Glaucoma Detection Using Fundus Images and OCT Images
7 Pages Posted: 7 Sep 2019 Last revised: 20 Oct 2019
Date Written: August 31, 2019
Glaucoma is a progressive and chronic eye disease which leads to the lack of peripheral vision and finally causes the permanent blindness. Glaucoma caused by the increased Intra Ocular Pressure (IOP) in the eye, which leads to the damage of the retinal nerve which is linked to the brain. The detection and identification of glaucoma in an image are important for controlling the decay of the vision. Currently used methods for the IOP measurement are pachymetry, gonioscopy, tonometer, etc. But the problem of these methods are the need for efficient medical practitioners and also the time consumption. So an efficient glaucoma detection system is presented here using retinal fundus images and Optical Coherence Tomography (OCT) images of the same eye images. K-means clustering and Otsu thresholding techniques are compared in the retinal fundus image for the structural feature analysis. Retinal Nerve Fibre Layer (RNFL) thickness is calculated from the OCT image. Support Vector Machine (SVM) with multi class is used for the classification of the fundus images and OCT images as glaucomatous, Non-glaucomatous and suspect for glaucoma. This work is applied on a public dataset and images collected from eye hospital containing glaucomatous, Non-glaucomatous and suspect for glaucoma. This proposed method gives accuracy of 90% for fundus images and 92% for OCT images. The identification of the glaucoma stage is achieved by comparing the results of both the retinal fundus image and the OCT image.
Keywords: Glaucoma, Optical Coherence Tomography, Intra Ocular Pressure, Retinal Nerve Fibre Layer, K-means clustering, Otsu segmentation, Support Vector Machine
Suggested Citation: Suggested Citation