Sensitized Glaucoma Detection Using a Unique Template Based Correlation Filter and Undecimated Isotropic Wavelet Transform

Glaucoma is the most common cause of vision loss and its identification using image processing techniques is apparently becoming more important. This paper reports the development of an automated Glaucoma detection system based on image features of eye fundus photographs, which can be used to detect Glaucoma at an early stage. We have improved the sensitivity of Glaucoma detection by using Neuroretinal rim thickness, Neuroretinal rim area and vessel information of fundus image as additional features along with the cup-to-disc ratio feature that is normally used. A unique template based correlation technique using Pearson-r coefficients is employed to extract the features like cup-to-disc ratio, rim area and rim thickness. We have used vessel information as a new feature which is obtained by segmenting the vessels by employing an undecimated isotropic wavelet transform. Analysis of the extracted proposed features stored as a data base during each visit of the patient helps in monitoring the progression of the disease. An efficient methodology is developed showing promising results with better sensitivity and specificity in the classification of Glaucoma and healthy images, respectively.