北卡指数对青光眼的诊断性能评估  

Evaluation of the Diagnostic Performance of the University of North Carolina Optical Coherence Tomography Index for Glaucoma

作　　者：杨丽丹 李青蒨 陈倩茵[3] 马红婕[4] 林晨 Lidan Yang;Qingqian Li;Qianyin Chen;Jacey Hongjie Ma;Chen Lin(Aier School of Ophthalmology,Central South University,Changsha 410035,China;Yantian District People's Hospital,Shenzhen 518081,China;Guangzhou Aier Eye Hospital,Guangzhou 510040,China;Changsha Aier Eye Hospital,Changsha 410015,China;Shenzhen People's Hospital,Shenzhen 518020,China)

机构地区：[1]中南大学爱尔眼科学院,长沙410035 [2]盐田区人民医院,深圳518081 [3]广州爱尔眼科医院,广州510040 [4]长沙爱尔眼科医院,长沙410015 [5]深圳市人民医院,深圳518020

出　　处：《中华眼视光学与视觉科学杂志》2024年第5期327-334,共8页Chinese Journal Of Optometry Ophthalmology And Visual Science

基　　金：湖南省临床医疗技术创新引导计划(2017SK50903);深圳市知识创新计划基础研究项目(JCYJ20170307155030786)

摘　　要：目的:验证北卡罗来纳光学相干断层扫描成像指数算法(北卡指数)在中国青光眼人群中的诊断效率,并对其可重复性进行评估。方法:回顾性病例对照研究。连续收集2016年1月至2020年6月长沙爱尔眼科医院,以及2021年11月至2023年3月广州爱尔眼科医院的临床数据库中所有符合纳入标准和图像质量要求的非青光眼受检者[共62例(114眼),分为非高度近视组68眼、高度近视组46眼]和青光眼患者[共62例(91眼),分为视野平均缺损(MD)<-8 dB组54眼、MD≥-8 dB组37眼(其中MD≥-4 dB组29眼)]的临床数据资料。非青光眼受检者均经过眼健康体检及青光眼排查;青光眼患者通过房角镜和视野等系列检查确诊为原发性开角型青光眼或原发性慢性闭角型青光眼。将所有研究对象的光学相干断层扫描(OCT)参数,包括4个象限和平均视网膜神经纤维层(RNFL)厚度、垂直杯盘比(VCDR)、平均杯盘比和盘沿面积及6个分区、平均和最小视网膜神经节细胞内丛状层(GCIPL)厚度,导入北卡指数算法模型进行运算,得出青光眼的预测概率。使用受试者工作特征曲线下面积(ROC-AUC)评估诊断性能;采用DeLong检验比较北卡指数与各OCT参数间ROC-AUC的差异。另外,同一位OCT操作者使用OCT对2020年12月至2021年1月于长沙爱尔眼科医院门诊行眼健康体检的正常受检者32例(64眼)(作为正常组)进行重复扫描,运用类内相关系数(ICC)评估该仪器及北卡指数的可重复性。结果:在区分非高度近视眼和青光眼方面,北卡指数的ROC-AUC值分别为0.92(MD≥-8 dB青光眼组)、0.95(MD≥-4 dB青光眼组),表现最佳的OCT参数单变量为:最小GCIPL厚度为0.62(Z=5.10,P<0.001)、0.60(Z=5.56,P<0.001);下方RNFL厚度为0.67(Z=4.48,P<0.001)、0.62(Z=5.18,P<0.001);VCDR为0.84(Z=1.91,P=0.057)和0.83(Z=2.70,P=0.007)。在区分高度近视眼和MD≥-4 dB的青光眼方面,北卡指数的ROC-AUC值为0.93,与VCDR和平均杯盘比差异无�Objective:To validate the performance of the University of North Carolina Optical Coherence Tomography(UNC OCT)index for glaucoma in Chinese and to assess its reproducibility.Methods:This was a retrospective case-control study.Data of non-glaucoma subjects(62 case,114 eyes:68 non-highly myopic eyes and 46 highly myopic eyes)and glaucoma subjects[62 case,91 eyes:the numbers of glaucoma with mean deviation(MD)<-8 dB,MD≥-8 dB were 54 eyes,37 eyes(and MD≥-4 dB were 29 eyes)]who met the inclusion criteria and image quality requirements were collected consecutively from January 2016 to June 2020 in Changsha Aier Eye Hospital and November 2021 to March 2023 in Guangzhou Aier Eye Hospital.Non-glaucoma subjects all underwent complete eye health examination and glaucoma screening,and glaucoma subjects were those who had been diagnosed with primary open angle glaucoma and primary chronic closed angle glaucoma through atrial angioscopy,visual field,and other examinations.The optical coherence tomography(OCT)parameters,including four quadrants and average retinal nerve fiber layer(RNFL)thickness,vertical CD ratio(VCDR),average cup-to-disc area ratio,and disc rim area,and six sectoral,average,and minimum retinal ganglion cell-inner plexiform layer(GCIPL)thickness was calculated through the UNC OCT index algorithm.Diagnostic performance was assessed using the area under the receiver operating characteristic curve(ROC-AUC);and the DeLong test was used to compare the difference between single parameters and the UNC OCT index.In addition,32 normal subjects(64 eyes)who underwent eye health examination at the outpatient clinic of Changsha Aier Eye Hospital(as the normal group)from December 2020 to January 2021 were repeatedly scanned by the same operator with OCT.The reproducibility of OCT and the UNC OCT index was evaluated using the intraclass correlation coefficient(ICC).Results:When distinguishing between non-highly myopic eyes and glaucoma,the ROC-AUC of the UNC OCT index were 0.92(glaucoma with MD≥-8 dB)and 0.95(glaucoma

关 键 词：青光眼 光学相干断层扫描 算法 诊断性能 可重复性 

分 类 号：R775[医药卫生—眼科]