University of Michigan Develops Prima AI for MRI

The University of Michigan has developed a new artificial intelligence system, named Prima, capable of analyzing brain MRI scans and delivering diagnoses with up to 97.5% accuracy in seconds. This vision-language model was trained on over 200,000 MRI studies and 5.6 million imaging sequences from University of Michigan Health, integrating patient clinical histories. Prima can also assess the urgency of medical care needed, automatically alerting healthcare providers for critical conditions like strokes. This technology, detailed in Nature Biomedical Engineering, aims to address the growing demand for MRI scans and the shortage of neuroradiologists, potentially reshaping neurological care in the United States. Researchers, including senior author Todd Hollon, envision Prima as a 'co-pilot' for radiologists, with potential applications for other imaging types.