This paper presents the results of deep learning models trained for the task of segmenting cardiac structures in echocardiography images and discusses the potential visual aid these approaches can provide in stress echocardiography exams. In this study, models were developed to segment the left ventricle, left atrium, and myocardium using a public dataset containing 2D apical four-chamber and two-chamber view sequences acquired from 500 patients, with structures manually segmented by experts. The dataset was divided into training, validation, and testing sets. Additionally, we trained models for segmenting all three classes—left ventricle, left atrium, and myocardium—simultaneously, as well as models for segmenting only the left ventricle. Two different architectures were tested: U-Net and LadderNet. The segmentation performance was evaluated using Dice coefficient and Loss metrics against the ground truth of the test set, along with a visual assessment specifically on stress echocardiography images from cases provided by the University of Campinas Hospital. Although the results are preliminary, they demonstrate the capability of deep learning models for the task of echocardiography image segmentation and suggest the potential for these models to offer visual assistance in stress echocardiography exams.