Epicardial Fat Segmentation Using Pyramid Neural Network: An Automated Approach for Cardiovascular Risk Assessment

This work focuses on the automated segmentation of epicardial fat using a Feature Pyramid Network (FPN) and subsequent classification of severity levels utilizing Support Vector Machine (SVM) models. Epicardial fat, an important biomarker for cardiovascular diseases, requires accurate segmentation for quantitative analysis. The proposed FPN architecture leverages multi-scale features for precise delineation of epicardial fat boundaries from cardiac imaging data. Following segmentation, an SVM-based classification system is implemented to categorize the segmented fat regions into low, high, and moderate severity classes based on specific features extracted from the segmented regions. The SVM models are trained and validated on labeled datasets with varying severity levels to ensure robustness and accuracy in classifying the severity of epicardial fat. The proposed framework exhibits promising results in both segmentation and severity classification, providing a reliable tool for automated assessment and risk stratification in cardiovascular disease diagnosis.