AI turns routine mammograms into heart risk forecasts

AI model – Researchers report an artificial intelligence model that scans standard mammograms to quantify breast arterial calcifications (BAC)—calcium “railroad-track” streaks linked to stiffer vessels. In two large patient groups, even small BAC increases correlated wit

When a woman goes in for a routine mammogram, she’s usually looking for one answer: whether breast cancer is hiding in the tissue. A new AI system is aiming to squeeze another critical clue out of that same scan—one that could flag heart disease risk before symptoms ever show up.

The model, reported in the European Heart Journal, analyzes mammograms to measure bright streaks of calcium buildup known as breast arterial calcifications, or BAC. Researchers say the severity of these calcifications can help predict associated heart disease risk.

“Can be run on every single mammogram without any additional work. ” says Hari Trivedi. an Emory University radiologist and study co-author. In the U.S. more than 40 million mammograms are conducted every year—numbers large enough that adding a second function to a test already built into many care pathways could change what clinicians notice in real time.

BAC appears differently than the calcium deposits that can be left behind after breast cancer cells rapidly die. On a mammogram. those breast cancer–related deposits show up as small clusters. an early sign of abnormal growth and tumors. BAC, by contrast, often looks like two bright “railroad-track” stripes zigzagging through breast tissue, Trivedi says.

Trivedi emphasizes that BAC itself doesn’t harm breast tissue and doesn’t increase the risk of breast cancer. Instead, it may signal that calcification is developing elsewhere in the body. That matters because stiffened vessels can mean poorer circulation, which can translate into higher cardiovascular risk.

Other groups have explored using mammograms to assess heart disease risk, and some radiologists already look out for BAC during screenings. But Trivedi says his team’s model is the first approach that measures and tracks the amount of BAC and ties that specific amount to higher disease risk.

Some other AI tools for BAC detection exist as well, including one cleared by the U.S. Food and Drug Administration. Trivedi’s point is that quantifying the amount of BAC could allow clinicians to predict risk level more precisely.

The researchers built their findings around two populations: 74. 124 people at Emory Healthcare in Atlanta and 49. 638 people at Mayo Clinic sites in Arizona. Florida. Minnesota and the upper Midwest. They report that even small increases in BAC were correlated with slightly raised cardiovascular risk. More severe BAC levels. they found. were associated with fourfold to eightfold increases in events such as heart attack and stroke compared with rates in people who have none.

The sequence is straightforward and hard to ignore: a measurement of BAC on a mammogram lines up with cardiovascular outcomes, and that relationship strengthens as BAC severity increases.

breast arterial calcifications BAC mammograms artificial intelligence heart disease risk cardiovascular events European Heart Journal Emory Healthcare Mayo Clinic