Radiology · Care-coordination & triage platform
Viz.ai
Viz.ai
Viz.ai's ContaCT was the first AI platform the FDA authorised for large-vessel-occlusion triage, and the De Novo it received in February 2018 established the regulatory category that the entire computer-aided-triage field now occupies. A decade later the platform spans more than ten FDA authorisations across stroke, vascular, and cardiac pathways, carries a CE mark in Europe, and is supported by an unusually deep real-world literature — including a 2025 systematic review and meta-analysis pooling 15,595 patients. But the evidence is observational and workflow-focused rather than randomised, and a third-jurisdiction approval beyond FDA and CE could not be independently verified, which places the platform at two marks rather than three.
Performance Metrics
Clinical Evidence
Viz.ai's evidence base is among the largest in acute-stroke AI, though its centre of gravity is care-pathway impact rather than standalone diagnostic accuracy — a distinction procurement reviewers should weigh. The pivotal validation supporting the ContaCT De Novo (DEN170073) drew on a multi-site dataset of consecutive patients and reported approximately 96% sensitivity and 94% specificity for large-vessel-occlusion identification on CT angiography. Independent real-world testing has tended to return lower sensitivity than the pivotal figures. In a single-centre evaluation at a comprehensive stroke centre (Yahav-Dovrat et al., American Journal of Neuroradiology 2021; 42(2):247–254), all 1,167 consecutive head-and-neck CTAs over roughly 14 months were processed by Viz LVO against senior neuroradiologist reports as ground truth: sensitivity was 0.81 with a negative predictive value of 0.99 and overall accuracy of 0.94, with reduced sensitivity for more distal (M2) occlusions — a recognised limit of CTA-based LVO detection generally. The platform's strongest claims concern workflow. An early hub-and-spoke analysis (Hassan et al., Interventional Neuroradiology 2020; 26(5):615–622) associated deployment with a roughly 66-minute reduction in the interval from CTA at a primary stroke centre to arrival at a comprehensive stroke centre, alongside a shorter neuro-ICU length of stay. A real-world series in transferred LVO patients (Morey et al., Cerebrovascular Diseases 2021; 50(4):450–455) reported faster door-to-notification times after implementation. A 2025 systematic review and meta-analysis (Translational Stroke Research 2025) pooled 12 studies and 15,595 patients and found consistent improvements in stroke workflow metrics — door-to-groin-puncture, CT-to-treatment, and door-in-door-out times — while noting that the magnitude of benefit depends materially on care-team engagement rather than the software alone. Beyond stroke, Viz.ai holds the first De Novo authorisation for a cardiovascular machine-learning notification device (Viz HCM, DEN230003), which screens routine 12-lead ECGs for signs associated with hypertrophic cardiomyopathy. Its clinical literature is younger than the stroke work and is still accumulating.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| ContaCT pivotal validation (FDA De Novo DEN170073) | RetrospectiveRetrospective | 2,544 | 96% | 94% | — | FDA De Novo decision summary, 2018 (multi-site LVO dataset) |
| Yahav-Dovrat A, Saban M, Eyal-Taraboulos T, et al. | RetrospectiveRetrospective | 1,167 | 81% | — | — | AJNR, Feb 2021 (42:247–254); NPV 0.99, accuracy 0.94 |
| Morey JR, Zhang X, Yaeger KA, et al. (transferred LVO) | RetrospectiveRetrospective | 55 | — | — | — | Cerebrovascular Diseases, 2021 (50:450–455); faster door-to-notification |
| Systematic review & meta-analysis (Viz.ai LVO workflow) | RetrospectiveRetrospective | 15,595 | — | — | — | Translational Stroke Research, 2025; 12 studies pooled |
Clinical Pulse
Verified clinician sentiment
Verified clinician reviews launching soon · Apply to contribute →
Reviews are signed by clinicians verified against their regulator's public register.
Inside the algorithm
Editorial featureHow Viz LVO reaches a notification.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
A standard CT angiogram is read in parallel to the radiologist.
The pathway begins with a routine study acquired for a suspected acute stroke. As the series leaves the scanner it is routed to Viz.ai's cloud service in parallel to the worklist — the study is never removed from the radiologist's standard reading queue.
No additional acquisition, contrast, or radiation is required. The platform is a -style parallel-workflow tool: it observes the same images the radiologist will read, at the same time.
Input
Head-and-neck CTA DICOM
Inference
Cloud, parallel workflow
Inside the Auris+ Listing
Five more sections complete this device’s Auris+ Listing.
Decision Ledger
ProPro unlocks a private, cross-vendor log of every case you read on this device — what the AI called, what you concluded, and a one-line reason if you overrode it. Ready for the EU AI Act's deployer logging obligations when they land in 2028.
Clinical Evidence Deep Dive
ProPro unlocks the structured clinical-evidence summary — study count, target patient population, and a tabular accuracy-metrics view drawn from peer-reviewed sources.
Peer-Reviewed Publications
ProPro unlocks the curated peer-reviewed publication list with PubMed cross-links — the citation backbone of every editorial verdict.
Post-Market & Regulatory Conditions
ProPro unlocks the post-market surveillance summary, recall record, and the conditions of approval that bound real-world use.
AI Algorithm Version History
ProPro unlocks the chronological record of algorithm version changes — what changed when, drawn from manufacturer changelogs and regulatory filings.
Regulatory Approvals
K250354
Class II
DEN230003
Class II
Safety Record
No Viz.ai device recalls or FDA safety communications were identified in publicly available sources as of the review date; the MAUDE adverse-event database could not be queried directly from this environment, so this should be read as "none found in public reporting" rather than an exhaustive MAUDE audit. The products are notification-only, parallel-workflow software — they do not alter image acquisition, remove studies from the standard reading queue, deliver therapy, or interact with implanted hardware, which limits the harm surface. The principal documented risk is reliance on a negative result: independent testing recorded LVO sensitivity around 81% with reduced sensitivity for distal occlusions, so the absence of an alert does not exclude an occlusion and does not substitute for clinician review.
Intended Use & Indications
Viz.ai's foundational product, cleared by the FDA as ContaCT (Viz LVO), analyses CT-angiography studies in parallel to the standard radiology workflow and, when an algorithm threshold for suspected large-vessel occlusion is met, sends a compressed preview image and a notification to a designated specialist's mobile device. It is a notification-only, parallel-workflow tool: it does not remove the study from the standard reading queue, mark or measure the image, or provide a diagnosis. Final diagnostic and treatment decisions remain with the treating clinician. Subsequent modules extend the same alert-and-coordinate pattern to CT perfusion (Viz CTP), intracranial and subdural haemorrhage (Viz ICH, Viz SDH/HDS), cerebral aneurysm (Viz ANEURYSM), pulmonary embolism and aortic disease, and — via De Novo authorisation — AI-enabled 12-lead ECG screening for signs associated with hypertrophic cardiomyopathy (Viz HCM). The HCM module analyses routine ECGs and flags suspected cases; it does not diagnose HCM and is not intended for patients with implanted pacemakers.