Diabetic retinopathy (DR) is a severe complication of diabetes and a leading cause of vision loss and even blindness for adults.Appropriate retinal screening can prevent the incidence of DR and be helpful for the early diagnosis and treatment,reducing the rate of blindness effectively.Automated retinal image processing is an important primary screening method for diabetic retinopathy,which reduces not only human error but also the burden on ophthalmologists.This paper reviews the research development of automatic diabetic retinopathy screening based on retinal digital fundus images,and discusses the results and flaws of current techniques in literatures.The future potential and challenges for research are also discussed to provide guidance to algorithm designers of the automatic detection of DR.
|