Tan Bo, Graham Andrew D, Tsechpenakis Gavriil, Lin Meng C
*PhD †MA ‡OD, PhD, FAAO Clinical Research Center, School of Optometry (BT, ADG, MCL), Vision Science Group, (MCL), University of California, Berkeley, Berkeley, California; and Department of Computer and Information Science, Indiana University-Purdue University Indianapolis, Indianapolis, Indiana (GT).
Optom Vis Sci. 2014 Jun;91(6):650-7. doi: 10.1097/OPX.0000000000000267.
To develop and test a novel quantitative method of describing the corneoscleral junction, including metrics that reflect both the angle and the topography in this region of the ocular surface.
Forty-eight neophyte subjects were recruited (16 Asian, 16 white, and 16 Latino). Optical coherence tomography images of the nasal, temporal, superior, and inferior quadrants in both eyes were taken. Custom image analysis software was written in Matlab to allow the observer to select a point defining the center of the junction, from which 20 concentric circles were automatically drawn. The surface of the junction in the image was automatically located by edge-detection routines, and the circles intersecting this edge defined a series of points in the Cartesian plane. A linear regression was fit to these points, and a set of metrics based on the regression residuals was calculated.
The sum of the squared orthogonalized residuals (SSRo) was the most repeatable metric and had the advantage of being unaffected by the orientation of the image. The SSRo was significantly greater in the nasal quadrant (p < 0.001), reflecting a more pronounced angle and/or rougher surface. The flattest and smoothest topography was found in the temporal quadrant. Whites had significantly higher SSRo than Asians and Latinos (p < 0.001).
This study presents a novel metric for characterizing the angle and topography of the corneoscleral junction using optical coherence tomography and establishes differences among quadrants and between ethnic groups.
开发并测试一种描述角膜缘的新型定量方法，包括反映眼表该区域角度和地形的指标。
招募了48名新手受试者（16名亚洲人、16名白人和16名拉丁裔）。拍摄双眼鼻侧、颞侧、上方和下方象限的光学相干断层扫描图像。用Matlab编写了定制的图像分析软件，以便观察者选择一个定义角膜缘中心的点，然后自动绘制20个同心圆。通过边缘检测程序自动定位图像中角膜缘的表面，与该边缘相交的圆在笛卡尔平面上定义了一系列点。对这些点进行线性回归拟合，并基于回归残差计算了一组指标。
正交化残差平方和（SSRo）是最具可重复性的指标，且不受图像方向的影响。鼻侧象限的SSRo显著更高（p < 0.001），反映出角度更明显和/或表面更粗糙。颞侧象限的地形最平坦、最光滑。白人的SSRo显著高于亚洲人和拉丁裔（p < 0.001）。
本研究提出了一种使用光学相干断层扫描来表征角膜缘角度和地形的新型指标，并确定了各象限之间以及不同种族之间的差异。
