Deep Learning Model Can Predict Breast Cancer up to Five Years in Advance

The new MIT system was trained on the mammograms and outcomes of more than 60,000 patients; from this data, the algorithm learned the subtle patterns in breast tissue that are precursors to malignant tumors. The system's creators hope that it will make late breast cancer detection a thing of the past.

Risk-based screening more accurate

The system will help doctors develop individual risk management plans for women that will determine how often they should be screened. Currently, the American Cancer Society recommends annual screening starting at age 45, the US.

Preventative Task Force recommends screening every two years starting at age 50. But for women with a high risk, this may not be enough.

“Rather than taking a one-size-fits-all approach, we can personalize screening around a woman’s risk of developing cancer,” says Barzilay, senior author of a new paper about the project out today in Radiology.

“For example, a doctor might recommend that one group of women get a mammogram every other year, while another higher-risk group might get supplemental MRI screening.”

A system more accurate than traditional methods

Barzilay is the Delta Electronics Professor at CSAIL and the Department of Electrical Engineering and Computer Science at MIT and a member of the Koch Institute for Integrative Cancer Research at MIT. The system accurately placed 31 percent of all cancer patients in its highest-risk category, compared to only 18 percent for traditional models.

The system proves that screening strategies can be determined on risk factors rather than just age. Previously a woman's risk factor of developing breast cancer was determined by a combination of age, family history of breast and ovarian cancer, hormonal and reproductive factors, and breast density.

Algorithms detect patterns too subtle for humans

These markers are weakly connected to the actual development of breast cancer, and risk-based screening is not widely supported. The MIT/MGH team developed a deep learning model that can identify patterns in mammograms that drive future cancer. Training on more than 90,000 mammograms, the model detected patterns too subtle for the human eye to detect.

“Since the 1960s radiologists have noticed that women have unique and widely variable patterns of breast tissue visible on the mammogram,” says Lehman.

“These patterns can represent the influence of genetics, hormones, pregnancy, lactation, diet, weight loss, and weight gain.

We can now leverage this detailed information to be more precise in our risk assessment at the individual level.”

The model will also close the gap in breast cancer detection and treatment between black and white women. Black women are 42 % more likely to die from breast cancer than white women due to a variety of factors, including access to healthcare.

The team hopes the system can become a standard part of healthcare across the U.S and the world.