糖尿病视网膜病变自动筛查方法研究进展

Research development of automatic diabetic retinopathy screening

糖尿病视网膜病变(diabetic retinopathy,DR)是糖尿病的严重并发症,是成年人视力下降甚至致盲的主要原因之一。适当的眼底筛查方法可预防糖尿病视网膜病变的发生,有利于疾病的早期诊断和治疗,有效降低失明率。自动的眼底图像处理是糖尿病视网膜病变的一种重要初步筛查方式,不仅可以减少人为误差,也大大减轻了眼科医生的负担。本文回顾了基于视网膜眼底数码图像的糖尿病视网膜病变自动筛查方法的研究进展,讨论了文献中多种现有技术的结果和不足,展望了其研究前景和面临的挑战,为糖尿病视网膜病变自动筛查算法的设计者提供了指导。

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.