Mammograms May Help Identify Heart Disease Risk (VIDEO)

By Carey Stanton

Heart disease is the leading cause of death among women, responsible for 1 in 3 deaths. It often develops without symptoms until a major event occurs. Despite this, there is not a standard for annual heart disease screening, a critical gap in early detection for women.

Doctors typically assess heart disease risk using factors like cholesterol, blood pressure and body weight. These tools don’t fully account for how heart disease develops differently in women or what’s happening inside the blood vessels themselves.

A missed opportunity in routine mammograms

Mayo Clinic recommends women at average risk begin annual mammogram screening at age 40, and most women do. Researchers say this creates an opportunity to assess heart disease risk during a test many women already receive without additional radiation or testing.

Breast arterial calcification (BAC) is captured during a mammogram. For two decades, studies have demonstrated that it can be used to help identify patients at risk for heart disease. However, the challenge has been measuring it accurately and consistently, and assigning a value to it.

“BAC is difficult to measure because it appears as a very subtle finding on standard 2D mammograms and cannot be consistently quantified by eye, even by experts. As a result, it typically requires specialized computational methods to reliably detect and measure it,” explains Dr. Imon Banerjee, scientific director of the Arizona Advanced AI and Innovation Hub at Mayo Clinic. “AI allows us to measure these calcium deposits quickly and more accurately than before.”

Large study shows strong link to risk

A retrospective cohort study published in the European Heart Journal included more than 120,000 women who had screening mammograms at two healthcare systems. They evaluated whether AI-derived BAC measurements could help clinicians predict a woman’s risk of cardiovascular disease and death, independent of traditional risk factors. Researchers note the study was observational and evaluated associations between BAC and future cardiovascular outcomes.

“Breast arterial calcification occurs when calcium builds up in the walls of the arteries within breast tissue,” says Dr. Banerjee. “Calcium seen on mammograms correlates with calcification in other parts of the body. This type of calcium in the breast is different from calcium in the heart. Instead of blocking blood flow, it makes the blood vessels stiffer and less flexible, signaling a higher risk of heart problems because it affects how blood moves.”

Using a deep learning model, the team developed a tool that identifies calcium deposits, measures their extent and classifies severity. Researchers then tracked whether higher levels were associated with cardiovascular events or death over time.

The findings were significant. Women with severe BAC had more than 10 times the risk of developing a cardiovascular event within five years compared to those with no or mild BAC. The model was validated across 12 institutions, comparing AI-generated measurements with radiologist assessments. The AI model is now undergoing Food and Drug Administration review.

WATCH: Dr. Imon Banerjee – AI can accurately measure heart disease risk

Moving towards clinical use

“We created an AI algorithm that allows us, with a single click, to measure BAC. It can now easily be added to the radiologist’s report,” Dr. Banerjee says. “That measurement becomes a new input for the prediction models. This could completely change the screening workflow for women and improve early detection of heart disease.”

Adoption remains the next challenge

“The challenge with BAC is that we have the algorithm, we have the validation, we know that it works. Now, how do the institutions adopt it?” acknowledges Dr. Banerjee. “There are currently no clinical regulations of BAC. Mayo Clinic physicians are translating this research into practice.”

Teams across Cardiovascular Medicine, Primary Care, Women’s Health and Radiology at Mayo Clinic’s Arizona and Florida campuses are working to integrate the technology into clinical practice and develop a framework for broader use before expanding to all Mayo Clinic sites.