ANALYSIS OF ATHEROSCLEROTIC PLAQUE IN CAROTID ARTERIES BY USING CONVOLUTIONAL NEURAL NETWORKS

The main aim of this study was to identify and classify the atherosclerotic plaque components such as lipid core, fibrous and calcified tissue, by applying the Convolutional Neural Networks (CNNs) on Ultrasound (US) imaging data. The U-Net, SegNet, and Pyramid Scene Parsing Network (PSPNet) architectures for multi-class image segmentation have been applied. Four common classification metrics are considered for one-vs-all quantitative evaluation, including accuracy (ACC), precision (P), recall (R) and F1-score. Our models showed a good accuracy in prediction of lipid core, fibrous and calcified carotid tissue based on US images. In the era of personalized medicine, plaque image analysis has the potential to extract valuable information about the plaques and thus to identify more accurately the patients at risk of plaque rupture. This type of computational methods can pre-estimate the risk of cardiovascular disease and stratify patients as a high/low risk, detecting the stabile and vulnerable atherosclerotic plaques