In this paper, we analyze the application of U-NeXT deep learning architecture to segment echocardiographic structures. Automation of segmentation of cardiac structures is a difficult task due to problems such as variable tissue contrast, tissue artifacts, ultrasound speckle noise (acoustic interference) and varying position, shape and movement of the cardiac structures obtained in pathological conditions. Also, the echocardiographic data is dynamic and requires optimization and appropriate selection before segmentation. We have used Echo Dataset 2.1 having 408 images containing 2601 structures. The result on UNeXT trained architecture showed IoU 0.9200. The output segmentation masks were qualitatively assessed and presented. The best contours identification accuracy by the Dice coefficient of the Right ventricular endocardium and left ventricular epicardium were 0.9692 and 0.9843 respectively. The results validates that UNeXT architecture achieves state-of-the-art performance with faster inference times

CardiacUltrasound, Cardiac Segmentation,UNeXT,Deeplearning,Echocardiography, Computer vision