Using AI to detect cancer
New research from a University of Minnesota Rochester alum is using technology to analyze potential melanoma cases one pixel at a time.
In medicine, early detection can mean the difference between life and death, and an alumnus from the University of Minnesota Rochester (UMR) is pioneering an innovative artificial intelligence (AI) model designed to detect melanoma. Led by Quincy Gu (who earned a PhD from UMR, and is currently an assistant professor at the University of Pittsburgh), this groundbreaking research holds the promise of revolutionizing skin cancer diagnostics, offering hope not just for Minnesotans but for people around the globe.
Melanoma, a type of skin cancer that arises from melanocytes, is one of the fastest-growing cancers in the United States. According to the American Cancer Society, about 100,000 new cases of melanoma are expected annually, and early detection is critical for successful treatment. The stakes are high: melanoma is the deadliest skin cancer, but when detected early, it is highly curable.
To diagnose melanoma, pathologists examine microscope slides of biopsied skin lesions. But pathologists can make mistakes.
To reduce these errors, the AI model developed by Gu is trained to understand how normal tissue structure varies from abnormal tissue. The technology can examine a whole slide image of the digitized microscope slide and rate each pixel as “melanoma” or “non-melanoma.” A pathologist can then give any “melanoma-rated” areas special attention before making a diagnosis.
His model has an 89 percent chance of successfully identifying a melanoma or non-melanoma pixel, which compares favorably with the track record of pathologists, Gu says.
AI generated images of normal skin patches
Scans with AI annotations of abnormalities
Since the model’s success is similar to that of a pathologist, the physician could use Gu’s AI model to examine the slides that had been evaluated as melanoma. If they agreed with its assessment, they could make a diagnosis earlier, using the extra time to more closely scrutinize the relatively few slides where the model may have missed a cancer.
“AI has the potential to transform healthcare,” Gu says. “It can make the process of diagnosing something like melanoma much quicker, which could end up saving lives.”
This research is particularly important for Minnesota, a state known for its outdoor lifestyle, which increases the risk of skin cancer due to higher sun exposure. Rates of melanoma are growing in Minnesota and the state has the third-highest rate of the disease, according to the National Cancer Institute.
A perfect fit
Gu experienced serious health issues before he started his undergraduate studies at the University of Minnesota as an international student in 2016.
“That was my motivation to pursue a career in medicine,” he says. “My health issues were initially misdiagnosed, and I don’t want that experience to happen to anyone else.”
Gu began his academic journey at the Twin Cities campus, earning a bachelor's degree in mathematics. He then continued his education by pursuing a PhD at the University of Minnesota Rochester in the Bioinformatics and Computational Biology (BICB) program. The BICB program provides interdisciplinary education in biomedical informatics and computational biology at the interface of quantitative sciences, medicine, and biology. The PhD program trains graduate students in the development and applications of computational methods and to work in interdisciplinary teams of life scientists and computational scientists.
"UMR students, just like Quincy, have extraordinary collaboration opportunities in America's City for Health, Rochester, Minnesota, as they work with others to design solutions to the grand health challenges of the 21st century.”
Lori J. Carrell
Chancellor, University of Minnesota Rochester
A good use of AI
UMR’s commitment to health sciences and Gu’s ability to use his mathematics degree and his interest in computer science made the program a natural fit for him.
Gu’s research, and others like it going on at UMR, could have a profound effect on healthcare going forward. This work is not only advancing the capabilities of AI in healthcare, but is also laying the groundwork for a future where early detection of skin cancer becomes the norm. By harnessing AI technology, doctors can improve diagnostic accuracy, enhance patient outcomes, and ultimately save lives.
“There is a lot of fear about AI out there, but it shouldn’t be scary,” Gu says. “It’s all about how you use it. This technology can do a lot of good things for you. In the case of our work, AI can dramatically change the healthcare industry.”