Application of Multiresolution Analysis for the Detection of Glaucoma

Increase in fluid pressure inside the eye damages the Optic Nerve Head (ONH) and may cause glaucoma. It may cause visual loss and blindness if it is not diagnosed at the early stage. Automated fundus image analysis and mass screening may help to detect the disease at early stage. In this work, we have developed an automated glaucoma detection system using relative energy and entropy features of Discrete Wavelet Transform (DWT). These features were tested for statistical significance using independent sample t-test. Then the selected features were fed to the set of supervised classifiers (Support Vector Machine (SVM), Decision Tree (DT), Fuzzy, k-Nearest Neighbour (k-NN), Naive Bayes (NB) and Probabilistic Neural Network (PNN)) to select the best classifier. Fuzzy classifier yielded the highest accuracy, sensitivity and specificity of 93.33%. The performance of the system can be further improved using better features and robust classifiers.