Algorithm as accurate as radiologists in assessing breast density, cancer risk

Software that assesses breast density is just as accurate in predicting women’s risk of breast cancer as a typical evaluation of images conducted by radiologists, according to researchers at UC San Francisco and the Mayo Clinic.

In the largest study of its kind to date, researchers pitted automated breast density software from New Zealand’s Volpara Solutions against Breast Imaging Reporting and Data System (BI-RADS) density categories estimated by radiologists.

What they found was that automated and clinical BI-RADS evaluations “similarly predict interval and screen-detected cancer risk, suggesting that either measure may be used to inform women of their breast density.” The results of the study were published May 1 in the Annals of Internal Medicine.

Currently, 30 states have laws requiring that women receive notification of their breast density, which is graded on the standard four-category BI-RADS scale. However, there have been concerns raised about the reliability of BI-RADS breast density measures, because assessments can vary for women based on the radiologist and the mammogram they’re analyzing, says UCSF Professor of Medicine Karla Kerlikowske, MD.

“You have this mandatory reporting, but there’s no standard way to report it,” says Kerlikowske, who— along with Mayo Clinic Professor of Epidemiology Celine Vachon—led the National Cancer Institute-funded study. “The advantage of the automated measure is that it can be very consistent from one facility to the next. It’s much more reproducible. As a clinician, if I’m going to act on something, I want to make sure I’m really doing the right thing for the patient. The automated measure allows for that consistency.”

Also See: Machine learning helps Mass General predict cancerous breast lesions

Researchers discovered that women with an automated BI-RADS assessment of extremely dense breasts had a 5.65 times higher risk of interval cancer and a 1.43 times higher risk of screen-detected cancer than women with scattered fibroglandular densities—the most common density category in average-risk women.

“It’s highly predictive,” adds Kerlikowske. “One of the interesting things in our study is that breast density was more predictive of interval or missed cancers than the ones they actually find in mammography.”

The study included 1,609 women with cancer detected within a year of a positive mammography result, 351 women with interval invasive cancer, and 4,409 control participants matched by age, race, state of residence, screening date and mammography machine.

Kerlikowske recounts how one radiologist told her that he could “save an hour a day” in his practice by instituting an automated system instead of having to read mammograms for breast density.

“This would be a way to standardize the reporting of breast density for women across the United States,” she concludes.