Summary: A multidisciplinary neuroengineering team has overcome a major imaging limitation by mapping the precise flow velocities of the brain’s waste-clearance network. Using a physics-informed artificial intelligence approach to decode magnetic resonance imaging (MRI) data, the study reveals the mechanics of the glymphatic system, the fluid pathway that clears metabolic waste—including amyloid-beta proteins implicated in Alzheimer’s disease.
Their AI models discovered a two-speed drainage pattern: fluid moves roughly fifty times faster across the brain’s outer surfaces than it does through the deep tissue of the brain.
Key Facts
- The Glymphatic Baseline: First described in 2012 by neuroscientist Maiken Nedergaard, the glymphatic system functions like the brain’s plumbing. During deep sleep, a waterlike fluid circulates through and around the central nervous system to remove metabolic debris associated with neurodegenerative diseases.
- The MRI Velocity Limitation: Measuring this circulation in a living brain has been difficult. Microscopes can resolve tiny regions with great detail but cannot capture whole-brain dynamics, while conventional 3D MRI lacks the sensitivity to measure the extremely slow flow velocities involved.
- Physics-Informed Neural Networks: To bridge that gap, mechanical engineers and computational scientists developed specialized, physics-informed AI. By training neural networks on time-series MRI videos of dye moving through brain tissue, the AI reconstructed fluid velocities and estimated tissue permeability.
- The Dual-Velocity Blueprint: The study shows the glymphatic system uses two distinct modes to clear particles such as amyloid-beta. A fast advective flow moves fluid at a few micrometers per second across open cortical regions (for example, between the skull and brain surface). A slower, diffusion-dominated flow moves through deep brain tissue at roughly fifty times slower velocities.
- Diagnostic Roadmap: Researchers have established baseline flow measurements in animal models and are refining the AI for human clinical use. The aim is to compare fluid dynamics across ages and disease states to better understand how clearance changes with aging and pathology.
- Clinical Implications: According to senior author Professor Douglas Kelley, accurate brain fluid mapping could help screen for impaired circulation linked to Alzheimer’s and assess whether concussions disrupt glymphatic flow, enabling earlier intervention and more personalized care.
Source: University of Rochester
How sleep supports brain clearance
When a person enters deep sleep, a waterlike fluid circulates around the brain and spinal cord, flushing away metabolic waste that has been associated with neurodegenerative disorders. This clearance mechanism, known as the glymphatic system, was first characterized in 2012 by Maiken Nedergaard and has since been a focal point for understanding how the brain maintains homeostasis.
“A microscope can show what’s happening in a very small patch of tissue, but it cannot capture whole-brain behavior,” explains Professor Douglas Kelley from the Department of Mechanical Engineering at the University of Rochester. “MRI provides a three-dimensional view of the entire brain but traditionally cannot resolve the extremely slow velocities of glymphatic flow. Combining MRI with physics-informed AI allows us to extract that velocity information from dynamic imaging data.”
In the published study, the team presents MR-AIV (magnetic resonance artificial intelligence velocimetry), a framework built with a physics-informed neural architecture and an optimization method that reconstructs three-dimensional fluid velocity fields from dynamic contrast-enhanced MRI (DCE-MRI). MR-AIV yields brain-wide velocity maps and provides estimates of tissue permeability and pressure fields—quantities that were previously out of reach with standard imaging or analysis techniques.
Applied to animal models, MR-AIV quantifies a landscape of interstitial and perivascular flows, separating slow diffusion-driven transport on the order of 0.1 micrometers per second from faster advective flow near 3 micrometers per second. These measurements define a functional map of clearance pathways and help explain how the brain removes potentially harmful proteins and metabolites.
The research team is using animal baseline data to train and validate the AI tools. Their near-term goal is to adapt MR-AIV for human clinical studies so clinicians can compare fluid dynamics in young versus old or healthy versus diseased brains. In clinical practice, such measurements could inform early screening for Alzheimer’s risk and provide a rapid, noninvasive assessment following head trauma to determine whether glymphatic circulation has been compromised.
Funding: The research was supported by the NIH National Center for Complementary and Integrative Health and the NIH BRAIN Initiative.
Collaborators on this project include Juan Diego Toscano, Yisen Guo, Zhibo Wang, Mohammad Vaezi, Yuki Mori, George Em Karniadakis, Kimberly A. S. Boster, and Douglas H. Kelley.
Key Questions Answered:
A: The brain continuously generates metabolic waste as it performs its functions. During deep sleep, the glymphatic system becomes more active, allowing a waterlike fluid to move through and around neural tissue to clear toxic materials—including amyloid-beta proteins—that are linked to Alzheimer’s disease. This process helps maintain brain health and homeostasis.
A: Standard MRI gives detailed three-dimensional images of brain anatomy but lacks the sensitivity to measure the extremely slow velocities of glymphatic flow. By combining dynamic MRI recordings of dye transport with a physics-informed neural network, researchers can infer flow velocities and tissue permeability from patterns in the imaging data—information that conventional MRI analysis cannot provide.
A: Concussive impacts can disrupt the delicate pathways of the glymphatic system, impairing the brain’s ability to clear waste. AI-driven fluid mapping could offer a rapid, noninvasive assessment of whether a patient’s internal fluid circulation has been compromised, guiding safer, individualized recovery plans and monitoring rehabilitation progress.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context provided by staff.
About this neuroscience and AI research news
Author: Luke Auburn
Source: University of Rochester
Contact: Luke Auburn – University of Rochester
Image credit: Neuroscience News
Original Research: Open access. “MR-AIV reveals in vivo brain-wide fluid flow with physics-informed AI” by Juan Diego Toscano, Yisen Guo, Zhibo Wang, Mohammad Vaezi, Yuki Mori, George Em Karniadakis, Kimberly A. S. Boster, and Douglas H. Kelley. Science Advances. DOI: 10.1126/sciadv.aeb0404
Abstract
MR-AIV reveals in vivo brain-wide fluid flow with physics-informed AI
The circulation of cerebrospinal and interstitial fluids plays a critical role in removing metabolic waste from the brain, and disruptions to these flows have been associated with neurological disorders. Directly measuring brain-wide fluid transport—particularly in deep tissue—has been difficult to achieve in vivo. MR-AIV introduces a physics-informed AI framework that reconstructs three-dimensional velocity fields from dynamic contrast-enhanced MRI. MR-AIV produces brain-wide velocity maps and yields estimates of tissue permeability and pressure fields, enabling quantitative separation of slow diffusion-driven transport (approximately 0.1 μm/s) from faster advective flow (approximately 3 μm/s). This approach opens new avenues for investigating brain clearance mechanisms and fluid dynamics in health and disease, and it may extend to other porous-media systems from tissue mechanics to geophysics.