Retinal Disease Classification from Retinal-OCT Images Using Deep Learning Methods

Abstract

Retinal diseases are the damage caused to any part of the retina. OCT images are used to diagnose retina related diseases. Cross sectional view of the retina is obtained through Optical Coherence Tomography (OCT). Medical disease prediction through images is time consuming and manual process. Computer vision technology and its progress have given a solution for the prediction and classification of medical diseases through images. The major aim of this research is to present a new deep learning-based classification model for automatically classifying and predicting different retinal disorders using OCT data. In this paper, retinal disease classification is performed using convolution neural networks (CNN). CNV, DME, DRUSEN, NORMAL images are the four classifications investigated in this study. Proposed Retina CNN architecture is compared with existing architecture for the detection of retinal disease. The best accuracy and better model prediction are obtained for the Retina CNN architecture with Adam optimizer with the training accuracy is 99.29 and 97.55% validation accuracy.