Artificial Intelligence Enhances Accuracy and Efficiency of Cardiovascular Angiography

As research utilizing artificial intelligence (AI) in the medical field is actively progressing, a new study has shown that AI can be used for the analysis and procedures of complex cardiovascular angiography. It is expected that the accuracy and efficiency of stent procedures, performed when cardiovascular vessels narrow or become blocked due to conditions such as angina or myocardial infarction, can be significantly improved in the future.

Professor Kang Si-hyeok, Department of Cardiology, Seoul National University Bundang Hospital.

Professor Kang Si-hyeok of the Department of Cardiology at Bundang Seoul National University Hospital and Professor Moon In-tae of the Department of Cardiology at Uijeongbu Eulji University Hospital recently announced research results showing that AI-based analysis is up to 80% similar to conventional intravascular ultrasound examinations. This study was published in the international academic journal 'JMIR CARDIO.'

It is known that approximately 70,000 patients undergo cardiovascular stent insertion procedures annually in Korea. Before stent procedures, cardiovascular angiography is performed to assess the shape of the vessels and the presence of stenosis. However, cardiovascular angiography has the disadvantage of complex imaging and difficulty in fully understanding the three-dimensional structure inside small vessels. As a result, one in four patients additionally undergoes intravascular ultrasound, an expensive medical device costing about 1.8 million KRW, for accurate cardiovascular evaluation.

Accordingly, the research team aimed to determine whether AI software (AI-QCA), which automatically analyzes cardiovascular vessels, could replace intravascular ultrasound. They examined 47 patients who underwent additional intravascular ultrasound and compared the results of AI software with measurements such as percentage diameter stenosis, percentage area stenosis, lesion length, and minimum lumen area.

Based on the cardiovascular angiography image (left), artificial intelligence analyzes and displays the length of lesions in each area (A~C) (bottom right). [Source: Bundang Seoul National University Hospital]

As a result, important indicators during procedures?vessel diameter and area, and lesion length?showed correlations ranging from at least 60% to up to 80% between measurements taken by intravascular ultrasound and those measured by AI software. Additionally, the lesion detection rate was 88.7%, and the difference in lesion size was about 10 mm, showing no significant discrepancy. The AI software demonstrated accuracy comparable to intravascular ultrasound.

In particular, since the AI software provides real-time information on the presence of cardiovascular lesions, lesion length, and diameter, it can assist in determining the length and diameter of stents. If AI software is used alongside stent procedures, which require high physician expertise, more effective examinations and procedures are expected to be possible.

Professor Kang Si-hyeok stated, "Through this study, we found that the analysis results of AI software correlate up to 80% with those analyzed by skilled cardiovascular interventionists using expensive intravascular ultrasound equipment." He added, "Utilizing AI in complex stent procedures could reduce economic burdens while improving procedural accuracy." Professor Moon In-tae explained, "The key to cardiovascular stent procedures is safely inserting an appropriately sized stent without complications. Although this study alone cannot fully evaluate AI's capabilities, it is significant that the numerical data analyzed by AI can be used as reference indicators during procedures."

This content was produced with the assistance of AI translation services.