This hospital is CT’s first to use AI to detect early signs of coronary artery disease. Here’s how

STAMFORD — A new artificial intelligence algorithm at Stamford Health’s Heart & Vascular Institute is helping doctors identify the first signs of coronary artery disease in patients receiving computed tomography scans, more commonly known as CT scans. 

Dr. David Hsi, chief of cardiology and institute co-director, said the U.S. Food and Drug Administration-approved algorithm can provide doctors with an accurate data analysis within seconds. 

Stamford Health is the first health care system in Connecticut to implement this specific software, according to the developer San Francisco-based BunkerHill Health. So far, Stamford Health has already performed more than 2,000 chest CT scans using the AI analysis.  

“You can imagine the FDA cleared many AI algorithms, the majority to cardiology, and this is by far the most easily implemented in a cost-effective way for a large patient population,” Hsi said. 

Coronary artery disease is the most common cause of death in the U.S. population and is often “silent in nature,” Hsi said. 

In 2021, heart disease accounted for nearly 20 percent of all Connecticut resident deaths, according to the state Department of Public Health. An estimated 7 percent of residents aged 18 years or older have been told by a health professional that they have had a stroke, heart attack, or coronary heart disease also known as CAD, which is the most common type. 

One of the early objective signs of CAD is coronary artery calcification, which occurs when calcium builds up along walls of the arteries. It’s traditionally found through a specialized test that is not routinely covered by insurance or Medicare, Hsi said. The out-of-pocket cost often presents a barrier to patients, so it’s not universally available for a majority of patients. 

The AI-powered cardiovascular screening algorithm aims to reduce these barriers by detecting the presence of coronary artery calcium in patients who are being scanned for other diseases. The scan almost immediately identifies CAD if it’s present, and at no added cost to the patient. 

Hsi said that they conduct thousands of CT scans a year for a variety of reasons. So, the AI provides a “window of opportunity for us to zero in to see if any of these patients may potentially have a coronary artery disease, if not previously identified.”

The algorithm runs in the digital background of chest CT scans in screenings for other disease, such as lung cancer. After obtaining the raw data from a scan, Hsi said staff put it into a graphic processing circuit workstation with the algorithm. The information is processed in less than five seconds, and the data comes back almost simultaneously. The AI calculates the amount of calcium and the likelihood of cardiac events, also known as the Agatston score.

If the score is high, the patient’s primary care provider or cardiologist is notified immediately. Doctors also receive the patient’s Multi-Ethnic Study of Atherosclerosis (MESA), which can provide percentiles for possible coronary blockage based on age, sex, and race. MESA can also help inform treatment options, like nutrition, exercise, medication or further testing.

The standard chest scan cannot see into coronary arteries, Hsi said, adding: “But with this AI technology, using the same data set, you can see coronary disease. That’s a huge step forward using the technology for patient care.”

Hsi said the algorithm was the first full-scale integration of AI at Stamford Health. AI is already used in cardiac care in the latest generation of medical machines, such as CT scanners and heart rhythm monitors. However, the adoption of more widespread use of AI in cardiology is driven by FDA clearance and validation, cost and IT resources.

The Stamford health organization has been working to implement this to its entire health system over the last year with Bunkerhill Health. The accuracy of the algorithm was verified by researchers across the nation and within its own Connecticut health system. 

Hsi said the algorithm is currently part of the health organization’s daily routine, and it is actively exploring other areas where AI can be implemented in cardiovascular imaging without adding additional burden to the health care system. 

“This is the beginning of AI adoption in cardiovascular care,” he said. “We have a way to go, but the initial results are exciting.”

Click here to read the entire article.