Retinal Disease Classification from Optical Coherence Tomographical Scans using Multilayered Convolution Neural Network

Classification of retinal diseases using Optical Coherence Tomographical (OCT) scans is a crucial task. Accurate detection and classification of these diseases is necessary for patient's survival. Presently, the analysis of retinal diseases are carried out by doctors by examining the OCT images regularly. However the manual diagnosis procedure is tedious. Therefore, in this paper, an automatic detection and classification technique of retinal diseases has been proposed to assist doctors in their diagnosis. A deep multilayered convolutional neural network (CNN) has been used to detect and classify the retinal abnormalities using OCT scans. The proposed technique has been applied on an open source retinal OCT dataset containing 59,142 images and 96.5% blind test accuracy has been achieved.