自动检测角膜塑形镜偏位量的算法研究

The study of Orthokeratology decentration using automatic algorithm

目的开发一种基于角膜地形图、客观、定量的算法,用于评估角膜塑形镜的偏位量。方法使用K-means算法对角膜地形图离焦环进行分析,并结合使用加权最小二乘法,在角膜地形图的离焦环区域上模拟出最适合椭圆形。将椭圆形的中点与瞳孔中心之间的距离定义为角膜塑形镜的偏位量。选用100例患者(200眼)对算法的准确性及可重复性进行评估,并与人工评估结果进行对比。结果在配戴角膜塑形镜后,114眼(57%)为轻度偏位,75眼(37.5%)为中度偏位,11眼(5.5%)为重度偏位。计算机算法结果与人工评估结果相比,有很高的匹配度和更好的可重复性。结论应用K-means及加权最小二乘法的算法对角膜塑形镜配戴后偏位量分析的准确性和重复性较高,为角膜塑形镜临床验配与研究提供了定量的自动分析方案。

Objective To develop an automatic and objective quantification algorithm based on corneal topography to evaluate the orthokeratology(Ortho-K)lens fitting.Methods K-means algorithm was used to analyze the defocus ring of corneal topography and fit the most suitable ellipse on the defocusing region of corneal topography combining with weighted least squares method.The decentration distance was defined as the distance between the pupil center and the center of ellipse.Corneal topography images from 200 eyes(100 patients)were used to evaluate the accuracy and repeatability of the algorithm.Results The decentration distance was calculated using this automatic algorithm.One hundred and fourteen eyes(57%)had mild decentration,75 eyes(37.5%)had moderate decentration and 11 eyes(5.5%)had severe decentration after wearing Ortho-K lenses.Compared with manual evaluation,the automatic algorithm showed higher accuracy and better repeatability.Conclusions Using K-means combining with weighted least squares algorithm,an objective and accurate algorithm was developed to evaluate Ortho-K fitting and could be used as an useful quantitative analysis tool for future clinical practice and research.