An investigation into the design of an automated glaucoma diagnostic system
In this thesis, an automated, computerized technique for facilitating the detection of glaucoma from fundus images is proposed and demonstrated. This process utilizes such techniques as pyramidal stereo-matching, fuzzy clustering, active contours, feature extraction, and others. High accuracy is achieved as evaluated through sensitivity and specificity calculations of the regions of interest within the digital fundus images. Furthermore, a review of such a system's weaknesses, necessary improvements, and areas of future research are included to complete the investigation.