Today we launched a new AI-powered technology to accurately identify bleeds, fractures and other critical abnormalities in head CT scans. We have also released the results of an unprecedented clinical validation study confirming the algorithms’ near-radiologist performance on 21,000 patients and made a dataset of almost 500 AI-analyzed head CT scans available for download.

We trained the new AI using a collection of 313,318 anonymized head CT scans, along with their corresponding clinical reports. Of these, 21,095 scans were then used to validate the algorithms. Finally, the AI was clinically validated on 491 CT scans, with the results compared against a panel of three senior radiologists. The panel of radiologists included Norbert G. Campeau, M.D., a senior neuro-radiologist from the Mayo Clinic’s Department of Radiology. The validation study found that our AI was more than 95% accurate in identifying abnormalities.

In addition to the study, We have made a dataset of 491 AI-interpreted head CT scans, as well as the corresponding interpretations from the three radiologists, publicly available for download. This dataset is from the Centre for Advanced Research in Imaging, Neurosciences and Genomics, and includes both out-patient and in-patient scans from 7 centers. To download the full dataset, visit: http://headctstudy.qure.ai.