Predictive Ability of Galilei to Distinguish Subclinical Keratoconus and Keratoconus from Normal Corneas

Purpose: To determine the predictive ability of different data measured by the Galilei dual Scheimpflug analyzer in differentiating subclinical keratoconus and keratoconus from normal corneas. Methods: This prospective comparative study included 136 normal eyes, 23 eyes with subclinical keratoconus, and 51 keratoconic eyes. In each eye, keratometric values, pachymetry, elevation parameters and surface indices were evaluated. Receiver operating characteristic (ROC) curves were calculated and quantified by using the area under the curve (AUC) to compare the sensitivity and specificity of the measured parameters and to identify optimal cutoff points for differenciating subclinical keratoconus and keratoconus from normal corneas. Several model structures including keratometric, pachymetric, elevation parameters and surface indices were analyzed to find the best model for distinguishing subclinical and clinical keratoconus. The data sets were also examined using the non‑parametric “classification and regression tree” (CRT) technique for the three diagnostic groups. Results: Nearly all measured parameters were strong enough to distinguish keratoconus. However, only the radius of best fit sphere and keratometry readings had an acceptable predictive accuracy to differentiate subclinical keratoconus. Elevation parameters and surface indices were able to differentiate keratoconus from normal corneas in 100% of eyes. Meanwhile, none of the parameter sets could effectively discriminate subclinical keratoconus; a 3‑factor model including keratometric variables, elevation data and surface indices provided the highest predictive ability for this purpose. Conclusion: Surface indices measured by the Galilei analyzer can effectively differentiate keratoconus from normal corneas. However, a combination of different data is required to distinguish subclinical keratoconus.