Automatic Identification of Diabetic Retinopathy in Fundus Images

Diabetic retinopathy (DR) is a condition where the retina is damaged due to fluid leaking from the blood vessels into the retina. In extreme cases, the patient will become blind. Therefore, early detection of diabetic retinopathy is crucial to prevent blindness. In this paper, a system for automatic identification of normal and abnormal retinal images is proposed through automatic detecting of the blood vessels, hard exudates microaneurysms, entropy and homogeneity. The measurements such as exudates area, blood vessels area, micro-aneurysms area, homogeneity and entropy are computed from the processed retinal images. These objective measurements are finally fed to the artificial neural networks (ANNs) classifier for automatic classification. Different approaches for Fundus image restoration are tested and compared. Furthermore, the effect of restoration on the automatic detection process is investigated. The proposed automatic identification system is proved to be reliable under a severe condition of blurred images and can achieve high detection rates with high fidelity features. The system can identify different stages with an average accuracy of more than 93.3 %, a sensitivity of more than 86.6 %, and a specificity of 100 %.