Measurements taken with a mini camera in the coronary arteries surrounding the heart can accurately predict whether someone will have another heart attack. Until now, interpreting these images has been so complex for the human eye that it has only been done in a few specialized labs. A new study by Radboudumc shows that AI can reliably take over these analyses and quickly assess blood vessels for weak spots.
A heart attack occurs when a coronary artery, which supplies the heart with blood, becomes blocked by a blood clot. This can happen when arteriosclerosis causes narrowing of the
blood vessels. The heart then receives too little oxygen. Treatment typically consists of angioplasty, in which a doctor inflates a small balloon in the artery, usually in combination with the placement of a tube, known as a stent. This procedure is performed approximately 40,000 times a year in the Netherlands.
Predictor
Nevertheless, approximately 15% of people who have had a heart attack will have another one within two years. Technical physician Jos Thannhauser and physician-researcher Rick Volleberg of Radboudumc are therefore working with their team to investigate whether they can better detect weak spots in the artery wall that could cause a new heart attack. The researchers analyzed the coronary arteries of 438 patients with a mini camera and a specially developed AI, and followed these participants for two years.
The study shows that AI detects weak spots in the vessel wall just as well as specialized labs—the international gold standard—and even more accurately predicts whether a new hea
rt attack or death will occur within two years. How does this benefit the patient? Volleberg: “If we know who has high-risk weak spots in their blood vessels and where they are located, we can adjust medication accordingly in the future or place a stent preventively.”
‘Healthcare is unprepared for the good advice by Doctor ChatGPT’
Professor of AI-Guided Imaging Henkjan Huisman wants better and more affordable healthcare through research into the possibilities of AI.
Vessel wall
The mini camera uses a technique called optical coherence tomography (OCT). The camera is inserted into the bloodstream via the arm and takes images of the vessels using near-infrared light. This allows it to visualize the structure of the vessel walls at a microscopic level.
“Doctors already use this technique in the clinic to determine the best way to place a stent during angioplasty and to check whether the stent is in the right position,” says Thannhauser. It has already been demonstrated that the use of OCT leads to fewer new heart attacks and fewer deaths from complications. But then doctors only look at a very small area in a coronary artery where the infarction occurred. The current study demonstrates that this technique has considerable potential and can also visualize the entire blood vessel.
Towards clinical application with AI
“One challenge with this technique is that doctors find it difficult to assess the OCT images,” says Thannhauser. And that’s not surprising, because the camera takes 540 images at a time. The analysis is challenging, even when the only objective is to place a stent. If you want to analyze the coronary arteries in their entirety, that would involve far too many images. “Currently, only a few specialized labs can assess these images, but they don’t look at everything. In addition, it is too expensive and labor-intensive to apply this as human work in the clinic.”
That is why Thannhauser’s team developed an AI that analyzes all the images. It is just as reliable as analysis by a specialized laboratory and much faster. Thannhauser: “Our AI can already help doctors with OCT when they place a stent. In addition, thanks to AI, we are getting closer to our goal of scanning entire blood vessels for weak spots in the clinic. But I expect that this will still take a number of years.”