Glaucoma Detection in Fundus Eye Images using Convolutional Neural Network Method with Visual Geometric Group 16 and Residual Network 50 Architecture

Abstract

Glaucoma is an eye disease usually caused by abnormal eye pressure. One of the causes of abnormal eye pressure is blockage of fluid flow, which if detected too late can lead to blindness. Glaucoma can be identified by examining specific areas on the retina fundus image. The aim of this study is to detect positive and negative glaucoma in fundus images. The image data was obtained from the glaucoma_detection dataset, consisting of 520 images, including 134 glaucoma-infected images and 386 normal images. This study uses the Convolutional Neural Network (CNN) method with Visual Geometric Group-16 (VGG-16) and Residual Network-50 (ResNet-50) architectures. The research and testing results using the VGG-16 architecture obtained an accuracy rate of 78%, while using the ResNet-50 architecture obtained an accuracy rate of 80%.