Innovative Methods and Applications of Soft Computing in Automated Diagnosis and Damage Assessment of Eye Disorders

Abstract

xi newlineAbstract newlineSince almost a decade, the detection, diagnosis and severity assessment of human visual system disorders has received a great momentum. This is due to astonishing exponential development in image processing techniques. In order to account for diversity and complexity of several intricate functionalities of human eye varied detection system, methodologies and algorithms have been developed. With such systems ophthalmologist can precisely detect exclusively the portion of the eye, the kind of ailment and severity with in no time. As a comparison the physical testing is time consuming and in certain cases of eye disorder it might be extremely hard for a precise diagnosis. Several points merit their consideration in support of computer aided detection systems. To mention a few, they provide accuracy, consistency; provide a high level of confidence in the mind of ophthalmologist besides reducing several visits of a patient in turn reducing the possible long haul treatment costs. newlineThis thesis reports, an exhaustive research done in developing automated eye disorder detection and severity assessment systems. The research documented in this thesis has a four pronged approach. newlineFirstly, development of automated detection systems for maculopathy. Maculopathy is a disorder which will affect the direct vision due to high concentration of hard substance that oozes out from blood vessels. To develop this system around 150 retinal images were collected from public databases as well as local eye hospitals. Three systems were developed; the one, based on conventional method developed using the data base of feature values. The second one is the back propagation neural network [BPNN] based and the third one is the fuzzy inference system [FIS] aided. All the three systems have one thing in common that they categorise the given input image into any one class of severity [the details of modalities, the system development procedure, and the attribute elicitation are detailed in chapter 2]. The three systems so developed were