FDA clears Viz.ai's algorithm to spot internal bleeds in brain scans

In the span of only about four years, Viz.ai has quickly racked up an impressive list of FDA permissions for its artificial intelligence algorithms designed to spot anomalies in emergency medical scans.

Lucky No. 7 arrived just last week, when the agency greenlighted Viz.ai’s latest AI tool: an algorithm that aims to detect both acute and chronic subdural hematomas in CT images of the brain.

Also known as subdural hemorrhage, the condition is typically caused by a head injury that bursts blood vessels, creating a pool of blood beneath the dura, the layer of protective tissue between the skull and brain. Hematomas can range in severity, from small and symptomless bleeds that only require observation to larger ones that lead to symptoms like headache, confusion and vomiting and may need to be treated with surgery.

According to data cited by Viz.ai, subdural hematoma is expected to become the most common neurosurgical diagnosis by the end of the decade.

If those projections come to fruition, the newly cleared Viz SDH algorithm could prove a handy tool for physicians hoping to quickly diagnose and begin treating subdural hematomas before they progress too far, potentially leading to coma, paralysis or death.

Viz SDH analyzes CT scans to identify suspected bleeds. It also automatically outlines the area affected by the possible hematoma and calculates its maximum thickness.

In a multi-center study that spanned more than 500 participants, the AI achieved a sensitivity of 94% and specificity of 92% in spotting subdural hemorrhages, according to its San Francisco-based maker.

“Subdural hemorrhages are growing in commonality, but can present different levels of urgency with different clinical pathways,” said Jayme Strauss, Viz.ai’s chief clinical officer. “Viz SDH supports physicians by detecting the presence of subdural hemorrhage and expediting communications and image sharing to improve the clinician workflow and more efficiently and effectively treat patients experiencing subdural hemorrhages.”

Strauss continued, “The algorithm is very sensitive and specific, significantly increasing the number of subdural hemorrhages detected and ensuring patients receive the necessary follow-up from this potentially life-threatening disease.”

The FDA clearance for Viz SDH comes just a few months after Viz.ai’s last agency nod. Handed down in February, it went to Viz Aneurysm, which uses AI to single out suspected cerebral aneurysms in CT angiograms.

Those two algorithms join a handful of others already authorized for clinical use. The first to be cleared, designed to spot large vessel occlusion strokes, came in 2018, then was followed by AI tools for CT perfusion, intracranial hemorrhage, pulmonary embolism and aortic disease.