An Enhanced RNN-LSTM Model for Image Classification using Deep Learning Techniques

Abstract

Glaucoma is a leading ocular disease which majorly damages the optic nerve head (ONH) of the retina. The main cause for the glaucoma is intraocular pressure of the eye and due to this it may leads to complete or partial vision loss. The regular screening and the early detection of glaucoma is the only one solution to avoid from vision loss. The Computer Aided Diagnosis (CAD) technique helps in the diagnosis of glaucoma in the early stage using retinal fundus images. The proposed enhanced RNN methodology is applied to classify the images from normal and glaucoma images. The segmented region of the optic disc and optic cup obtained from the enhanced CNN technique is considered for RNN-LSTM classification model. This CNN-RNN-LSTM model helps in increasing the performance of the system to classify the images. The DRISHTI-GS database with ground truth images including data augmentation process helps to train and test the model. This helps to increase the performance of the model and finally achieved with a 96% accuracy.