Prospects for using machine learning to improve coronary angiography

Abstract

Cardiovascular diseases pose the main threat to the population health of the Russian Federation and rank the first among the causes of death. Coronary heart disease has the highest standardized mortality rates among the population of the Russian Federation. Comprehensive diagnosis of coronary artery disease includes assessment of coronary atherosclerosis using both non-invasive methods, such as multispiral computed tomography of the coronary arteries, and invasive ones, including coronary angiography, and sometimes intravascular imaging. First two methods are the two most important diagnostic methods for coronary heart disease. The widespread use of medical technologies based on artificial intelligence in recent years has led to the emergence of new diagnostic and therapeutic opportunities. Artificial intelligence has bridged the gap between massive datasets and useful information by processing and analyzing important data at an unprecedented rate. The review identifies five potential cases with machine learning having significant prospects in the field of coronary angiography: improving quality and effectiveness, determining plaque characteristics, assessing hemodynamics, predicting disease outcomes and diagnosing non-atherosclerotic lesions of the coronary arteries. While machine learning has transformative potential in the field of coronary angiogram analysis, careful consideration of limitations, including data exchange protocols and interpretability of models is essential to fully exploit its potential and ensure optimal diagnosis and treatment of patients.