Mammograms Hide A Silent Heart Warning

Your routine mammogram may already hold a hidden heart risk score that no one is telling you about yet.

Story Snapshot

Doctors can now use artificial intelligence to measure calcium in breast arteries on mammograms as a hard number, in square millimeters, across more than 120,000 women.^[1]^[2]
That calcium, called breast arterial calcification, is a powerful warning sign for heart attacks, strokes, and death, even after standard risk scores say you are “fine.”^[2]^[4]
Each tiny step up in calcification raises risk in a smooth, dose-response way, not an on–off switch.^[2]^[4]
The tool sits in review at the United States Food and Drug Administration while big medical groups move slowly, leaving patients in the dark.^[2]

An overlooked signal hiding in plain sight

Every year, about 40 million American women get screening mammograms, yet most of those images are used for one job only: finding breast cancer.^[3] On those same images, the arteries can show bright, rail-like streaks of calcium called breast arterial calcification, which has long been brushed off as “incidental.”^[6] Large studies now show this “incidental” finding tracks strongly with heart and vessel disease, the top killer of women.^[2]^[6]

A Mayo Clinic and Emory team led by Dr. Imon Banerjee built an artificial intelligence system to stop guessing and start measuring that calcium.^[1]^[3] Instead of a rough visual score, the model outlines every speck of breast arterial calcification and totals the area in square millimeters.^[1]^[2] They ran it on mammograms from 123,762 women across many hospitals, turning fuzzy notes into clean numbers that statisticians and cardiologists can trust.^[1]^[2]

What the numbers say about risk

The team sorted women into groups by how much calcified area they had: none, mild, moderate, and severe.^[2] Women with more breast arterial calcification had more major heart events and more deaths, even after adjusting for age, blood pressure, cholesterol, diabetes, and a modern American Heart Association and American College of Cardiology risk score called PREVENT.^[2]^[4] In plain terms, the calcium signal stayed strong even when the usual boxes were already checked.^[2]^[4]

When the researchers treated breast arterial calcification as a continuous number, the picture sharpened.^[2]^[4] Each 1 square millimeter gain in calcified area brought about a 2% to 3% jump in risk for heart attack, stroke, or cardiovascular death after adjusting for the PREVENT score.^[2]^[4]

How the AI model works in real practice

The artificial intelligence algorithm reads standard digital mammograms and, with a single click from the radiologist, outputs a breast arterial calcification area number and category.^[2] Radiologists in the validation study compared its measurements with their own reads across 12 institutions, and the tool held up well.^[2]^[4] Because it runs on images already taken for cancer screening, it adds no extra radiation, scan time, or cost to the patient; it simply mines more value from the same picture.^[2]^[4]

The model is now under review by the United States Food and Drug Administration for clinical use, which means regulators are checking whether the code is reliable and safe enough to feed into real medical reports.^[2] Once cleared, it could show up in mammogram results as a new line item, sitting next to breast density and cancer findings, flagging which women are quietly building dangerous artery disease.^[2]

Where evidence is strong and where it is thin

The dataset is large, multi-institutional, and racially diverse, which raises confidence that the finding is not tied to one hospital or hometown.^[2]^[3] The association between higher breast arterial calcification and more bad outcomes appears consistent across different ways of slicing the data, including women who start out “low risk” by traditional scores.^[2]^[3] Reviews of multiple studies now call breast arterial calcification a clinically meaningful marker, not just background noise.^[6]

At the same time, all of this is still observational work, mainly retrospective cohort analyses that look backwards at who had events.^[2]^[9]^[12] These studies prove prediction, not that acting on the artificial intelligence result saves lives. There is no randomized trial yet where one group’s breast arterial calcification is reported and treated, and another group’s is hidden, to see if heart attacks drop. That gap keeps major guideline writers cautious, and they are not wrong to ask for that higher bar.^[12]