Abstract
Abstract: :
Purpose: To evaluate the discriminatory validity of three indices of corneal geometry for detecting subclinical keratoconus. Methods: Study was conducted in a high referral hospital in Mexico City (APEC). Patients were clinically evaluated and categorized into three groups by an expert : Keratoconus, Keratoconus Suspects and Healthy eyes. A corneal topography was performed with a Placido's ring-based system and diagnostic indices were obtained from the pathfinder computer algorithm: Corneal irregularity (CMI), Shape factor (SF) and Toric keratometry Mean (TKM). Analysis was performed in two steps : I.- Sensitivity and Specificity were calculated using : A.-The cutoff point defined by the system and B.- The individual cutoff point for each index: first at the normal value defined by the system and second at 1,2 and 3 standard deviations from the healthy sample, and II.- Discriminant Function Analysis (DFA) was performed. Results: Seventy two eyes were clinically diagnosed as keratoconus or suspects, sixty two were considered as normal. Sensitivity increased up to 95% when only keratoconus eyes were considered and, decreased to 82% when only suspects were included. TKM showed a sensitivity of 97% when it was analized by the normal value defined by the system (best sensitivity obtained) and showed a specificity of 96% when analized at two standard deviations from the healthy sample (best specificity obtained). TKM index had the highest weight (discriminatory ability) by DFA. The classification ability of the model was 95%, 37% and 80% for healthy, suspects and keratoconus eyes, respectively, with a total of 78.7% eyes correctly classified. Conclusions: Ring-based system topography appeared to be a reliable method for screening keratoconus. This system seems to be more sensitive for the detection of keratoconus than for keratoconus suspects. TKM had the highest weight in the discriminant function model.
Keywords: keratoconus • topography • imaging methods (CT, FA, ICG, MRI, OCT, RTA, S