Detection and Classification of Retinal Disorders using Deep Learning Techniques

Retinal disorders among patients may result in loss of vision or blindness. These disorders can be identified from Optical coherence Tomography (OCT) images of retina. OCT plays a vital role in retinal screening thereby diagnosing the retinal disorders well in advance to avoid huge loss. The traditional screening methods have the drawback of consuming more time due to annotations. To overcome these issues, a deep ensemble model, called EyeNet is proposed. EyeNet is a deep ensemble CNN model that aims in detecting the retinal disorders at an earlier stage. This model uses DenseNet121 and ResNet50 models. The Retinal Dataset obtained from OCT is used for experimental evaluation. Performance of the proposed EyeNet model is evaluated using Accuracy, Precision, Recall and F1 score.