Accuracy of corneal elevation with four corneal topography systems

PURPOSE: To compare the elevation accuracy of placido-based videokeratoscopes with a raster-photogrammetry-based topography system using models of an aspheric surface and a simulated myopic ablation. METHODS: Test surfaces representative of normal corneas and corneas after excimer ablation were fabricated from a material measurable using both reflective (placido) and projective (raster-stereographic) systems. The "normal" surface was a rotationally symmetric asphere, while the simulated photorefractive keratectomy ablation had a central 5.5 mm diameter clear zone with a 1 mm blend radius to eliminate surface discontinuity. The surfaces were measured with their axes of symmetry aligned with the instrument axis, and again with their axes tilted downward by 10 degrees to produce a nonsymmetric shape. RESULTS: All of the topography systems performed reasonably well in measuring the aspheric surface, with root mean square elevation error ranging from 1.2 to 14.3 mu m. The raster-stereographic system had a root mean square error of 4.7 mu m on the simulated ablation, while the placido systems had root mean square errors of 10.6 to 29.5 mu m. Tilting the surfaces did not significantly alter the results. CONCLUSIONS: The tested placido videokeratoscopes do not provide reliable elevation data for shapes that have nonlinear changes in curvature. This may have serious implications if these systems are used for postoperative analysis or contact lens fitting on corneas that have been modified by refractive procedures.