THE PROGNOSTIC ROLE OF A NEW DEEP LEARNING ALGORITHM FOR THE QUANTIFICATION OF EPICARDIAL ADIPOSE TISSUE VOLUME THROUGH CMR IMAGES IN CORONARY ARTERY DISEASE

Epicardial adipose tissue (EAT) represents the fat deposit located between the myocardium and the visceral pericardial layer. Human epicardial fat is a metabolically active organ and a source of several bioactive molecules which have numerous exocrine and paracrine effects. Pericoronary adipose tissue (PCAT) of the EAT directly surrounds the coronary arteries. It has a complex bidirectional interaction with the underlying vascular wall.In physiological conditions it is fundamental in maintaining the homeostasis of the vascular wall; while when dysfunctional (e.g., in inflammatory conditions) it plays a key role in the development and progression of atherosclerosis by the production of bioactive molecules (pro-inflammatory cytokines, adipocytokines, growth factors, etc.). Over the past two decades, EAT has become the subject of increasing scientific investigation, with emerging evidence that it may be associated with the development of coronary artery disease (CAD) and major cardiovascular events. Therefore, in an attempt to improve cardiovascular risk assessment, noninvasive imaging has been used increasingly to characterize EAT. Echocardiography was the first method used for the assessment of EAT, by measuring its thickness along the free wall of the right ventricle. Cardiac magnetic resonance (CMR) is considered the reference modality for imaging total body fat. CMR provides excellent visualization of the visceral and parietal pericardium enabling easy assessment and volumetric quantification of EAT. In the recent years applications of artificial intelligence in cardiac imaging have increasingly developed, thus enabling EAT to be quantified faster. In this study we propose a fully automated quantification of EAT based on the analysis of standard routinely acquired CMR cine images. Moreover, we demonstrate that EAT calculated with this tool and indexed for the body mass index (BMI) is a reliable method for predicting myocardial infarction and cardiac death more effectively than other clinical and imaging variables. Our method does not require the administration of contrast agents. If further validated in other cohorts of patients, it could be applied in the CMR clinical routine to assess the risk of adverse cardiovascular events non-invasively and without the use of contrast agents or ionizing radiation.