When Maiken Nedergaard first described the glymphatic system in 2012, she revealed one of the brain's most elegant housekeeping mechanisms: during deep sleep, waterlike fluid washes away metabolic waste linked to Alzheimer's and other neurodegenerative diseases. But more than a decade later, a fundamental question remained unanswered—how fast does this fluid actually move through the brain? The answer, which researchers at the University of Rochester have now mapped using artificial intelligence, turns out to be more complex and revealing than anyone expected.
The challenge of studying the glymphatic system in a living brain is formidable. Traditional microscopy can capture exquisite detail, but only from tiny patches of tissue. MRI scans offer a whole-brain view, yet they cannot measure the slow fluid flows that scientists need to understand. This impasse is precisely why Professor Douglas Kelley from the University of Rochester's Department of Mechanical Engineering and his collaborators turned to physics-informed artificial intelligence. Working with colleagues from Brown University and the University of Copenhagen, they built neural networks trained on videos of dye spreading across brain tissue over time, allowing the AI to deduce both fluid flow velocity and tissue permeability from standard MRI data.
The results, published in Science Advances, revealed something striking: the glymphatic system operates at two distinctly different speeds. Around the brain's surface—in the open regions between the skull and brain—fluid moves at a few microns per second. But as that same fluid penetrates deeper into the brain's tissue, it slows dramatically, trickling through at a rate approximately 50 times slower than the surface flow. This two-speed system suggests that the brain's waste-clearing mechanism is far more nuanced than previously understood, with different regions relying on different circulation patterns to flush away particles like amyloid beta, the protein implicated in Alzheimer's disease.
So far, Kelley's team has been establishing baseline measurements in animal brains—primarily mice—to refine and validate their AI tools. But they are working toward something far more ambitious. The next step is comparing fluid flow between healthy and diseased brains, young and old brains. Eventually, they hope to study circulation in human subjects, opening doors to clinical applications that seemed distant just years ago.
"We're working hard toward being able to measure the flow of waterlike fluids in and around human brains because then the clinical applications get a lot more important and exciting," Kelley said. The possibilities are tantalizing: doctors could one day screen Alzheimer's patients to see whether they have poor brain circulation, or use early detection of impaired flow to intervene before symptoms emerge. The same approach could assess whether a concussion has disrupted a person's fluid circulation, offering a new diagnostic window into brain injury.
This work represents a crucial intersection of neuroscience and engineering—a reminder that understanding the brain's deepest mysteries sometimes requires stepping outside traditional laboratory methods. By harnessing AI to unlock secrets hidden in MRI data, Kelley and his colleagues have moved us closer to a future where maintaining the brain's waste-clearing system might become as routine and preventive as any other aspect of medical care.