Eye Disease Detection Using Transfer Learning on VGG16

Deep learning has emerged as a breakthrough technology in varied fields like health care, computer vision, natural language processing and many more. Ocular infections like diabetic macular edema (DME), choroidal neovascularization (CNV) and DRUSEN are commonly found eye diseases in humans and can lead to temporary or permanent loss of eyesight. The optical coherence tomography (OCT) technique is often used for the preliminary screening of mentioned ocular ailments and provides high resolution cross-sectional imaging. In this work, we have focused on classification of normal and abnormal optical coherence tomography by making use of visual geometry group (VGG16) convolution neural network (CNN) model for prompt diagnosis and timely proper medical treatment of the eye diseases mentioned. OCT image is high resolution imaging technique capable of capturing microstructures within human eye. Here, we endeavored to develop a CNN model for classifying OCT images into normal and abnormal category. Our model achieves an accuracy of 99% and precision of 98.8% which is quite improved results in comparison with other state-of-the-art works that we reviewed.