Texture-based Feature Extraction from Fundus Images for Glaucoma Diagnosis

Glaucoma is a retinal disease, which causes permanent vision loss when the disease increases to a severe stage. Glaucoma screening in the mild stage is necessary for good retinal health. The clinical instruments used in the medical department are manual and less accurate. In this paper, we proposed the texture feature-based computerized method for glaucoma classification. In this research work, the bi-dimensional empirical mode decomposition (BEMD) algorithm has been used for the decomposition of the preprocessed fundus photographs into intrinsic mode functions (IMFs). Then, texture-based features have been calculated from the decomposed IMFs. Then, ReliefF method is utilized for the selection of the useful features. Finally, least-square support vector machine (LS-SVM) classifier has been utilized for categorization purpose. The proposed method is evaluated on different openly available databases. The obtained parameters show that the developed method achieved better performance than the previously deployed methods on RIM-ONE database.