A combination of a miniature camera and artificial intelligence can now accurately predict the likelihood of a recurrent heart attack, according to a new study from Radboud university medical center. The research, published in the European Heart Journal, demonstrates that AI can rapidly and reliably analyze detailed images of coronary arteries to identify weak spots that may lead to future cardiac events. Heart attacks occur when a coronary artery becomes blocked by a blood clot, often due to atherosclerosis—narrowing caused by plaque buildup. Treatment typically involves angioplasty, where a balloon widens the artery, followed by stent placement. In the Netherlands, this procedure is performed around 40,000 times annually. Despite treatment, about 15% of patients experience another heart attack within two years. To address this, researchers led by technical physician Jos Thannhauser and physician Rick Volleberg analyzed the coronary arteries of 438 patients using a tiny camera and advanced AI. The patients were followed for two years to track outcomes. The results show that the AI system detects vulnerable plaques—weak areas in the artery wall—just as accurately as specialized labs, which are considered the international gold standard. Moreover, the AI outperformed traditional methods in predicting future heart attacks or death within two years. Volleberg said the findings could lead to more personalized care. “If we know who has high-risk plaques and where they are located, we may one day tailor medications or even place preventive stents to stop another heart attack before it happens.” The imaging technique used is called optical coherence tomography (OCT). A miniature camera is inserted through the arm into the bloodstream and uses near-infrared light to capture high-resolution images of the artery wall. While OCT is already used during angioplasty to guide stent placement and confirm proper positioning, it has largely been limited to examining only the affected area. Thannhauser explained that analyzing the entire artery using OCT produces hundreds of images, making manual review extremely time-consuming and impractical. “Even assessing stent placement is difficult. Evaluating entire vessels manually is nearly impossible,” he said. Currently, only a few specialized labs can interpret these images, and the process is costly and labor-intensive. To overcome this, Thannhauser’s team developed AI capable of analyzing all OCT images quickly and accurately. “AI can already help during stent placement,” Thannhauser noted. “Now, we’re one step closer to routinely scanning entire coronary arteries for dangerous plaques in clinical practice.” He added that while the technology shows great promise, widespread adoption will likely take several years. Thannhauser leads the CARA Lab—a collaboration between Radboudumc, Amsterdam UMC, and Abbott—focused on advancing cardiology through innovation.