Deep Learning based ROI Segmentation using Convolution Neural Network

Abstract

Atherosclerosis is a chronic degenerative disease that results in cardiovascular diseases (CVDs) and is detected either by cardiac arrest or stroke. Early diagnosis of CVDs is made possible by identifying Intima Media Thickness (IMT) and elasticity. B-mode ultrasound imaging has on no account ionizing radiation and is economical and non-invasive to assess CVDs. This paper proposes an effective automatic image segmentation method using deep learning CNN for segmenting the region containing intima media of far wall carotid artery. The proposed approach is compared with SVM classifier and RBF neural network and is proven to be robust with improved accuracy and F1 score.