AI Tool Shows Promise in Detecting Hard-to-Find Breast Cancers

An artificial intelligence (AI) tool may help identify women at higher risk of developing interval breast cancers—tumors that appear between regular screening mammograms—according to a new study in Radiology, published by the Radiological Society of North America (RSNA).

“Interval cancers generally have a worse prognosis compared with screen-detected cancers because they tend to be either larger or more aggressive,” said co-author Fiona J. Gilbert, M.B.Ch.B., professor of radiology at the University of Cambridge and honorary consultant radiologist at Addenbrooke’s Hospital. “That’s why it’s important to minimize the number of interval cancers in any screening program.”

Dr. Gilbert and lead researcher Joshua W. D. Rothwell, a medical student and Ph.D. candidate at Cambridge, used AI to analyze mammograms from the U.K.’s triennial breast screening program to determine whether AI could help identify women who might benefit from supplemental imaging or more frequent screening.

The study analyzed 134,217 mammograms from women aged 50 to 70, including 524 interval cancers, performed between 2014 and 2016 at two screening centers. Negative mammograms were processed using Mirai, a deep learning model that produces a risk score for developing interval breast cancer based on mammographic features such as tissue density.

Mirai’s three-year risk predictions correctly identified 3.6% of interval cancers among the top 1% of risk scores and 42.4% among the top 20%. “Our results suggest that further workup of mammograms within the top 20% of scores could yield 42.4% of interval cancers,” said Rothwell. “Mirai could be used to identify women for supplemental imaging or a shortened screening interval.”

The AI performed best at predicting cancers diagnosed within one year of screening and outperformed traditional risk models, though it was less accurate in women with extremely dense breast tissue.

In the U.K., 2.2 million women are screened for breast cancer annually. Dr. Gilbert noted that AI could help optimize the national screening program by targeting women most likely to benefit from additional imaging, such as MRI or contrast-enhanced mammography. “If we called back 20% of women for supplemental imaging, we’d have to find the capacity to offer it to 440,000 women,” she said.