Increase in intraocular pressure (IOP) is one of the causes of glaucoma which can lead to blindness if not detected and treated at an early stage. Glaucoma symptoms are not always obvious; hence patients seek treatment only when the condition progressed significantly. Early detection and treatment will decrease the chances of vision loss due to glaucoma. This paper proposes a novel automated glaucoma diagnosis method using various features extracted from Gabor transform applied on digital fundus images. In this work, we have used 510 images to classify into normal and glaucoma classes. Various features namely mean, variance, skewness, kurtosis, energy, and Shannon, Rényi, and Kapoor entropies are extracted from the Gabor transform coefficients. These extracted features are subjected to principal component analysis (PCA) to reduce the dimensionality of the features. Then these features are ranked using various ranking methods namely: Bhattacharyya space algorithm, t-test, Wilcoxon test, Receiver Operating Curve (ROC), and entropy. In this work, t-test ranking method yielded the highest performance with an average accuracy of 93.10%, sensitivity of 89.75% and specificity of 96.20% using 23 features with Support Vector Machine (SVM) classifier. Also, we have proposed a Glaucoma Risk Index (GRI) developed using principal components to classify the two classes using just one number.
All Science Journal Classification (ASJC) codes
- Signal Processing
- Health Informatics