后发性白内障计算机图像定量分析系统建立与评价

目的:建立一套后发性白内障计算机图像分析系统并对其进行评价.方法:建立家兔后发性白内障模型.充分散瞳后(瞳孔直径＞6.5mm),用裂隙灯显微镜拍摄系统(含数码相机、电脑)在同一条件下进行拍照.所得后照图像用于后发性白内障的评分.后发性白内障的评分(PCO score)=∑(OD×AI in CR)×2 ∑(OD×AI in OR),其中OD为后发障混浊密度(0到4级),AI为该级别后发障所占评估范围(以视轴为中心直径6mm的后囊膜)的面积百分比,CR为中央3mm区域,OR为外周3～6 mm环行区域.为检验系统的可信度,由6名检查者分别评估5幅图像,以检验系统的个体间差异;由同一名检查者在不同日期对5幅图像重复评估5次,以检验系统的个体内差异.结果:当后发障评分为0到1.5时,个体间可信度检验的标准差仅为0.05～0.087,Kruskal-Wallis非参数分析显示差异无统计学意义(H=0.314,P=0.9726).个体内可信度检验的标准差为0.041 ～0.067,差异也无统计学意义(H=0.613,P=0.9616).结论:该后发性白内障定量评估系统具有良好的可信度,方便、实用,可用于对后发性白内障进行比较精确的定量分析.为检验不同人工晶状体和手术技巧对后发性白内障形成及发展的作用提供了重要工具.

Evaluation of computerized image analysis system for quantification of posterior capsular opacification

AIM: To set up and evaluate a computerized image scoring system for quantification of posterior capsular opacification (PCO) that is independent of examinee.METHODS: The PCO model was made in rabbits. Following dilation of the pupil (＞6.5mm), standardized digital images of the PCO were obtained using slit-lamp photographic system(digital camera, computer included). PCO was scored by evaluating retroillumination photograghs. The PCO score was calculated by the formula: PCO score=∑ (OD×AI in CR)×2 +∑(OD×AI in OR), in which OD is opacity density (class 0 to 4),AI is the opacity area fraction, CR is the central region (diameter=3mm), OR is the outer region (3-6mm ring region).To evaluate the reliability of the scoring system, 6 examiners scored the pictures of 5 same eyes (interindividual reliability),and one examiner scored the pictures of 5 eyes on five different days (intraindividual reliability).RESULTS: The PCO scoring system was very reliable. With PCO score range from 0 to 1.5, the interindividual reliability showed standard deviation between 0.05-0.087. Kruskal-Wallis analysis showed no significant difference (H=0.314,P=0.9726). The intraindividual reliability showed standard deviation between 0.041-0.067, no significant difference either (H =0.613, P =0.9616).CONCLUSION: This PCO quantification analysis system evaluates the central area (diameter=6mm) of posterior capsule which is larger than visual acuity or other objective testing. The system revealed high reliability and insignificant investigator-dependent difference. With a standardized photographic setup, systematic errors by the photographic technique were not relevant. This system was proved to be an easy and useful tool to accurately evaluate PCO.