应用近视正常值数据库后对近视性青光眼眼OCT颜色概率编码的评估
比较应用内置与近视正常值数据库后OCT颜色概率编码的一致性,并评估近视正常值数据库对近视性青光眼诊断能力的影响。研究纳入305只眼评估一致性,另120只眼评估诊断性能;应用近视正常值数据库后,象限RNFL厚度、钟点RNFL厚度和GCIPL厚度的特异度均提高(P值均<0.001),ROC曲线值也增加。
要点
- 应用近视正常值数据库后,象限RNFL厚度、钟点RNFL厚度和GCIPL厚度的特异度均高于内置数据库。
- 应用近视正常值数据库后,象限RNFL厚度、钟点RNFL厚度和GCIPL厚度的ROC曲线值增加。
- 共识性分析中,象限RNFL厚度、钟点RNFL厚度和GCIPL厚度的平均加权kappa系数分别为0.636、0.627和0.564。
核心结论
- 应用近视正常值数据库后,OCT颜色编码检测近视性青光眼的诊断能力显著提高。
- 在近视眼中使用OCT颜色概率编码时,需要实施近视正常值数据库以实现更精确的解读。
研究设计与样本
- 比较效度研究,纳入305只眼(154只近视健康眼和151只近视青光眼眼)评估一致性,另纳入120只眼(60只近视健康眼和60只近视青光眼眼)评估诊断性能。
- 基于近视健康眼获取近视正常值数据库,分别评估应用内置数据库与近视正常值数据库后OCT颜色概率编码之间的一致性。
主要发现
- 象限RNFL厚度、钟点RNFL厚度和GCIPL厚度的平均加权kappa系数分别为0.636、0.627和0.564。
- 近视正常值数据库在象限RNFL厚度、钟点RNFL厚度和GCIPL厚度上的特异度均高于内置正常值数据库(P值分别为<0.001、<0.001和<0.001)。
- 使用近视正常值数据库时,象限RNFL厚度、钟点RNFL厚度和GCIPL厚度的受试者工作特征曲线值增加(P值分别为0.011、0.004和<0.001)。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
PURPOSE: To evaluate the optical coherence tomography (OCT) color probability codes based on a myopic normative database and to investigate whether the implementation of the myopic normative database can improve the OCT diagnostic ability in myopic glaucoma. DESIGN: Comparative validity study. METHODS: In this study, 305 eyes (154 myopic healthy eyes and 151 myopic glaucoma eyes) were included. A myopic normative database was obtained based on myopic healthy eyes. We evaluated the agreement between OCT color probability codes after applying the built-in and myopic normative databases, respectively. Another 120 eyes (60 myopic healthy eyes and 60 myopic glaucoma eyes) were included and the diagnostic performance of OCT color codes using a myopic normative database was investigated. RESULTS: The mean weighted kappa (Kw) coefficients for quadrant retinal nerve fiber layer (RNFL) thickness, clock-hour RNFL thickness, and ganglion cell-inner plexiform layer (GCIPL) thickness were 0.636, 0.627, and 0.564, respectively. The myopic normative database showed a higher specificity than did the built-in normative database in quadrant RNFL thickness, clock-hour RNFL thickness, and GCIPL thickness (P < .001, P < .001, and P < .001, respectively). The receiver operating characteristic curve values increased when using the myopic normative database in quadrant RNFL thickness, clock-hour RNFL thickness, and GCIPL thickness (P = .011, P = .004, P < .001, respectively). CONCLUSION: The diagnostic ability of OCT color codes for detection of myopic glaucoma significantly improved after application of the myopic normative database. The implementation of a myopic normative database is needed to allow more precise interpretation of OCT color probability codes when used in myopic eyes.
引用
- Seol BR, Kim DM, Park KH, et al. Assessment of Optical Coherence Tomography Color Probability Codes in Myopic Glaucoma Eyes After Applying a Myopic Normative Database. Am J Ophthalmol. 2017. PMID: 28911994. doi:10.1016/j.ajo.2017.09.010
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。