AI System Quickly, Accurately Diagnoses Cancer

With the growing use of artificial intelligence (AI) in medical applications, a team of scientists at the University of Waterloo, Ontario, Canada, have developed a system to quickly and accurately diagnose cancer. While advances in cancer detection and treatments have resulted in growing survival rates for women and men in the United States, mortality rates are still significantly less in developing or low- to middle-income countries.

As part of a five-year study, which received $3.2 million in funding from the Ontario Government, the Waterloo team turned to the National Cancer Institute’s public Cancer Genome Atlas Program to test their AI algorithm system to advance digital pathology in an effort to find enhanced solutions for earlier detection.

“AI can help us tap into our medical wisdom, which at the moment is just sitting in archives,” said Hamid Tizhoosh, director of the Laboratory for Knowledge Inference in Medical Image Analysis (KIMIA Lab) at Waterloo. “When you use AI like this, its performance is astounding.”

Hamid Tizhoosh and his team have developed an AI system to more quickly diagnose cancer. (Image courtesy of the University of Waterloo.)

The system’s AI searches the confirmed cancer biopsies in order to detect cases similar to the undiagnosed digital image. Via the verified findings of similar cancer cases, the system is able to recommend a diagnosis from a new image. During four months of testing, which involved approximately 30,000 slides from around 11,000 patients, the system was able to diagnose 32 forms of cancer in 25 organs and body parts with 100 percent accuracy.

“We showed it is possible using this approach to get incredibly encouraging results if you have access to a large archive,” Tizhoosh said. “It is like putting many, many pathologists in a virtual room together and having them reach consensus.”

While additional research will be needed to enhance the system, the researchers said that it has already shown how it can be beneficial as a screening tool to assist pathologists in improving and speeding up diagnoses. For parts of the world that may lack resources, it could be an inexpensive alternative that could be remotely accessed to promote early treatment.

“This technology could be a blessing in places where there simply aren’t enough specialists,” Tizhoosh said. “One could just send an image attached to an email and get a report back.”