A team of researchers at Seoul National University Bundang Hospital (SNUBH) and Uijeongbu Eulji Medical Center (UEMC) has demonstrated that artificial intelligence (AI) software can be used to analyze complex cardiac angiograms to improve the accuracy of stent procedures.

Stenting is a medical procedure performed to expand narrowed or blocked blood vessels, typically in cases of angina or myocardial infarction. Approximately 70,000 patients in Korea undergo cardiovascular stenting annually to address these conditions.

Prior to stenting, an angiogram is conducted to assess the morphology of the blood vessel and identify any narrowing. However, angiograms have limitations as they can be intricate and may not provide a comprehensive visualization of all three-dimensional structures within small blood vessels.

Consequently, one in four patients who require stent procedures additionally uses intravascular ultrasound (IVUS), an expensive medical device costing about $1.8 million, to accurately assess the cardiovascular system. 

A team led by Professor Kang Si-hyuck of Cardiology at SNUBH and Professor Moon In-tae of Cardiology at UEMC conducted a study to determine whether AI–based quantitative coronary angiography (AI-QCA) can replace IVUS.

The researchers tested 47 patients who conducted additional IVUS to see how well the stenosis diameter percentage, stenosis area percentage, lesion length, and minimum lumen area matched the results of AI-QCA.

Accordingly, the results showed that the diameter and width of the blood vessel and the length of the lesion, which are important indicators during the procedure, demonstrated a correlation between 60 and 80 percent with those measured by IVUS. 

Additionally, the identification rate of the lesion was 88.7 percent, and the difference in the size of the lesion was less than 10 mm.

In particular, the AI software can also help determine the length and diameter of the stent to discover lesions in the cardiovascular system in real-time, so that the software can be used in conjunction with stent procedures, for more effective testing and procedures.

"The study showed that the results of IVUS analyzed by an experienced cardiologist using expensive medical equipment, correlated with the results of the AI software by up to 80 percent," said Professor Kang. "If AI is used for complex stent procedures, it is expected to improve the accuracy of the procedure while reducing the economic burden." 

Professor Moon said that the key to cardiovascular stenting is to safely implant a properly sized stent without complications. 

 "Although this study cannot evaluate the capabilities of AI alone, it is significant that we have confirmed that numerical values analyzed by AI can be used as indicators for reference during the procedure,” he added.

The study was published in the international journal JMIR CARDIO.