自动化分割算法分析RTVue100 OCT黄斑区视网膜各层厚度的重复性和再现性研究

Repeatability and reproducibility of macular thickness profiles of intra-retinal layers determined by an automated algorithm with RTVue100 OCT

背景视网膜各层厚度的检测和评估对多种眼科疾病的诊断、治疗及预后的监测至关重要，频域光学相干断层扫描（OCT）是目前常用的测量工具。以往的测量方法采用软件分析法，但关于视网膜自动分层软件可靠性的研究较少。目的评估自动化分割算法分析RTVue100 OCT测量黄斑区视网膜各层厚度的重复性和再现性。方法采用横断面研究设计，纳入正常受试者18人18眼，用RTVue100 OCT拍摄受试者右眼以黄斑中心凹为中心6mm扫描区的视网膜各层厚度图像，采用自动化分割算法将视网膜图像分割成神经纤维层（RNFL）、神经节细胞层和内丛状层（GCL＋IPL）、内核层（INL）、外丛状层（OPL）、外核层（ONL）、光感受器内节（IS）、光感受器外节（OS）、视网膜色素上皮（RPE）层，并用Matlab软件进行分析，通过将自动探测得到的相邻两条边界相对应的位置相减获得结果。每个受检眼先由一位操作者连续拍摄2次，然后由另一位操作者拍摄1次，采用组内相关系数（ICC）和重复性系数（COR）分析测得结果的重复性和再现性。结果RTVue100 OCT测量得到水平方向视网膜全层厚度为（303．22±14．10）μm，垂直方向为（306．68±13．32）μm。视网膜各层结构中，最厚的为GCL＋IPL以及ONL。同一检查者测量2次的重复性结果和不同检查者之间的再现性结果均表明，无论在水平方向还是在垂直方向，OPL、IS和OS的ICC和COR均〈0．60，而RNFL、GCL＋IPL、INL、ONL和RPE层的ICC和COR均〉0．70。结论自动化分割算法分析RTVue OCT图像是一种可靠的工具，可提供重复性的OCT视网膜厚度资料，有利于视网膜疾病的诊断和监测。

Background Evaluation of intra-retinal layer thickness plays an important role in the diagnosis and monitor of various eye diseases, and spectral domain optical coherence tomography （OCT） is a frequently used tool. Software analysis method was used to measure the retinal thickness in previous study, but the study on the reliability of automatic layered software is lack. Objective This study was to evaluate the repeatability and reproducibility of thickness profile measurement of intra-retinal layers determined by an automated algorithm applied to OCT images from RTVue100 OCT instrument. Methods In this prospective cross-sectional study, retinal thickness images at 6 mm around fovea were obtained from 18 right eyes of 18 normal subjects with RTVue100 OCT instrument. The retinal images were segmented into retinal nerve fiber layer （RNFL） , ganglion cell layer and inner plexiform layer（GCL＋IPL） ,inner nuclear layer （INL）, outer plexiform layer （OPL）, outer nuclear layer （ONL）, inner segment （ IS）, outer segment （OS） and retinal pigment epithelial （RPE） layer using automated algorithm method. Then Matlab software was used to analyze the measuring outcome. Interclass correlation coefficient （ICC） and coefficients of reproducibility （COR） were calculated from the results of two-time examination by the same examiner to evaluate the repeatability and from the results of two different examiners to assess the reproducibility. Written informed consent was obtained from each subject prior to any medical procedure. Results The entire retinal thickness measured by RTVue-OCT was （303.22± 14. 10） μm in the horizontal meridian and （306.68± 13.32） μm in the vertical meridian, with the maximum values of retinal thickness in the GCL ＋ IPL and ONL. Whether in the horizontal meridian or in the vertical meridian,the ICC and COR were 〈0. 60 in the OPL, IS and OS ; while those in the RNFL,GCL＋IPL,INL,ONL and RPE layer were 〉0.70. Conclusions RTVue OCT with automated algorithm is a useful and reliable approach to the measurement of intra-retinal layer thickness. Automated segmentation can offer accurate and repeatable thickness profile of OCT retinal image. This method may improve the diagnosis and monitoring of retinal diseases.