A research team at Radboud University Medical Center in the Netherlands has developed an artificial intelligence system capable of accurately predicting the risk of heart disease recurrence in patients by analyzing coronary optical coherence tomography (OCT) images. The study, led by technician Jos Thannhauser and physician Rick Volleberg, was published in the European Heart Journal.
Approximately 15% of patients with coronary artery disease may experience recurrence within two years after undergoing angioplasty. Traditional OCT image analysis relies on manual interpretation by specialized laboratories, which is complex and limited in efficiency. The research team used artificial intelligence algorithms to analyze coronary OCT data from 438 patients and followed them for two years. The results showed that the AI achieved accuracy in identifying vulnerable regions of the arterial wall comparable to the international gold standard, and outperformed traditional methods in predicting the risk of new infarction or death.
Volleberg said: "If we know who has high-risk plaques and where they are located, we may be able to tailor medication or even place preventive stents in the future." OCT technology uses a miniature camera inserted into the blood vessel to capture near-infrared light images, providing microscopic resolution of the vessel wall structure. It is currently used in clinical practice to guide procedures and evaluate stent placement.
Thannhauser pointed out: "This technology is already used in clinical practice to guide angioplasty and check whether stents are placed correctly. Our study shows that combining OCT with AI has even greater potential for assessing the entire vessel." Currently, OCT image interpretation faces challenges such as the large volume of data and high analysis costs. Each examination can produce hundreds of images, and only a few specialized laboratories have the capacity to analyze them.
The introduction of artificial intelligence technology has significantly improved analysis efficiency and reliability. Thannhauser added: "AI is already able to assist doctors in using OCT for stent placement. Thanks to our artificial intelligence, we are now one step closer to scanning the entire coronary artery in clinical practice to identify vulnerable areas." The study was advanced by the CARA laboratory in collaboration with Abbott, Radboudumc, and Amsterdam UMC. It is expected to take several more years to achieve clinical application.
