Fundus Eye Images Classification for Diabetic Retinopathy Detection Using Very Deep Convolutional Neural Network

Abstract

Diabetic retinopathy is an anomaly responsible for causing microvascular and macrovascular damage to the retina and occurs as a consequence of the worsening of diabetes. According to the World Health Organization (WHO), diabetic retinopathy is the most common cause of avoidable blindness in patients with diabetes worldwide. Early detection is important for the efficiency of treatments. Fundus Eye Image can be used to identify early disease development and monitor the patient’s clinical condition. The diagnostic process using this type of image may require some expertise from the ophthalmologist since not all retina anomalies are clearly visible. Thus, this paper proposes the development of a classification method based on Convolutional Neural Networks, but highly dense and deeper. The proposed method obtained a total of 92% AUC in the given experiments.