AI analysis of routine mammograms may uncover signs of cardiovascular disease in women patients that would otherwise go undetected, according to research by doctors in Israel.
The analysis looks at existing X-ray images for subtle features associated with hypertension, ischaemic heart disease – also known as coronary artery disease – and stroke.
Details of the research, by a team at Sheba Medical Center in Ramat Gan, were presented last month at the European Society of Cardiology’s annual congress, the world’s largest heart conference.
Dr Viana Copeland, who leads the research team at Sheba, told the JC that the goal is to use a test already performed routinely for breast cancer detection to identify women who might otherwise go without cardiovascular assessment.
“Cardiovascular disease is the leading cause of mortality worldwide,” she said. “We’re trying to implement a screening tool for a different disease to diagnose it in women and prevent the mortality and development of these diseases. We wanted to identify a way that we can essentially make it easier for women to get this diagnosis and not miss out, because they’re under-treated and under-diagnosed as well. You’re literally on the same platform and you’re getting the results for two tests, one for cancer and one for cardiovascular disease.”
The approach could potentially identify women with cardiovascular disease or significant risk who remain undiagnosed and could benefit from further assessment.
Dr Copeland is a clinician and research fellow specialising in heart failure, specifically in women.
It was her idea to investigate whether mammograms, a routine screening test for millions of women, most commonly between the ages of 40 and 70, could contain information that an AI model could use to identify women at higher cardiovascular risk or women who already have cardiovascular disease.
The researchers analysed 97,364 mammograms from 29,921 women, and compared the images with existing medical records showing which women had hypertension, ischaemic heart disease or a history of stroke.
Among the women studied, 16 per cent had high blood pressure, 2.5 per cent had ischaemic heart disease and 2.5 per cent had suffered a stroke.
The AI model performed better than chance at distinguishing women who had each condition from those who did not, with its strongest performance for stroke. The AI does not “see” the heart in a mammogram. Rather, it analyses features within the breast X-ray images that may be associated with cardiovascular health but are not routinely examined or interpreted for that purpose.
“There are findings in the mammography that are indicative of heart disease. It’s not that you can see the heart on the mammogram… But there are findings in it,” Dr Copeland said. Further testing is needed before the technology can be used as a clinical screening or referral tool.
Associate professor Elena Arbelo of the European Society of Cardiology said the concept was compelling, while emphasising the need for further validation.
“As both a cardiologist and a woman, I find this concept compelling: a mammogram may one day do more than look for breast cancer, it may also offer a window onto cardiovascular health,” she said.
To get more news, click here to sign up for our free daily newsletter.