Glaucoma classification using brownian motion and discrete wavelet transform

Glaucoma is a neurodegenerative disease that affects the eye. The early stages of glaucoma normally go unnoticed as initial vision loss affects the peripheral vision. However, glaucoma damage to the eye is irreversible and will eventually lead to blindness if the condition is not controlled. There are many ways to detect glaucoma and in this work we use digital fundus images. Textural features are extracted using fractal dimension and Brownian motion. A two dimensional level 2 discrete wavelet transform is performed on the images to extract energy and entropy values. All these features are then input into classifiers with the support vector machine of radial basis function kernel giving the best accuracy of 95.5%.