KLASIFIKASI PENYAKIT DIABETIC RETINOPATHY PADA CITRA FUNDUS BERBASIS DEEP LEARNING

Abstract

Diabetic Retinopathy is one of the complications of diabetes and if it is treated too late, the patient will experience permanent blindness. Diabetic Retinopathy cannot be detected directly. This is because the hallmark of Diabetic Retinopathy is on the retina of the eye and can only be detected by an ophthalmoscope which produces an image of the fundus. However, the stage of detecting and classifying the type of Diabetic Retinopathy using an Ophthalmoscope still takes a long time to get results, so a system that can detect Diabetic Retinopathy is needed quickly to detect Diabetic Retinopathy. The Diabetic Retinopathy detection system that will be built is a Deep Learning-based system by detecting the eye fundus image which will go through several stages of process such as preparing data, image training stage and image testing stage. The data set used is from the kaggle.com and strare sites. This system will detect and classify Diabetic Retinopathy based on Deep Learning based on the characteristics of the appearance of mycroaneurysms, hard exudates, soft exudates, and bleeding in the form of dots, lines, and spots on the retina of the eye. The results obtained from the learning process obtained an accuracy of 86.7% and an error of 13.3%. So it can be concluded that the googlenet architecture can classify diabetic retinopathy well.