AI algorithm enables tracking of vital white matter pathways (news.mit.edu)

Researchers from MIT, Harvard University, and Massachusetts General Hospital have developed an AI-powered tool named the BrainStem Bundle Tool (BSBT), capable of automatically segmenting eight distinct white matter bundles in the brainstem using diffusion MRI sequences. This advancement is significant as it addresses the long-standing challenge of imaging the brainstem, a critical region that regulates essential physiological functions such as breathing and heart rate. The BSBT allows for unprecedented observation of brainstem changes in various neurodegenerative diseases, including Parkinson’s, multiple sclerosis, and Alzheimer’s, providing crucial insights into disease progression and recovery. The algorithm employs a convolutional neural network to generate a probabilistic fiber map based on diffusion MRI scans, which helps to visualize the directionality of water diffusion along myelinated axons. Through rigorous training and validation against post-mortem dissections, BSBT achieved reliable identification of neural bundles, demonstrating superior accuracy compared to traditional classification methods. This capability not only enhances diagnostic imaging but also holds promise for developing novel biomarkers by tracking changes in bundle volume and structural integrity during disease progression. Notably, BSBT has already been employed to observe recovery in a traumatic brain injury patient, showcasing its potential in prognostic assessments and facilitating a deeper understanding of brainstem-related pathologies.