At the annual meeting of the Radiological Society of North America (RSNA) on November 29, 2022, researchers presented a new deep learning model that uses a single X-ray to predict a patient’s 10-year risk of dying from cardiovascular disease.
Researchers recently developed a deep learning model that predicts the 10-year risk of a heart attack or stroke using only a single chest X-ray.
A deep learning model is an advanced artificial intelligence trained to search X-ray images and find patterns related to cardiovascular disease (CVD).
Medical doctors currently use a risk score that is effective but more cumbersome to know when to prescribe a statin drug for CVD prevention.
A model using just an X-ray may be easier to help doctors identify and treat high-risk patients.
New model is simple and accurate
Cardiac disease is the world’s leading cause of death. By using only a routine chest X-ray, the new deep learning model simplifies the process for doctors, which could help identify more high-risk people.
Medical doctors currently use diagnostic tools like the atherosclerotic cardiovascular disease (ASCVD) risk score to decide when to prescribe a statin drug for high cholesterol. Current medical guidelines recommend statins – a medication to reduce cholesterol – for anyone with a 10-year risk of major cardiac events like a heart attack or a stroke.
"The beauty of this approach is you only need an X-ray, which is acquired millions of times a day across the world,... Based on a single existing chest X-ray image, our deep learning model predicts future major adverse cardiovascular events with similar performance and incremental value to the established clinical standard [ASCVD risk score]."Jakob Weiss, MD, in RSNA's press release
To calculate the risk, the ASCVD risk score weighs multiple variables like sex, race, age, blood pressure, smoking, Type 2 diabetes, high blood pressure treatment, and blood tests. Collecting this data for each patient can be cumbersome.
AI finds patterns in chest X-rays
A deep learning model is a type of advanced artificial intelligence (AI). Researchers trained the model to search X-ray images and find patterns related to cardiovascular disease.
The researchers call the model CXR-CVD risk. CXR is an acronym for chest X-ray. CVD is an acronym for cardiovascular disease.
To develop the model, the team of scientists used 147,497 chest X-rays from 40,643 people in a randomized controlled trial. They then tested the model using a different group of 11,340 patients.
Over 10 years, 9.6%, of the second study group suffered a major cardiac event. The deep learning model adequately predicted these events for this subset of patients. Participants had a mean age of 60.1 years and were 42.9% male.
Further research is still needed before doctors use the model as a decision-making tool, but the CXR-CVD risk model successfully predicted the 10-year risk similarly to the ASCVD risk score.
"What we've shown is a chest X-ray is more than a chest X-ray," Dr. Weiss said. "With an approach like this, we get a quantitative measure, which allows us to provide both diagnostic and prognostic information that helps the clinician and the patient."
Cardiovascular disease and X-rays
Cardiovascular diseases describe disorders of the heart and blood vessels.
According to the World Health Organization, about 17.9 million people died from CVDs in 2019, 32% of all global deaths. 85% of these deaths were due to heart attack and stroke. In the U.S., someone has a heart attack every 40 seconds.
Some examples of CVDs include:
Coronary artery disease – damage to the arteries supplying blood to the heart muscle, which can cause a heart attack.
Cerebrovascular disease – damage to the blood vessels supplying the brain, which can cause a stroke.
Peripheral artery disease – damage to the blood vessels supplying the arms and legs.
Chronic heart failure – the inability of the heart to pump blood effectively to the body.
Deep vein thrombosis and pulmonary embolism – blood clots that start in leg veins but can break off and move to the lungs.
Fortunately, exercise, diet, weight-loss, stress-reduction, and substance abuse treatment can prevent most cardiovascular diseases. Early diagnosis is essential to manage cardiovascular disease successfully.
Doctors can't see most heart conditions by looking at an X-ray, but they can assess size and shape of a heart. They also see fluid backed up in the lungs due to heart failure.
While the CXR-CVD risk model may not diagnose specific heart conditions, researchers hope further testing confirms that it can consistently predict a patient’s 10-year risk of cardiovascular death.
"We've long recognized that X-rays capture information beyond traditional diagnostic findings, but we haven't used this data because we haven't had robust, reliable methods," Dr. Weiss said. "Advances in AI are making it possible now."
The American Heart Association (AHA) and the National Academy of Medicine partly funded the research and development of the CXR-CVD risk model.
AHA advises people 20 and older to talk with their doctor about their cardiovascular risks and recommendations for heart check-ups and monitoring.