The development of new treatments for diseases like Alzheimer’s could be significantly impacted by a new artificial intelligence-based method for measuring fluid flow around the blood vessels in the brain.

The cerebral blood vessels are surrounded by perivascular spaces that assist in the removal of waste and transport water-like fluids throughout the brain. Although they are difficult to measure in vivo, changes in the fluid flow have been linked to neurological conditions like Alzheimer’s, small vessel disease, strokes, and traumatic brain injuries.

New AI velocimetry measurements for accurately calculating brain fluid flow were developed by a multidisciplinary group of mechanical engineers, neuroscientists, and computer scientists led by Associate Professor Douglas Kelley of the University of Rochester. A study that was published in the Proceedings of the National Academy of Sciences provides an overview of the outcomes.

Kelley, a professor in the University of Rochester’s Department of Mechanical Engineering, explains, “In this study, we combined some measurements from inside the animal models with a novel AI technique that allowed us to effectively measure things that nobody’s ever been able to measure before.”

Maiken Nedergaard, coauthor of the study and codirector of Rochester’s Center for Translational Neuromedicine, led years of experiments that served as the foundation for this work. By injecting tiny particles into the fluid and measuring their position and velocity over time, the group has previously been able to conduct two-dimensional studies of the fluid flow in perivascular spaces. However, researchers required more perplexing estimations to grasp the full multifaceted design of the framework – – and investigating such an indispensable, liquid framework is a test.

The team worked with George Karniadakis from Brown University to use artificial intelligence to solve that problem. They created unprecedented high-resolution views of the system by combining the existing 2D data with physics-informed neural networks.

According to Kelley, “this is a way to reveal pressures, forces, and the three-dimensional flow rate with much more accuracy than we can do otherwise.” The pumping mechanism that drives all of these brain flows is still unknown, so the pressure is important. This is a brand-new area.”

The researchers received assistance from the Army Research Office’s Multidisciplinary University Research Initiatives program, the National Institutes of Health Brain Initiative, and the Collaborative Research in Computational Neuroscience program.