November 20, 2018
Regina Barzilay makes a splash in The New York Times discussing the AI technologies that she hopes will transform the cancer clinic with George Church, Jennifer Egan, Catherine Mohr and Siddhartha Mukherjee. Barzilay talks about the personal experience with mammography that inspired her to apply her expertise in AI to the cancer clinic, for instance helping to build an automated database of the records of more than 100,000 patients with breast disease at Massachusetts General and other partner hospitals. The database allows clinicians and researchers to mine data with a simple query, rather than resorting to the arduous process of extracting data by hand.
In other work profiled at MIT News, Barzilay worked with colleagues from MIT and Massachusetts General Hospital to develop an AI model that uses deep-learning to identify dense breast tissue in mammograms. The convolutional neural network-based model–the first of its kind to be successfully used in a clinic on real patients–evaluated mammograms as reliably as expert radiologists, according to the study published in Radiology. An estimated 40 percent of U.S. women have dense breast tissue, which increases the risk of breast cancer in addition to making screening more difficult. The researchers hope the model will bring greater reliability to the assessment of breast density, and plan to scale the model into other hospitals.