Brain fluid dynamics: In vivo measurements and data-driven modeling
University of Rochester
Abstract
The glymphatic system circulates water-like fluid around and through brain tissue during sleep, clearing metabolic wastes whose accumulation might otherwise cause diseases like Alzheimer’s and Parkinson’s. However, because the modern glymphatic model was formulated just a decade ago, many fluid dynamical questions remain open. Working closely with neuroscientists, my team and I make in vivo measurements of flow velocities and of the shapes, sizes, and pulsations of fluid passageways in the brain. Combining measurements with machine learning, we infer flow rates, shear stresses, and pressure gradients. We have learned that motion of adjacent vessel walls is a key flow driver, that fluid passageways seem to be shaped to minimize their hydraulic resistance, that vortices sometimes appear, that small pressure gradients can drive substantial flow, and that contrast-enhanced magnetic resonance imaging (DCE-MRI) may be combined with machine learning to measure flows deep in the brain. I will present those findings and close with a discussion of some important open questions in brain fluid dynamics.
About the speaker
Dr. Douglas H. Kelley is Professor of Mechanical Engineering at the University of Rochester. He and his group study biophysical and magnetohydrodynamic fluid mixing, especially cerebrospinal fluid flow in the brain and liquid metal flow in technology and planets. His studies of brain fluid flow have implications for Alzheimer’s disease, traumatic brain injury, stroke, and related pathologies that involve brain waste clearance or swelling. Doug earned a Ph.D. from University of Maryland and did postdoctoral research at Yale University and MIT. He won a National Science Foundation CAREER Award, the University of Rochester’s David T. Kearns Faculty Teaching and Mentoring Award, and the Hajim School of Engineering’s Edmund A. Hajim Outstanding Faculty Award. He is a Moore Foundation Experimental Physics Investigator.

