MGH, MIT researchers create AI tool that predicts risk of breast cancer

A deep learning model is able to accurately assess breast tissue density—an independent risk factor for breast cancer—in mammograms.

In a new study of the model, developed by Massachusetts General Hospital and MIT, nearly 90,000 full-resolution screening mammograms from about 40,000 women were used to train, validate and test the deep learning model.

“We developed a deep learning model that uses full-field mammograms and traditional risk factors, and found that our model was more accurate than the Tyrer-Cusick model (version 8), a current clinical standard,” conclude researchers in a study published earlier this week in the journal Radiology.

Also See: Algorithm as accurate as radiologists in assessing breast density, cancer risk

“Unlike traditional models, our deep learning model performs equally well across diverse races, ages and family histories,” says Regina Barzilay, senior author and the Delta Electronics Professor at MIT’s Computer Science and Artificial Intelligence Laboratory and the Department of Electrical Engineering and Computer Science. “Until now, African-American women were at a distinct disadvantage in having accurate risk assessment of future breast cancer. Our AI model has changed that.”

Among the study’s findings were that patients with non-dense breasts and model-assessed high risk had 3.9 times the cancer incidence of patients with dense breasts and model-assessed low risk.

“These results support the hypothesis that mammography contains informative indicators of risk not captured by traditional risk factors, and DL models can deduce these patterns from the data,” states the study. “These models have the potential to replace conventional risk prediction models. Further research is required to validate our model across institutions and vendors before it can be broadly implemented, and to this end, we made our trained model and code available for research.”