Computer Based Identification of Classification of Eye Abnormalities – Cornea Using Optical Images

Abstract

"A Computer Aided-based system for classification eye diseases may prove vision-saving, as many severe eye diseases do not exhibit early warning sign before they begin to diminish vision. The incidence of ocular pathology is on rise with increasing aging population. The most common causes of eye disorder and visual impairment with resultant of aging in elderly are cataracts. This is followed by Iridocyclitis, which is an inflammation at the iris (the colour part of the eye) and of the ciliary body. Corneal Haze is another eye disorder where it is a complication of refractive surgery characterized a cloudiness of the usually clear cornea. For proper care and management of eyes, an automatic system is necessary to identify the eye disease which is the aim of this project.

This project presents a classification of four different classes of eye that made up of one normal and three diseases using neural network classifiers. Big Area Ring, Small Area Ring, Homogeneity, BWMorph and entropy are five features that are extracted from the raw images using the image processing techniques and fed to the classifiers for classification. In this work, the feedforward architecture for the neural network classifier has been used.

In this project, 151 subjects consisting of four different classes of eye conditions from different race and age have been used. A sensitivity of 92.86% with the specificity of 93.75% for these classifiers has been demonstrated. The systems are clinically readied to run on large amount of data sets. "