Eye Disease Detection Model Using Convolutional Neural Network with RMSprop, SGD, and Adam Optimizers

Eye disorders such as cataracts, glaucoma, and diabetic retinopathy can cause abnormalities in eye function, including blindness. This research aims to develop a machine-learning model and find the best model for detecting eye diseases. Deep Learning methods based on the Convolutional Neural Network (CNN) model will be used to create eye disease detection models by adding the RMSProp, SGD, and Adam optimizers techniques. The dataset used consists of 4 classes, namely normal (1074 data), glaucoma (1007 data), cataract (1038 data), and diabetic retinopathy (1098 data), totaling 4217 fundus images. The evaluation results reveal the CNN model with the RMSProp optimizer has better model performance with a value of 88% accuracy, 88% precision, 87% recall, and 88% f1-score, and based on the AUC value on the ROC-AUC curve for each class is above 0.90, which means that the model can differentiate data between classes well and is included in the Excellent classification category.