Deep Autoencoder Network for Automatic Classification of Echocardiograms Views

The standard views in echocardiography capture distinct slices of the heart which can be used to assess cardiac function. Determining the view of a given echocardiogram is the first step for analysis. Due to the expanding use of echocardiography by non-experts it would be useful to automate the analysis pipeline starting with view identification. Long axis echocardiogram clips were automatically classified using a neural network. To compensate for the small amount of data and to promote learning, the weights were pre-trained using deep autoencoders. This architecture showed reasonable accuracy when trained on a small dataset of echocardiograms even without preprocessing.