Development of an OCT image analysis algorithm for differential diagnosis of retinal edema based on deep learning

The aim of this work is to develop an algorithm for differential diagnosis of retinal edema and study deep learning methods and their application to image analysis. The application of convolutional neural networks for the task of semantic segmentation of retinal layers is investigated and its efficiency is proved for two selected layers (pigment epithelium and retina). An algorithm of disease classification based on the intellectual analysis of the layers selected by the neural network is implemented. A proof of its applicability for differential diagnostics of retinal edema is presented. The accuracy of disease detection amounts to 90%.