上海市北新泾社区糖尿病视网膜病变远程筛查系统的应用研究

The application study of community-based tele-screening system for diabetic retinopathy in Beixinjing Blocks, Shanghai

目的探讨社区糖尿病视网膜病变（DR）远程筛查系统的应用效果。方法横断面研究。以免散瞳数码眼底照相机和计算机网络技术为基础，构建社区DR远程筛查系统。采用随机数字表法抽取109例已进入筛查系统的糖尿病居民进行系统评价。评价内容：（1）经过培训的视力检查员和眼科医师对同一患者双眼远视力检查结果的一致性，采用配对≠检验进行分析；不同眼病的分析人员分别按该系统中免散瞳眼底照相的读片法和散瞳后应用检眼镜、裂隙灯显微镜联合90D非接触镜检查法（传统检查法），对同一患者双眼DR诊断和分级结果的一致性进行评价，采用kappa值和组内相关系数进行分析；（2）对眼底照相获取的图像进行不同比例的压缩，通过该系统网络传输后，采用组内相关系数分析压缩后的图像质量，获得合适的压缩比例；（3）计算筛查系统运行所需的时问，并与传统检查法进行比较。结果109例（218只眼）受检者中，由眼科医师检查获取的视力分布情况：视力〈0．05者13只眼，视力0．05—0．3者61只眼，视力≥0．3者144只眼，与其他检查人员的视力检查结果比较，差异无统计学意义（t=-0．572，P=0．568）。采用传统检查法与眼底照相法诊断为DR的眼数相同，均为52只眼，两者间比较的kappa值为0．885，95％可信区问为0．807～0．963；DR不同分级诊断的一致性：组内相关系数为0．91，95％可信区间为0．85～0．94，表明两者的一致性好。将眼底照相获取的图像，以最大限度压缩至原始图像的15％（526x350像素）时，图像清晰度仍不受影响。每位受检者实时远程筛查所需的时间为5—7min，略少于传统检查法所需时间。结论社区DR远程筛查系统可满足DR患者筛查的需求。

Objective To design the community-based tele-screening system for diabetic retinopathy and evaluate the feasibility of it. Methods Cross-sectional study. The tele-screening system was based on non-mydriatic digital eye fundus camara photography and computer network technology. 109 type 2 diabetes mellitus residents were randomly selected for system evaluation, which included ： （ 1 ） The consistency of the far visual acuity examined by an ophthalmologist and a trained inspector, evaluated by paired t-sample test; the consistency of diagnosis of diabetic retinopathy by tele-screening system and traditional screening method by ophthalmoscope, slit-lamp bimicroscope combined with non-contact lens after pupil dilation, evaluated by kappa value and intraclass coefficient correlation. （2）The proper compression ratio of the fundus photographs transferring through the internet, evaluated by intraclass coefficient correlation. （3）The working time for the tele-screening on the residents, comparing with the traditional screening method. Results The visual acuities of the 218 eyes in 109 residents examined by an ophthalmologist were 〈 0. 05 in 13 eyes, between 0. 05 and 0. 3 in 61 eyes, f〉 0. 3 in 144 eyes. No significant difference was found between the vision acuity given by different examiner （ t = - 0. 572, P = 0. 568）. 52 eyes were diagnosed as DR by traditional screening method, while 51 eyes were diagnosed as DR by the tele-screening method, so in DR diagnosis, high consistency were found with kappa value as 0. 885, 95% CI 0. 807 to 0. 963, and in DR degree diagnosis with ICC value as 0. 91, 95% CI 0. 85 to 0. 94. The most compression ratio of fundus photographs was as low as 15% （526 × 350）. It took 5 to 7 minutes for the tele-screening system to examining and giving diagnosis of a diabetes mellitus resident, a little bit sooner than traditional screening method. Conclusions This community-based tele-sereening system can meet the requirements of mass screening for diabetic retinopathy.