Diagnosis of Retinal Disorder by using Deep Learning Algorithm

Diagnosis of diseases related to the retina probably is the best preventive measures from blindness and vision loss. A method for automatically identifying and classifying retinal disorders using deep learning-based techniques on pictures involving optical coherence tomography is proposed. From the research undertaken, a deep learning model, Convolutional Neural Networks, was utilized in classifying diseases like glaucoma, diabetic macular edema, drusen, and choroidal neovascularization. To ensure robust performance, the model was tested with a training and evaluation dataset of 207,130 OCT pictures. Besides that, it was evaluated using certain key metrics such as accuracy, precision, sensitivity, and F1-score. Results were reportedly greater than 97% accuracy in most of the cases. The presented methodology consisted of preprocessing, feature extraction and classification utilizing logistic regression, neural networks, SVM, and kNN models.