Glaucoma classification based on texture and morphological features

Retina nerve fibre layer (RNFL) is indicated as an early sign of glaucoma. In fundus image, RNFL represented as a brighter bundle of fibres and presents in the surrounding of optic nerve head (ONH). Based on medical report, evaluation of early glaucoma can be done by measuring of ONH as morphological feature and analysis textures of RNFL loss. Loss of RNFL on glaucoma eyes is represented as a dark area and causes the change of ONH size. In this study, we proposed a scheme to classify glaucoma by using combination of texture and morphological features on a grayscale image excluding blood vessels and optic disc area. Gray level co-occurrence matrices (GLCM) and gray level run length matrices (GLRLM) is adopted to extract texture features. Extracted features were classified by support vector machine (SVM) and k-nearest neighbour (k-NN). The results show that k-NN (k=5) can classify images into glaucoma and normal with accuracy, sensitivity, and specificity are 88.3, 93.3, 83.3 respectively. It can be concluded that combination of texture and morphological features can be used to evaluate of glaucoma. © 2019 IEEE.