New AI foundation model offers wide adaptability across neuroradiology
Experts at Mass General Brigham have developed an artificial intelligence tool they believe has broad applicability in neuroimaging, with the potential to help providers personalize treatment pathways.
The brain imaging adaptive core—BrainIAC for short—is a foundation model that was designed to analyze large MRI datasets using self-supervised learning techniques. This enables it to spot patterns and features in unlabeled datasets, which in turn makes the model more adaptable. Unlike most AI tools that were built with specific goals, like identifying stroke, BrainIAC can be used for a wide array of applications, as it can learn from other AI frameworks.
“The [AI] field has been constrained, thus far, by limited training data and task-specific models that do not generalize well across patient populations and medical tasks,” corresponding author Benjamin Kann, MD, of the Artificial Intelligence in Medicine (AIM) Program at Mass General, and colleagues wrote in Nature Neuroscience. “By leveraging self-supervised learning, pretraining and targeted adaptation, foundation models present a promising paradigm to overcome these limitations.”
BrainIAC was trained and validated on nearly 49,000 MRIs across seven different tasks that varied in complexity. This revealed the model to be more accurate than task-specific alternatives included in its training. It performed well across all tasks, both simple and complicated, even accurately specifying brain tumor mutation types.
What’s more, the model maintained its performance with minimal training data. This, the authors contend, suggests it could hold up in real world settings in the absence of annotated datasets.
“BrainIAC has the potential to accelerate biomarker discovery, enhance diagnostic tools and speed the adoption of AI in clinical practice,” the team suggested. “Integrating BrainIAC into imaging protocols could help clinicians better personalize and improve patient care.”
Read more here.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.