Qure.ai launches AI platform to identify brain bleeds and fractures

By Laura Lovett
Share

This morning healthcare AI company Qure.ai announced the launch of its new AI technology which can identify bleeds, fractures, and other critical abnormalities in the head CT scans. 

“Qure.ai’s new head CT scan technology rapidly screens scans in under 10 seconds to detect, localize, and quantify abnormalities, as well as assess their severity,” Prashant Warier, cofounder and CEO of Qure.ai, said in a statement. “This enables patient prioritization and the appropriate clinical intervention.”

The technology can generate an automated report that pinpoints the anatomical locations of lesions. The reports are aimed to help providers prioritize care, make better and faster decisions, and reduce costs, according to a statement. 

The launch comes after the results of a clinical validation study, which found that the technology was 95 percent accurate in identifying clinical abnormalities. A writeup of the trial, authored by researchers from the Mayo Clinic and Qure.ai, was recently released online.

“This study demonstrates that deep learning algorithms can accurately identify head CT scan abnormalities requiring urgent attention,” the authors of the paper wrote. “This opens up the possibility to use these algorithms to automate the triage process. They may also provide a lower bound for quality and consistency of radiological interpretation.”

The technology was trained on 313,318 anonymized CT scans and their accompanying clinical reports, according to the research paper. To validate the algorithm, the researchers compared the algorithm's readings  of 491 CT scans readings to those of three trained radiologists. The radiologists — each of whom held between eight and 20 years of experience, and one of whom was a study author —  were asked to examine each scan for certain types of hemorrhages, midline shift and mass effect, and fractures.

Results showed that Qure.ai’s product achieved an area under the curve of .9 or higher in detecting intracranial, intraventricular, subdural, etradural, and subarachnoid hemorrhages. It also proved capable of detecting calvarial fractures, as well as midline shifts and mass effect. 

“We are delivering near-radiologist accurate AI to support radiologists, physicians, and healthcare providers,” Sasank Chilamkurthy, an AI scientist at Qure.ai, said in a statement. “Our deep learning algorithms can accurately detect and highlight head CT scan abnormalities, reducing the chances of missing a diagnosis. Our technology can also localize the brain regions affected and quantify the bleed regions in a fully-automated report.”