Research & Innovation

Developing a Clearer Vision of Biomedical Imagery

Through an NSF CAREER grant, this Stevens researcher is using artificial intelligence to deliver biomedical images at a super-resolution, faster speed and lower cost so clinicians can more accurately diagnose and treat diseases

Biomedical imaging technology has come a long way since the 1890s, when X-ray imaging techniques first gave clinicians revolutionary new ways to see inside the body. However, today’s advanced technology still has limitations in quality, speed and affordability that can impact medical professionals’ ability to properly diagnose and treat diseases. 

Yu Gan, an assistant professor in the Department of Biomedical Engineering at Stevens Institute of Technology, aims to shatter those limitations with his groundbreaking research. Most recently, he was granted $600,000 through the National Science Foundation’s CAREER award program for his five-year study on “Developing Algorithms for Object-Adaptive Super-Resolution in Biomedical Imaging.”

The project focuses on delivering a triple play of higher quality, higher speed and lower cost for a diverse range of imaging systems, such as optical coherence tomography (OCT), histological microscopy, confocal images, MRI and ultrasound. 

“As you acquire more pixels and more data, speed is generally decreased,” Gan said. “We're using artificial intelligence tools to break that balance, developing a more intelligent sampling strategy that combines object detection and super-resolution to clearly and quickly deliver deeper insights that meet clinical needs on a variety of diagnostic modalities.”

This research partially builds on Gan’s 2020-22 National Science Foundation Computer and Information Science and Engineering (CISE) Research Initiation Initiative (CRII) award for “CRII:SCH: A Generative Deep Learning (GDL) based Platform for Super-resolution, Virtual-Pathological Visualization of Coronary Images.” The CRII  grant was more specifically directed at improving OCT images for the diagnosis and treatment of coronary artery disease at an arbitrary magnification.

“The data from the CRII study and feedback from our collaborators at the Heersink School of Medicine at the University of Alabama at Birmingham all support our new work to build a super-resolution system that’s more intelligent and more powerful,” he said. “One of the biggest improvements we expect to make is multiple magnification, and, of course, we’re working to bring this beyond a single method or disease state to the broader context of general biomedical imagery.”

This spring, Gan launched a graduate-level Stevens course, “Machine Learning in Biomedical Engineering,” in which he shares his research into machine learning and artificial intelligence for biomedical engineering applications. Through his latest CAREER grant, he’s expanding that educational outreach component of his work. In addition to having Ph.D. students on his research team, Gan is bringing his expertise and enthusiasm to local middle schools with high-tech demonstrations designed to help inspire the next generation of researchers to explore artificial intelligence and its potential to improve human lives.

“I'm excited about the vision that this CAREER grant can bring about to more quickly make medical images with more details at a lower expenditure,” he said. “It will save clinicians a lot of work and save patients financial stress. It will give them an interface that integrates multiple imaging systems. And, as a result, it will help them with analysis, diagnosis and treatment so they can save more lives.”

Learn more about academic programs and research in the Department of Biomedical Engineering: