Effect of translation and rotation fitting on analysis of corneal topography

The purpose was to assess the suitability of quadratic equations for the accurate representation of corneal topography and consider the effect of translation and rotation fitting on the quality of fit and the curvature results. Topography images were recorded for the anterior and posterior surfaces of 490 corneas of 490 myopic patients using Pentacam. Elevation data were fitted to four shape models, three of which considered translational and/or rotational fitting. Differences between the models in the estimates of radii of curvature (R) and asphericity coefficients (Q) and in the quality of fit (as measured by the root mean square (RMS) error and the structural similarity index (SSIM)) were statistically analysed. The general shape model that considered both translational and rotational misalignments provided the best fit for the anterior (RMS = 1.18 ± 0.56 µm, SSIM = 0.99 ± 0.01) and posterior (RMS 3.64 ± 1.23 µm, SSIM = 0.99 ± 0.01) corneal surfaces in all subjects. The quality of fit degraded significantly (with p < 0.01 in all cases) when misalignments were not considered, increasing RMS to 5.20 ± 2.27 µm (anterior) and 17.10 ± 6.08 µm (posterior) and decreasing SSIM to 0.84 ± 0.18 (anterior) and 0.68 ± 0.22 (posterior) when both translational and rotational misalignments were ignored. The estimates of Rx, Ry, Qx and Qy as obtained when both forms of misalignment were considered varied, respectively, by as much as 0.18 mm, 0.23 mm, 0.27 and 0.54 for the anterior surface, and 0.25 mm, 0.39 mm, 0.32 and 0.37 for the posterior surface when misalignments were ignored. The variations were statistically significant, with p remaining below 0.01 in all cases. In conclusion, consideration of geometric misalignments helps improve the accuracy of describing corneal topography. The effects of misalignments on the estimates of corneal radius and asphericity are statistically significant and may in some cases be clinically significant.