Efficient detection of glaucoma using double tier deep convolutional neural network

Glaucoma is an ocular disease which causes the eyes’ optic nerves to suffer from irreversible blindness because of increased intraocular pressure. Early detection and glaucoma screening can prevent loss of vision. A common way to diagnose the progression of glaucoma is through examination by a special ophthalmologist of the dilated pupil of the eye. But this approach is difficult and takes a lot of time, so automation can solve the problem with the concept of double tier deep convolution neural networks. This network is well suited for resolving this type of problem, as it can deduce hierarchical data from the image that allows them to discern among glaucoma and non-glaucoma diagnostic patterns. It is composed with different layers. Every layer contains hidden layers like turbidity, max pooling, output and the fully connected layer. The retinal images are processed in the hidden layers, and the results obtained are combined and classified as normal or glaucomatous. The efficiency of the double tier deep convolutional neural network has been compared with different existing recognition methodologies. The outcomes show that the double tier deep convolutional neural network gives better performance when compared with other methodologies in terms of accuracy of 92.64%, sensitivity of 92.18%, specificity of 91.20% and precision of 90.76%.