Somerville, Mass.-based Mass General Brigham has developed a new AI foundation model that can extract disease risk signals from routine brain MRIs.
The model was trained on nearly 49,000 brain MRI scans and outperformed task-specific AI models, according to a Feb. 5 news release. Researchers said it uses self-supervised learning to identify inherent features from unlabeled datasets. They validated the framework on 48,965 diverse scans across seven tasks.
Investigators also said the tool could estimate “brain age,” predict dementia risk, detect brain tumor mutations and predict survival from brain cancer. Results were published Feb. 5 in Nature Neuroscience.
Moreover, the model could accelerate biomarker discovery and support clinical adoption of AI. The study received support from the National Institutes of Health and the National Cancer Institute.
