Breast cancer may become easier to detect before it even develops with the help of new AI-based software created in St. Louis.
Prognosia Breast, which was developed by a team of Washington University School of Medicine researchers, analyzes mammograms to determine a person’s risk of developing breast cancer within five years.
The technology recently cleared a major hurdle earning a breakthrough device designation from the Food and Drug Administration. Joy Jiang, a codeveloper of the software and an associate professor of surgery at the university, said the designation means it is a step closer to full market approval.
“[This] is a huge step forward,” Jiang said. “It’s being recognized that breast cancer is a huge burden and early detection and prevention is very important right now.”
The current standard procedure for breast cancer prevention is through a questionnaire that includes BMI and general family health history, as well as cancer history. Jiang said this tool will improve accuracy.
“We were motivated to invent this new tool by using mammogram images alone, so women wouldn’t have to fill in these questionnaires anymore,” she said. “Because everybody would get a mammogram every year or two. So given that availability of data, we can directly use that data to predict accurately who’s at risk.”
Prognosia Breast can reach up to 85% accuracy for early cancer detection in comparison to the standard process of up to 65%.
The process is straightforward. Once a patient gets a mammogram, the image will go through Prognosia Breast. The software will then analyze it and provide a risk score for the likelihood of developing breast cancer within five years.
“Based on that output, there may be a conversation between the provider and the woman in terms of what can be done,” Jiang said. “And how can we prevent the breast cancer from happening if the woman is actually at high risk.”
A clinical trial for the software will start by the end of the year.
