Validation of a New Scoring System for the Detection of Early Forme of Keratoconus

Abstract

Purpose: To evaluate the accuracy of a new objective method for the detection of ectasia susceptible eyes. Methods: One hundred and eighty-three elevation and placido topographies were retrospectively evaluated by one experimented refractive surgeon and classified as 'normal' or 'at risk for LASIK'. An objective automated system built on the combination of topography and tomography data in a discriminant function was also used to classify the corneas. The concordance between the objective and the subjective classification was evaluated and the usefulness of the objective scoring system was assessed by receiver operating characteristic (ROC) curve analysis. Results: The mean age of the studied group was 37 ± 8 years old. One hundred and fifty-nine eyes were subjectively classified as 'normal' and 24 as 'At risk for LASIK'. The scoring system correctly classified 153 eyes as 'normal' and 22 eyes as 'at risk for LASIK'. Six eyes were wrongly detected as 'at risk' by the automated system (false-positive) and two eyes were wrongly classified as 'normal' (false-negative). The sensitivity and specificity of the automated system were 92 and 96% respectively. Conclusion: An automated system built on the combination of topography and tomography parameters can help in creating a sensitive and specific artificial intelligence for the detection of corneas at risk for refractive surgery.