Segmentation of Neuro Retinal Rim Area using Histogram Feature-based for Glaucoma Detection in Retinal Fundus Image

Abstract

Neuro retinal rim is an area between optic disc and optic cup. Neuro retinal rim area consists of four parts, namely inferior, superior, nasal, and temporal (ISNT). The diagnosis of glaucoma can be done by observing the shape of the neuro retinal rim. In normal eyes, the size of the four parts in the neuro retinal rim follows the ISNT rule, from the largest to the smallest, as follows: inferior, superior, nasal, and temporal. This research was conducted to examine an adaptive threshold of neuro retinal area segmentation. The neuro retinal rim area was obtained by reducing the optic disc and the optic cup parts. Before that, optic cup and optic disc was segmented by analyzing histogram features likes mean and standard deviation values. Normal image has the area from the largest to the smallest according to the ISNT rules, however, in glaucoma, it is usually the inferior and superior areas attacked first, so that they experience notching or narrowing and violate the ISNT rules. The accuracy of the system was 91.25%, with 73 images from DRISTHI-GS retinal image database successfully diagnosed with the ISNT area that experienced notching, while the 8 images were still failed to diagnose due to a poor level of brightness.