Please use this identifier to cite or link to this item: http://hdl.handle.net/10266/4220
Title: Detection of Glaucoma Using Retinal Fundus Images
Authors: Garg, Disha
Supervisor: Singla, Sunil Kumar
Keywords: detection of gluacoma;retinal fundus images
Issue Date: 1-Sep-2016
Abstract: Now a days Computer aided systems along with image processing techniques are used for the diagnosis and screening of various abnormalities in human body. Fundus image processing or Fundus Photography is one such technique which is used to examine the fundus of the Eye and detect its diseased conditions. Several eye disorders, like Glaucoma, are asymptomatic at the early stages. If not treated at appropriate time, it may lead to loss of vision. In order to prevent any such damage, it needs to be detected as early as possible. The purpose of this research is the early detection of glaucomatous eye using the image processing techniques which are used to study size, shape and orientation of the fundus. The morphology of the fundus undergoes various changes in pathological conditions that can be examined with the help of Fundus Image processing. Fundus image consists of various features like Optic Disc, Optic Cup, Neuro-Retinal Rim tissue etc. The proposed methodology deals with extraction of Optic Disc (OD) and Optic Cup (OC) followed by calculation of the Cup-to-Disc Ratio. The image database used for this purpose has been collected from a local physician. By estimating areas of Optic cup and Optic disc, CDR ratio has been estimated. This ratio helps in isolating glaucomatous images from normal. After CDR calculation, Support Vector Machines (SVM) classifier using radial basis Gaussian function has been used in this methodology to classify the normal and glaucomatous eye. The accuracy of the algorithm has been found to be 98.33% and sensitivity is 100%.
URI: http://hdl.handle.net/10266/4220
Appears in Collections:Masters Theses@EIED

Files in This Item:
File Description SizeFormat 
4220.pdf1.91 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.