AN IMAGE PROCESSING AND NEURAL NETWORK APPROACH FOR GLAUCOMA IDENTIFICATION

Glaucoma is a chronic eye disease, where a loss of vision occurs as a result of progressive optic nerve and astrocytes damage caused by high intraocular pressure (IOP). It is the second major cause of visual impairment and blindness worldwide. Early detection of the disease is critical for stopping the progression toward the complete vision loss. Due to the complex and diverse nature of disease pathology of glaucoma, its diagnosis heavily relies on the experience of glaucoma expert ophthalmologist. It is important to detect glaucoma in its early stages so that a patient’s vision can be preserved. Recent advances in image processing and associated machine learning techniques have allowed to design algorithms which can automate and accurately detect glaucoma from fundus images. This paper presents an image enhancement, segmentation and denoising approach for fundus images. Further statistical features of the image have been computed which would aid the detection of glaucoma. A deep neural network is subsequently used for classification which attains an accuracy of 98%. A comparison with existing techniques show that the proposed work outperforms baseline techniques in terms of classification accuracy.