Sicam Victor Arni D P, Snellenburg Joris J, van der Heijde Rob G L, van Stokkum Ivo H M
Department of Physics and Medical Technology, VU University Medical Center, Vrije Universiteit, Amsterdam, The Netherlands.
Optom Vis Sci. 2007 Sep;84(9):915-23. doi: 10.1097/OPX.0b013e3181559d70.
A pseudo forward ray-tracing (PFRT) algorithm is developed to evaluate surface reconstruction in corneal topography. The method can be applied to topographers where one-to-one correspondence between mire and image points can be established.
The PFRT algorithm was applied on a corneal topographer designed and constructed at the VU University Medical Center, Amsterdam, The Netherlands. Performance of the algorithm was evaluated using artificial test surfaces and two sample eyes. The residual output of the PFRT algorithm is displayed as pixel displacements of actual feature points on the corneal image. Displacement of 1 pixel indicates submicrometer corneal height accuracy.
PFRT residual increases with complexity of the measured surface. Using Zernike radial order 6, the mean residual for the artificial surfaces is subpixel. The mean residual for the regular cornea and the irregular cornea is 1.16 and 2.94 respectively. To some extent, increasing the Zernike radial order improves the accuracy. The improvement from order 6 to 20 is factor 2.3 for the irregular cornea. Using the residuals to further improve the accuracy brought local changes as high as 0.28 D in some areas of the reconstructed corneal power map.
PFRT can be used to evaluate how close a reconstructed corneal surface is to the actual one. The residue information obtained from this algorithm can be displayed simultaneously with the corneal image. This provides accurate information about the corneal shape that is useful for application in laser refractive surgery.
开发一种伪前向光线追踪(PFRT)算法,以评估角膜地形图中的表面重建。该方法可应用于能在视标与图像点之间建立一一对应关系的地形图仪。
将PFRT算法应用于荷兰阿姆斯特丹VU大学医学中心设计并构建的角膜地形图仪。使用人工测试表面和两只样本眼评估该算法的性能。PFRT算法的残余输出显示为角膜图像上实际特征点的像素位移。1像素的位移表示亚微米级的角膜高度精度。
PFRT残余随着被测表面的复杂度增加而增大。使用泽尼克径向阶数6时,人工表面的平均残余为亚像素级。规则角膜和不规则角膜的平均残余分别为1.16和2.94。在一定程度上,增加泽尼克径向阶数可提高精度。对于不规则角膜,从阶数6到20的改善系数为2.3。利用残余进一步提高精度在重建角膜屈光力图的某些区域带来了高达0.28 D的局部变化。
PFRT可用于评估重建的角膜表面与实际表面的接近程度。从该算法获得的残余信息可与角膜图像同时显示。这提供了有关角膜形状的准确信息,对激光屈光手术的应用很有用。