A groundbreaking AI system developed by researchers at the University of Michigan can analyze brain MRI scans in a matter of seconds, effectively identifying urgent medical conditions. This innovative technology, named Prima, demonstrates an accuracy rate of up to 97.5% in diagnosing various neurological disorders, surpassing existing advanced AI tools. The findings were published on February 10, 2026, in the journal Nature Biomedical Engineering.
The urgency of improving brain imaging processes has been underscored by the increasing global demand for MRI scans. As healthcare systems face significant pressure, the need for rapid and precise diagnostics has never been more critical. According to Todd Hollon, M.D., a neurosurgeon at University of Michigan Health and the senior author of the study, “Our AI model has the potential to reduce burden by improving diagnosis and treatment with fast, accurate information.”
Testing and Performance of the Prima System
The evaluation of Prima involved over 30,000 MRI studies over the course of a year. The system was trained on an extensive dataset comprising more than 200,000 MRI scans and 5.6 million imaging sequences, integrating both clinical histories and the reasons for imaging requests. This comprehensive approach enabled Prima to outperform other advanced models across more than 50 different radiologic diagnoses.
In instances where immediate medical intervention is necessary, such as strokes or brain hemorrhages, Prima can automatically alert healthcare providers. This feature is essential for ensuring timely action and prioritizing care for patients in critical condition.
Understanding the Technology Behind Prima
Prima is classified as a vision language model (VLM), a type of artificial intelligence capable of processing images, video, and text in real time. Unlike previous models, which were typically limited to specific tasks like identifying lesions, Prima was developed using a more comprehensive dataset. This broad training allows it to function similarly to a radiologist by synthesizing patient information and imaging data for a holistic view of health.
Co-first author Samir Harake, a data scientist in Hollon’s lab, emphasized that “Prima works like a radiologist by integrating information regarding the patient’s medical history and imaging data to produce a comprehensive understanding of their health.” This capability enhances performance across a wide range of diagnostic tasks.
As healthcare systems face challenges such as staffing shortages and diagnostic delays, the introduction of innovative technologies like Prima has the potential to significantly improve access to radiology services. Vikas Gulani, M.D., Ph.D., chair of the Department of Radiology at U-M Health, noted the need for solutions that can address the growing demand for MRI scans worldwide.
The research team is aware that while Prima shows promising results, it is still in the early evaluation phase. Future studies will focus on incorporating more detailed patient information and electronic medical record data to refine diagnostic accuracy further.
Hollon likened Prima to “ChatGPT for medical imaging,” suggesting that similar AI technologies could eventually be adapted for other imaging modalities, such as mammograms and chest X-rays. He stated, “We believe that Prima exemplifies the transformative potential of integrating health systems and AI-driven models to improve healthcare through innovation.”
This pioneering work received support from the National Institute of Neurological Disorders and Stroke and various foundations, highlighting a collaborative effort to advance medical imaging technology. The results not only promise to enhance the efficiency of MRI analysis but also aim to improve patient outcomes through timely and accurate diagnoses.
As AI continues to evolve within the healthcare sector, initiatives like Prima could redefine the landscape of medical imaging, offering hope for enhanced diagnostic capabilities and better patient care.
