Quantitative Analysis of Epicardial Adipose Tissue: 120K Shades of Heart

Abstract

Background: Epicardial adipose tissue (EAT) is the fat deposited between the myocardium and epicardium. Due to its unique anatomical position, EAT has both protective and harmful effects on the heart, influencing conditions such as coronary artery disease, atrial fibrillation, and heart failure. This study aimed to quantify the amount of EAT by analyzing the color shades of the heart's anterior surface during coronary artery bypass grafting (CABG) procedures.

Objective: To assess the number of color shades in different sub-regions of the heart and quantify EAT using real-time 2D images captured during CABG procedures, and to correlate these findings with clinical conditions and risk factors.

Methods: The study was conducted at Rehman Medical Institute, Peshawar, from October 2023 to April 2024. Images were captured using an iPhone 11 with a 12-megapixel camera during CABG procedures, specifically before cannulation, after opening the pericardium, and tucking the pericardium to the skin on a beating heart. Photographs were taken at a 90-degree angle and one-foot distance during systole, including surrounding tissues and a self-retaining retractor with a ruler for measurement reference. The images from three patients were processed to form the "HEART ANTERIOR VIEW THROUGH STERNOTOMY (HATS)" dataset. The data were cleaned and standardized for consistency. The surgical team annotated and labeled the images using the LabelMe tool, identifying the full heart region and its sub-regions: Aorta, Right Ventricle (RV) Myocardium, RV and Pulmonary Artery (PA) Epicardial Fat, and Right Atrium (RA) Appendage. Image segmentation techniques isolated the heart region and identified fat deposits. The total area of fat on the anterior surface of RV, PA, and RA was quantified using appropriate algorithms. Pixel analysis was conducted to determine the color shades, with each pixel having three color channels (Red, Green, Blue) and 256 intensity values per channel.

Results: The total pixel count for the full image (heart and surrounding region) was 1600x1200 for Patient 1, 480x624 for Patient 2, and 480x848 for Patient 3. The heart regions contained 218,864 pixels (Patient 1), 44,020 pixels (Patient 2), and 77,919 pixels (Patient 3). The EAT areas were found to be 158,213 pixels (Patient 1), 35,608 pixels (Patient 2), and 52,723 pixels (Patient 3). The percentage areas of the sub-regions varied, with RV and PA Epicardial Fat comprising 72.3%, 80.9%, and 67.7% of the heart regions for Patients 1, 2, and 3, respectively. The top 100 color shades were identified, with unique colors in the Aorta (23,323), Appendage (7,030), Epicardial Fat (80,257), and Myocardium (10,131).

Conclusion: The study demonstrated that EAT and the color shades of heart sub-regions could be accurately quantified using advanced imaging and computational techniques. These findings provide valuable insights into the correlation between EAT and cardiac risk factors, enhancing the ability to predict postoperative morbidity and mortality and enabling early interventions to improve patient outcomes.