A new artificial intelligence system, named Prima, developed at the University of Michigan, can diagnose neurological conditions from brain MRI scans within seconds.
The system achieved 97.5% accuracy in identifying neurological conditions and could also assess the urgency of medical care required by patients. This technology aims to redefine brain imaging practices within U.S. health systems.
The findings regarding Prima were published in the journal Nature Biomedical Engineering.
Dr. Todd Hollon, a neurosurgeon at University of Michigan Health and assistant professor of neurosurgery at U-M Medical School, led the research team. Dr. Hollon is also a senior author of the study.
Over a one-year period, Dr. Hollon’s team evaluated Prima using more than 30,000 MRI studies. The system demonstrated superior diagnostic performance compared to other advanced AI models across over 50 different radiologic diagnoses of major neurological disorders.
Prima can identify conditions such as strokes and brain hemorrhages, which require immediate medical attention. In these instances, the system can automatically alert healthcare providers, including subspecialists like stroke neurologists or neurosurgeons, for rapid intervention. Feedback becomes available immediately after a patient completes imaging.
Yiwei Lyu, a co-first author and postdoctoral fellow of Computer Science and Engineering at U-M, noted that timely diagnosis is critical for improved patient outcomes.
Prima is classified as a vision language model (VLM), capable of processing images, video, and text in real time. Unlike earlier AI models trained on limited MRI subsets for narrow tasks, Prima was trained on a broad dataset encompassing over 200,000 MRI studies and 5.6 million imaging sequences collected since radiology records were digitized at University of Michigan Health.
The model also incorporated patients’ clinical histories and the reasons physicians ordered each imaging study.
Samir Harake, a co-first author and data scientist in Hollon’s Machine Learning in Neurosurgery Lab, explained that Prima integrates patient medical history and imaging data, similar to a radiologist, to provide a comprehensive understanding of a patient’s health.
Millions of MRI scans are performed globally each year, with a rising demand for neuroradiology services. This increasing demand, coupled with shortages in neuroradiology staff, contributes to diagnostic delays and errors. Patients often experience wait times of days or longer for MRI results.
Dr. Vikas Gulani, co-author and chair of the Department of Radiology at U-M Health, stated that innovative technologies are necessary to improve access to radiology services in both large health systems and rural hospitals with limited resources.
Researchers consider Prima to be in an early evaluation phase. Future research will focus on integrating more detailed patient information and electronic medical record data to enhance diagnostic accuracy.
This approach mirrors how radiologists and physicians interpret medical imaging studies in clinical settings. Dr. Hollon compares Prima to “ChatGPT for medical imaging,” suggesting similar technology could be adapted for other imaging types, including mammograms, chest X-rays, and ultrasounds.
