3 Minute ReadMIT Researcher’s AI System Uses Machine Learning To Detect Early Breast Cancer

There is an AI system in the works that will use machine learning to improve early breast cancer detection and reduce false positives and unnecessary surgeries.

1 in 8 women will be diagnosed with breast cancer during their lifetime. In the UK, incidence rates for breast cancer are projected to rise by 2%. This equals to 210 cases per 100,000 females by 2035. These scary stats could be a thing of the past if Manisha Bahl, director of the Massachusetts General Hospital Breast Imaging Fellowship Program; MIT Professor Regina Barzilay; and Constance Lehman, a professor at Harvard Medical School and chief of the Breast Imaging Division at MGH’s Department of Radiology have anything to do with it.

MIT
Pictured, left to right, are Manisha Bahl, Regina Barzilay (centre); and Constance Lehman. Image Credit: Jason Dorfman/CSAIL

Although Mammograms are the best tests available, Barzilay, Bahl and Lehman have come up with a new way to use artificial intelligence (AI) to improve early breast cancer detection. They have developed an AI system that uses machine learning to predict if a high-risk lesion identified on needle biopsy after a mammogram will upgrade to cancer at surgery. According to MIT news, when the process was tested on 335 high-risk lesions, the model correctly diagnosed 97 percent of the breast cancers as malignant and reduced the number of benign surgeries by more than 30 percent compared to existing approaches.

“When there’s this much uncertainty in data, machine learning is exactly the tool that we need to improve detection and prevent over-treatment.”

Such a development means that women will no longer have to go through painful, expensive, scar-inducing surgeries if they are not necessary. As the first project to apply AI to improving detection and diagnosis, Regina Barzilay, MIT’s Delta Electronics Professor of Electrical Engineering and Computer Science, a breast cancer survivor herself, said, “Because diagnostic tools are so inexact, there is an understandable tendency for doctors to over-screen for breast cancer. When there’s this much uncertainty in data, machine learning is exactly the tool that we need to improve detection and prevent over-treatment.”

Breast Cancer
“Machine learning is exactly the tool that we need to improve detection and prevent over-treatment” -Regina Barzilay

The Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL), Massachusetts General Hospital, and Harvard Medical School explain that they have come up with a method known as a “random-forest classifier”. It is a model that has brought about results that meant fewer unnecessary surgeries compared to the strategy of always doing surgery. Specifically, the new model diagnosed 97 percent of cancers compared to 79 percent. “In the past, we might have recommended that all high-risk lesions be surgically excised,” Lehman says. “But now, if the model determines that the lesion has a meagre chance of being cancerous in a specific patient, we can have a more informed discussion with our patient about her options. It may be reasonable for some patients to have their lesions followed with imaging rather than surgically excised.”

Also Read: Science Meets Fashion: A Smart Cancer Fighting Wonder Bra

With the team still working to further hone on the model, they hope that in the future they will be able to tweak it so it can be applied to other kinds of cancer and even other diseases entirely. For now, they look forward to MGH radiologists incorporating the model into their clinical practice over the next year (2018).

Share Your Tips & Corrections

Muchaneta Kapfunde
Founding Editor in Chief at | editor@fashnerd.com

Founding editor-in-chief & WearableTechStylist of FashNerd.com has worked in the fashion industry for over 14 years. She is currently one of the leading influencers speaking and writing about the merger of fashion with technology and wearable technology. She also contributes to other digital news sites like Wareable.