RapidAI Outperforms VizAI in Detecting Large Vessel Occlusion

Harmeet Sachdev, MD

Findings from the DUEL study showed that use of RapidAI led to a higher percentage of large vessel occlusion (LVO) cases being identified, as well as a more accurate recognition of LVO-negative cases in comparison with Viz LVO, another FDA-cleared deep-learning software. All told, the substantial number of LVOs missed by the Viz software could have significant implications on LVO diagnosis and treatment times.¹

Presented at the 2025 International Stroke Conference (ISC), held February 5-7 in Los Angeles, California, the primary analysis of the study featured 1525 CTAs from consecutive patients with stroke between June 2022 and June 2024. Among cases that met inclusion criteria (M = 709; F = 816; unknown = 1), 149 (10%) had an LVO, which was confirmed by radiology reports combined with an expert review of both CTA and CT perfusion (CTP) imaging.

Led by Harmeet Sachdev, MD, a neurologist at Good Samaritan Hospital, use of RapidAI software successfully processed more than 99% (n = 1523) of eligible cases, whereas Viz successfully processed 90% (n = 1432). In comparison, Rapid software identified 98% (n = 146) of LVOs compared with 74% (n = 110) by Viz (P = .000000002). More importantly, Rapid correctly identified 94% of LVO-negative cases compared with 91% for Viz (P = 0.004), while also detecting 39 LVOs missed by Viz, whereas Viz identified 3 LVOs missed by Rapid.

"This study underscores the importance of evaluating the performance and reliability of the tools clinicians use for decision-making support," Sachdev told NeurologyLive^®. "While AI has significant potential, it is essential that these technologies consistently provide accurate and dependable results in critical clinical situations."

Viz LVO and RapidAI are both artificial intelligence (AI)-powered tools designed to improve stroke care, but they have different functionalities and approaches. Viz LVO, which primarily focuses on the detection of LVOs, primarily uses deep learning algorithms for LVO detection and care coordination whereas RapidAI uses AI-based imaging analysis, including perfusion maps, to assess stroke severity, treatment eligibility, and potential outcomes. RapidAI is considered more ideal for a full comprehensive stroke evaluation, including identifying LVOs, analyzing salvageable tissue, and managing hemorrhages.

READ MORE: Advancements in Mechanical Thrombectomy and Imaging: A Stroke Neurologist’s Perspective

Regarding Rapid LVO, Sachdev added, "When an LVO is detected, the system automatically notifies the entire care team within minutes of the scan, delivering results along with compressed source images via PACS, email, and the mobile app. By replacing traditional linear workflows with real-time, simultaneous alerts, Rapid LVO helps streamline communication across the care team. This accelerated process not only improves code stroke efficiency but also supports faster, more coordinated decision-making."

In the study, LVO was defined as an occlusion or high-grade stenosis of the intracranial internal carotid artery (ICA) or the M1 segment of the middle cerebral artery (MCA), anatomically traced from the MCA origin to its genu at the inferior aspect of the Sylvian fissure. High-grade stenosis was confirmed using CTA and/or evidence of a corresponding perfusion deficit in the ICA/MCA territory on CTP. A total of 66 cases were excluded due to issues with AI processing (n = 33) or if there was poor bolus, metal artifact, or brain hemorrhage present (n = 33).

In late 2024, the Alaska Stroke Coalition (ASC) announced a partnership with RapidAI on the “Rapid AK” project, a 3-year initiative focused on advancing health equity for Alaskans impacted by stroke.² As part of its launch, 6 hospitals across the state were given access to RapidAI’s core stroke imaging and workflow products to help overcome roadblocks associated with rural care. Following this announcement, NeurologyLive® pieced together a long-form feature highlighting the impacts of this technology on a state like Alaska, which has some of the most diverse and challenging geographic landscape to provide care for.

Click here for more ISC 2025 coverage.

REFERENCES
1. Hudson A, Sachdev H, Marklein S, Ong K, Flores M. Detection of Large Vessel Occlusion Using AI: Evaluating RapidAI versus VizAI in Detection of LVO on CT Angiography in a Large Consecutive Patient Series of 1,591 Code Strokes (DUEL). Presented at: ISC 2025; February 5-7; Los Angeles, CA. ABSTRACT LBP68
2. RapidAI and Alaska Stroke Coalition join forces to overcome distance barriers and deliver equitable stroke treatment to alaskans. News release. RapidAI. September 4, 2024. Accessed November 21, 2024. https://www.rapidai.com/press-release/rapidai-asc-join-forces-to-overcome-distance-and-deliver-equitable-care