Automatic Glaucoma Diagnosis in Digital Fundus images using Convolutional Neural Network

The World Health Organization (WHO) approximates that more than 42 million people are currently blind in the world, 80 percent of which could have been prevented or cured by early detection. According to a survey, Glaucoma is the second most leading cause for blindness after cataract. It is an irreversible eye disease, once the vision is lost it can not be recovered. Thus, it is vital to develop an automatic computerised tool to diagnose the disease. In this paper, an oveland robust deep learning based convolutional neural networks(CNN) architecture has been proposed to deal with the problem. The network consists of six convolutional layers, with various activation functions, and pooling layers to get the abstract and detailed information of the input image. The proposed architecture predicts the probability of an image being Glaucoma. The model has been experimented with Refugee and Drishti datasets. Our proposed model is able to diagnose the Glaucoma disease automatically with an accuracy of 95%, sensitivity of 100%, and specificity of 90% respectively.

I. Introduction

Globally eye diseases are increasing at an alarming rate. The top three eye diseases are Cataract, Glaucoma, and Age-related Macular degeneration(AMD). Talking about “Glau-coma”, which is the second leading cause of blindness in the world and number one blinding disease among African Americans, scientific evidences show that if early diagnosis and treatment is given during early stages, then permanent loss of vision can be prevented in at least 60% of the cases [1]. Thus there is a pressing need to develop a cost effective computer based automatic diagnostic system that can assist medical experts in early diagnosis, thus save their time and efforts wasted on the analysis of healthy people. Glaucoma is a chronic eye disease in which the optic nerve get gradually damaged, and thus lead to permanent loss of vision [2], [3]. It is also called the “silent thief of sight” because the loss of sight usually occurs over a long period of time. Eye experts use specialized devices for the eye monitoring such as a slit lamp or an ophthalmoscope to look at the back of the eye. These ophthalmologists are able to evaluate the health of an eye by looking at the various characteristics of cup and disc such as color, contour, and diameter and thus able to predict the presence of disease based on their experiences and knowledge. Usually, for the Glaucoma diagnosis, the two main indicators are the cup-to-disc ratio (CDR) and Inferior-superior-nasal-temporal(ISNT) area rule [4]and for the calculation of these parameters, it is necessary to segment the optic disc and cup simultaneously. Figure 1,2,3 shows some of the examples of healthy and glaucomatous retinal images from the database. Figure 2 clearly shows the glaucoma image with high cup-to-discratioandFigure3 showstheinferiorrimlossretinalimage respectively. As mentioned previously, glaucoma is frequently known as “the sneak thief of sight” because of gradual rise in intraocular pressure and damage of the optic nerve or vision. The disease shows no early symptoms in the majority of the cases [10]. Thus it is important to raise awareness among mass public for early detection and regular examination of eyes as this is still successfully controllable at the early stage if treated properly. Fig. 1.

Healthy image

Fig. 2.

High CDR(glaucoma)

Fig. 3.

Inferior rim loss(glaucoma)