科研之友
微信
新浪微博
Facebook
分享链接
摘要
Glaucoma is an eye disease that will eventually lead to blindness if not diagnosed and treated early. Sight deterioration due to glaucoma is irreversible and the early treatment will help to stop further degradation of the eye. There are two main types of glaucoma: open angle and close angle. There is no known cause for these types of glaucoma and the more common is the open angle glaucoma with no initial symptoms. This is because glaucoma affects the peripheral vision and we normally see with our central vision. Hence it is important for regular eye check-ups especially for people aged over 40 years old. In this paper, we have used digital fundus images of normal and glaucoma subjects for the classification. The RGB (Red, Green, Blue) vector of the images are input into K-means clustering classifier to get the cluster centroids. Then the clinically significant features are fed to Support Vector Machine (SVM), K-Nearest Neighbour (KNN), Fuzzy Sugeno (FS) classifier, Decision Tree (DT) and Probabilistic Neural Network (PNN) to select the best classifier. Our results show that, FS classifier performed better with average accuracy of 92%, sensitivity of 85% and specificity of 99%.
- 出版日期2013-12
