Diagnostic Accuracy of Spectralis SD OCT Automated Macular Layers Segmentation to Discriminate Normal from Early Glaucomatous Eyes
Diagnostic Accuracy of Spectralis SD OCT Automated Macular Layers Segmentation to Discriminate Normal from Early Glaucomatous Eyes
复制标题
DOI：
10.1016/j.ophtha.2017.03.044
复制
发表时间：
2017-08-01
期刊：
影响因子：
13.7
通讯作者：
Anton, Alfonso
中科院分区：
文献类型：
--
作者：
Pazos, Marta;Anna Dyrda, Agnieszka;Anton, Alfonso
目的：为了评估Spectralis光谱域（SD）光学相干断层扫描（OCT）装置（Heidelberg Engineering，Inc.，海德堡，德国）区分健康和早期青光眼（EG）眼睛。设计：前瞻性、横断面研究。参与者：包括40只EG眼睛和40只健康对照。方法：使用Spectralis OCT设备的标准后极和乳头周围视网膜神经纤维层（pRNFL）协议对所有参与者进行检查。使用黄斑水平的早期治疗诊断视网膜病变研究圈，应用自动视网膜分割软件来确定以下参数的厚度：总视网膜厚度、视网膜内层（IRL）、黄斑视网膜神经纤维层（mRNFL）、黄斑神经节细胞层（mGCL）、黄斑内丛状层（mIPL）、黄斑内核层（mINL）、黄斑外丛状层（mOPL）、黄斑外核层（mONL）、光感受器（PR）和视网膜色素上皮（RPE）。通过将mRNFL、mGCL和mIPL参数相加来确定神经节细胞复合体（GCC），并且通过结合mGCL和mIPL参数来确定神经节细胞层-内丛状层（mGCLIPL）。比较各组各层的厚度，并确定具有最佳受试者工作特征曲线（AUC）下面积的层和扇区。主要指标：比较各组不同面积的pRNFL、IRL、mRNFL、mGCL、mIPL、mGCC、mGCL-IPL、mINL、mOPL、mONL、PR和RPE参数以及视网膜总厚度及其相应AUC。在EG组中，在所有评估的6个扇区中，毛细血管周围RNFL均明显变薄（P < 0.0005）。对于黄斑变量，EG组的视网膜总厚度、mIRL、mRNFL、mGCL和mIPL的视网膜厚度显著降低。区分两组的2个最佳分离参数是pRNFL（AUC，0.956）和mRNFL（AUC，0.906）。当mRNFL，mGCL，和mIPL测量相结合（mGCC和mGCL加上mIPL），那么它的诊断性能提高（AUC，0.940和0.952，分别）。结论：黄斑RNFL，mGCL-IPL，和mGCC测量显示了很高的诊断能力，以区分健康和EG参与者。然而，黄斑视网膜内测量仍然没有克服标准pRNFL参数。
Purpose: To evaluate the accuracy of the macular retinal layer segmentation software of the Spectralis spectral-domain (SD) optical coherence tomography (OCT) device (Heidelberg Engineering, Inc., Heidelberg, Germany) to discriminate between healthy and early glaucoma (EG) eyes.Design: Prospective, cross-sectional study.Participants: Forty EG eyes and 40 healthy controls were included.Methods: All participants were examined using the standard posterior pole and the peripapillary retinal nerve fiber layer (pRNFL) protocols of the Spectralis OCT device. Using an Early Treatment Diagnostic Retinopathy Study circle at the macular level, the automated retinal segmentation software was applied to determine thicknesses of the following parameters: total retinal thickness, inner retinal layer (IRL), macular retinal nerve fiber layer (mRNFL), macular ganglion cell layer (mGCL), macular inner plexiform layer (mIPL), macular inner nuclear layer (mINL), macular outer plexiform layer (mOPL), macular outer nuclear layer (mONL), photoreceptors (PR), and retinal pigmentary epithelium (RPE). The ganglion cell complex (GCC) was determined by adding the mRNFL, mGCL, and mIPL parameters and the ganglion cell layereinner plexiform layer (mGCLIPL) was determined by combining the mGCL and mIPL parameters. Thickness of each layer was compared between the groups, and the layer and sector with the best area under the receiver operating characteristic curve (AUC) were identified.Main Outcome Measures: Comparison of pRNFL, IRL, mRNFL, mGCL, mIPL, mGCC, mGCL-IPL, mINL, mOPL, mONL, PR, and RPE parameters and total retinal thicknesses between groups for the different areas and their corresponding AUCs.Results: Peripapillary RNFL was significantly thinner in the EG group globally and in all 6 sectors assessed (P < 0.0005). For the macular variables, retinal thickness was significantly reduced in the EG group for total retinal thickness, mIRL, mRNFL, mGCL, and mIPL. The 2 best isolated parameters to discriminate between the 2 groups were pRNFL (AUC, 0.956) and mRNFL (AUC, 0.906). When mRNFL, mGCL, and mIPL measurements were combined (mGCC and mGCL plus mIPL), then its diagnostic performance improved (AUC, 0.940 and 0.952, respectively).Conclusions: Macular RNFL, mGCL-IPL, and mGCC measurements showed a high diagnostic capability to discriminate between healthy and EG participants. However, macular intraretinal measurements still have not overcome standard pRNFL parameters.