Comparison of three computerized videokeratoscopy systems with keratometry

To compare the reproducibility of computerized videokeratoscopy systems by using normal eyes and calibrated objects.

We evaluated the reproducibility of three commercially available videokeratoscopes [EyeSys, TechnoMed C-Scan, and PAR Corneal Topography System (CTS)] with the manual keratometer (Bausch & Lomb) by using calibration spheres and 10 normal subjects (20 eyes). All videokeratoscopy and keratometer results were obtained by one investigator (R.M.). Each eye and calibration sphere were submitted to 10 serial examinations by using each system. The average K of all points within the central 3.0 mm of the topography systems (central 3.0 mm) was compared with the average K of the manual keratometer.

All videokeratoscopy systems correlated well with each other and manual keratometry when accessing aspheric and spherocylinder calibration balls. EyeSys central keratometry clinical results had the strongest correlation with the average keratometry results at 35%, followed by PAR-CTS at 25% and C-Scan at 5%. Among the videokeratoscopy units, EyeSys and PAR-CTS had the strongest correlation at 65%. The correlation between the TechnoMed C-Scan and both the EyeSys and PAR-CTS systems was 25%. There was a statistically significant difference (p < 0.05) between the systems when analyzing the results obtained from clinical subjects. The average keratometry (K) difference of human eyes between videokeratoscopy systems is <0.35 diopters (D) (p < 0.05), which may be clinically significant. The average manual K reading (42.97 D) is statistically significantly flatter (p < 0.05) than each of the videokeratoscopy units (EyeSys = 43.49 D; PAR = 43.48 D; C-Scan = 43.83 D). Comparing the 10 measurements of each eye or calibration object in the same videokeratoscopy system verified that the devices give reproducible results. The average standard deviation (ASD) of the keratometer was 0.10 D. The ASD of the videokeratoscopy units was 0.05 D for the EyeSys, 0.29 D for the PAR-CTS, and 0.31 D for the C-Scan systems.

Based on this study, we should not assume that the results of different topography systems can be interchanged in clinical studies.