Strokes are the number one cause of death for Americans, according to the CDC. Responsible for 5% of the deaths each year, strokes kill more than 130,000 people annually in the U.S. Strokes aren’t all the same, however, and some types are more deadly than others. Large vein occlusion (LVO) strokes are thrombotic strokes that block large blood vessels in the brain; these are especially dangerous. As is the case with any severe medical condition, early diagnosis is crucial with strokes, and particularly with LVO strokes.

San Francisco-based startup Viz.AI sprung from neurovascular studies at Stanford University in 2015. Five controlled trials identified endovascular thrombectomy as the best way, the new gold standard, for treating LVO strokes. Endovascular thrombectomy is a procedure in which a surgeon mechanically removes a blood clot. Intravenous recombinant tissue plasminogen activator treatment, which involves using a protein to break down blood clots, is the current standard treatment for many types of strokes, but it’s too slow for LVO strokes. Unfortunately, most LVO stroke victims are not diagnosed in time for the faster blot removal procedure. Too often the result is that LOV stroke victims die or suffer greater damage than they would have with clot removal.

At the same time that researchers recognized endovascular thrombectomy as the best treatment for LVO strokes, advances in artificial intelligence, especially in deep learning, made it possible to interpret medical images much faster and more accurately than before. Viz.AI’s founders, neurosurgeon Chris Mansi and machine learning researcher David Golan, saw the opportunity to help LVO stroke victims with rapid AI-assisted image analysis to enable timely intervention with endovascular thrombectomy.

Viz.AI’s mission is “to bring the right patient to the right doctor at the right time.” The company’s signature product, Viz LVO, uses AI to triage patients by automatically detecting large vessel occlusion strokes to move the patients directly to stroke specialists. A second product, Viz CTP analyzes CAT (CT) perfusion images to generate perfusion color maps to identify LVO patients who can be saved. The approved guidelines call for mechanical thrombectomy in a short window of time from the onset of a stroke, just six to 24 hours, so the LVO diagnosis and perfusion mapping needs to take place quickly.

Viz LVO and Viz CTP have both been cleared by the FDA for LVO stroke identification, stroke triage, and notification. The FDA is also creating a framework from which to encourage the developers at Viz.AI  to expand their software functionalities to aid in diagnosing and treating additional diseases.