Efficient Detection of Glaucoma Using Statistical Feature of Fundus Image

The optical disk contains important physiological structures of eye abnormalities that are visible in the optical disk of the retinal fundus image. Glaucoma is the primary cause of blindness and incurable disease. Early identification of glaucoma in retinal fundus image is mandatory in order to reduce the possibility of blindness. Here proposed an algorithm to detect glaucoma in optical disk based on the texture, statistical and wavelet features. At first, the Otsu threshold is applied to localize the optical cup and the suitably cropped the Region of Interest (ROI) area. Then, the ROI image is enhanced by applying a gamma correction-based histogram equalization process. Now, extract the blood vessels and remove them from the ROI image. After that, the well-known Chan-Vese algorithm is applied to segment the optical nerve head (ONH). Finally, the feature extraction is done from the ONH and classified using Artificial Neural Network. To measure the effectiveness of the algorithm is examined and compared with the existing methods. The experiment results have shown that the proposed algorithm achieved better accuracy (98.21) than the existing methods.