More than 500,000 women worldwide die of breast cancer, a statistic that keeps Connie Lehman, MD, Chief of Breast Imaging at Massachusetts General Hospital (MGH) awake at night. Her desire to change the statistic, she says, requires accelerated innovation. “We have to think differently,” she says.
Dr. Lehman is leading the charge toward that paradigm shift. In collaboration with Massachusetts Institute of Technology (MIT) professor Regina Barzilay, PhD, she’s part of a growing movement to apply computer algorithms to make mammography easier, faster, and smarter. Dr. Barzilay took on the research after her own battle with breast cancer revealed the lag in technology available to support treatment.
Together, Drs. Lehman and Barzilay have developed a deep-learning algorithm they believe can soon be trained to read mammograms as well as an average radiologist. By turning to technology to quickly identify normal mammograms, they say, radiologists could be freed for more demanding tasks. In the future, it could be used to better identify which patients require a biopsy—currently a subjective undertaking that can result in unnecessary procedures. They’ve also developed a computer program that analyzes vast data to predict a patient’s risk of developing breast cancer in the future.
The work Drs. Lehman and Barzilay have created was presented during the 2019 World Medical Innovation Forum’s First Look session, where rising stars from Harvard-affiliated hospitals highlight the potential of their research in artificial intelligence, cognitive computing, machine learning, and big data to accelerate the application of high impact technologies and projects to the benefit of patients.
“We’re so excited about it, because it’s a stronger predictor than anything else out there,” Lehman says.
Read and listen for more on their research and partnership at NPR.