News Roundup: Using Light and AI to Improve Stroke Treatment

Researcher Yihao Zheng is shining a light on blood clots to improve treatment for strokes caused by blockages in the brain.

Zheng, an assistant professor in the Department of Mechanical and Materials Engineering, and a team of researchers with expertise across science and engineering are developing a fiber-optic probe that will use light and advanced calculations to determine in real time whether a blood clot in the brain is soft, stiff, sturdy, or weak. The technology will give doctors information to guide decisions about how best to remove the blockages during clot-removal procedures.

“Physicians typically treat blockages in the brain by inserting a long, flexible tube into a patient’s leg, guiding it through arteries to the site of a clot, and then using suction or mesh tools to capture and remove the clot,” Zheng says. “Too often, these procedures, known as thrombectomies, fail on the first attempt and may even cause complications if clots shatter and send pieces further into the brain. Our research suggests that better information about clots and how they react to mechanical forces would provide critical insights to doctors and streamline these procedures.”

This four-year project is being funded with a $1.1 million grant from the National Science Foundation. Zheng will serve as principal investigator (PI), and Yuxiang Liu, associate professor of mechanical and materials engineering, will be a co-PI. Researchers at the University of California San Francisco, the University of Georgia, and UMass Chan Medical School will collaborate on the project.

The researchers will develop a slender fiber-optic probe that can be guided through a catheter to a blood clot in a patient’s brain. Their prototype probe measures about 1 millimeter in diameter and is much thinner than current commercial products that deploy similar technology. The probe will emit light and gather data on the clot’s chemical fingerprint, using a method called Raman spectroscopy. Specifically, the probe will gather data on blood components involved in clot formation—fibrin, platelets, and red blood cells.

A type of artificial intelligence known as a convolutional neural network will read the light signals to model how the clot would respond to mechanical forces such as compression and tension. Those predictions could help physicians determine the best way to capture and remove the clot. To validate their device, the researchers will test it in laboratory experiments and models.

“Stroke is the third-leading cause of death worldwide and a leading cause of disability, and it poses a serious problem as our population ages,” Zheng says. “By bringing together researchers with expertise in engineering, physics, data science, and medicine, we are developing a technology that can address the urgent need for a solution.”