelligence (AI)-based technologies in clinical practice for patients with cerebral aneurysms or abdominal aortic aneurysms (AAAs).
“An aneurysm, whether in the brain or aorta, is a serious condition that can be life-threatening, prompting quick action to be taken to prevent rupture and potentially fatal bleeding,” said Molly Madziva Taitt, vice president of global clinical affairs at Viz.ai. “These studies, using diverse, real-world patient data, show the potential of both Viz AAA and Viz Aneurysm to be effective and highly accurate tools that can assist with a radiologist’s ability to triage and monitor patients appropriately, possibly saving lives.”
The first study, titled “Performance and validation of a machine learning-based aneurysm detection tool in the clinical setting”, assessed the performance of Viz Aneurysm through a retrospective evaluation of 703 head computed tomography angiography (CTA) exams from University of California (UC) Irvine Health System (Orange, USA).
The Viz Aneurysm algorithm successfully detected positive aneurysm cases with a sensitivity of 96.6% and specificity of 98.1%. The overall positive predictive value (PPV) of Viz Aneurysm in this study was 82.4% and the negative predictive value (NPV) was 99.7%. Additionally, it was noted that eight cases were detected with multiple aneurysms and 21 cases with aneurysms <4mm, suggesting Viz Aneurysm may be used to detect and manage patients with a lower risk of rupture, according to a company press release.
“Our findings show that the Viz Aneurysm algorithm successfully detected most of the aneurysms in our clinical dataset and performed consistently across subset analyses,” said Jennifer Soun (UC Irvine School of Medicine, Orange, USA). “Viz Aneurysm’s high accuracy shows its potential use to help detect and manage aneurysms in patients.”
The second study, titled “Performance of AAA AI triage software on real-world patient data”, evaluated the performance of the Viz AAA algorithm across 341 CTA exams of the abdomen from Montefiore Health System (New York City, USA). In this retrospective analysis, the Viz AAA algorithm successfully detected positive aneurysm cases with a sensitivity greater than 85%—whether cases with graft repair were included (86.5%) or not (88.6%)—and specificity of 98.3%. Additionally, it was noted that 25 CTAs with AAA and graft repair were correctly identified as positive by the Viz AAA module.
“These data suggest that this AI triage tool is both highly sensitive and specific, indicating high reliability for cases determined as positive across our diverse patient scans,” said Matthew Lazarus (Montefiore Medical Center, New York City, USA). “We believe incorporation of Viz AAA may help radiologists and other clinicians positively identify patients with AAA and potentially facilitate optimal treatment, management and health outcomes.”
