A University of Idaho researcher is developing a new technology for detecting breast cancer in women.
Ultrasound technology, which breastcancer.org reports is a supplementary test for other cancer screenings, allows doctors to examine abnormalities in mammograms. Min Xian is working on a way to connect AI and smartphones to ultrasound technology. This software would allow cancer patients to monitor the size of their tumors daily from the comfort of their home, Xian said.
Ten years ago, as computer science assistant professor at UI, Xian began studying breast cancer detection. He was assigned the task to work on a portable method of detecting breast cancer as part of his master’s thesis work at the Harbin Institute of Technology. Xian continued his research at Utah State University as he worked towards his PhD.
“We basically design new methodologies for biomedical data analysis,” Xian said. “For example, right now we are doing breast cancer research. That’s one of our applications. We design new approaches and then apply them into biomedical applications.”
Alex Brizee | Argonaut
Cancer detection methods can be expensive. Xian said X-rays can run around $280 per scan, while women are encouraged to get scanned once per year after turning 40 years old to ensure cancer is detected before it reaches its later stages. Women in low-income situations and developing countries do not have the resources necessary for this kind of care, he said. Xian said he hopes his portable scanner, an ultrasound device connected to a smartphone app, will make medical diagnostic equipment more accessible to these demographics.
Xian’s portable scanning technology would allow people to keep a scanner at home. He said costs for the scanner itself and the software it requires will need to be lowered. As of now, each scanner costs $5,000. Xian hopes the price will become lower over time, eventually becoming reasonable for every household in the world to have access to a scanner.
Xian said policies about the use of artificial intelligence in medicine must change as well. Xian said some people have concerns about where blame would be laid if an AI misdiagnosed — or failed to diagnose — a problem. That’s why he suggests using AIs and human doctors together, so one can catch the possible mistakes of the other.
“We need experts to do the cancer detection but we also can design artificial intelligence-based algorithms to replace the cancer detection process performed by the radiologists, by the experts,” Xian said. “Our strategy is we design new intelligence algorithms to do the cancer detection automatically. That’s also a good thing for cancer detection because (AI) is objective, not subjective. It does depend of the experience of the radiologist.”
Xian works with fifteen other researchers on this project. Four of these researchers are UI graduate research assistants.
Haotian Wang and Longze Li, two of these assistants, focus on developing an algorithm that allows software to identify cancerous cells. Wang said this software would be used to determine the rate at which cancer is spreading in a patient already diagnosed with cancer.
Wang and Li both discussed difficulties in the process of teaching the technology where one cell ends and another begins, or “segmenting” the images. Ultrasound imagery is not high resolution, Li said, which makes it more difficult to tell. Once the software can effectively segment cells, the researchers can “teach” it which cells are healthy and which cells are cancerous.
Xian continued to study breast cancer detection after he earned his doctorate because doctors cannot reliably detect breast cancer in its early stages. Xian said the high mortality rates associated with breast cancer are a major problem. He now works with a large interdisciplinary team across the world to work on improving these devices.“I think in the future, everybody can have one at home,” Xian said. “I think for somebody having breast cancer, they can monitor their cancer progress every day. They can do that.”
Lex Miller can be reached at [email protected]