An Automatic System to Detect Exudates in Mobile-Phone Fundus Images for DR Pre-screening

Diabetic Retinopathy (DR) is an eye disease blamed to be a major cause of vision loss. However, this worst outcome is preventable if a patient is detected at the early stage. In this work, we make a system that detects an exudate which is one of abnormality signs of DR. The system is designed to work specifically on retinal images obtained from a mobile-phone with a portable retinal lens. Our system can achieve a true positive rate and accuracy are up to 96% and 82%, respectively. High accuracy, quick processing, portabililty, ease of use, and economy are main advantages of this system that are potentially accelerate pre-screening rate.