An estimated 40 percent of American women have dense breast tissue, which can mask cancers on mammograms and make for challenging breast cancer screenings. Researchers from MIT and Massachusetts General Hospital (MGH) have developed an automated model that can assess dense breast tissue as reliably as expert radiologists — an innovation they hope will make mammograms more reliable.
Across the country, 30 states currently mandate that women must be notified if their mammograms indicate they have dense breasts. But that assessment can vary greatly among radiologists. MIT and MGH researchers trained a deep-learning model based on tens of thousands of high-quality digital mammograms to learn to distinguish fatty to extremely dense tissue based on expert assessments.
Given a new mammogram, the model can now identify a density measurement that aligns with expert opinion 90 percent of the time — and it takes less than a second.
Source: Massachusetts Institute of Technology