Abstract
Purpose :
Currently, the method for Orthokeratology (OrthoK) lenses evaluation relies on ophthalmologists’ experience and alters the normal distribution of tear films. Due to these disadvantages, we proposed a non-contact, objective and quantitative evaluation scheme for the OrthoK lens fitting based on corneal topographic data.
Methods :
Our evaluation scheme mainly have two assessment methods for analysis: the first one is combining K-means with weighted least squares and the other one uses boxplot instead of K-means algorithm. The treatment zone of OrthoK lens is fitting as an ellipse due to its shape. The procession of our scheme is shown as follows: K-means and boxplot are used to find out the isolated points from the set composed OrthoK treatment edge respectively. Then, the ellipse is computed by weighted least squares where the weights are determined by the results of K-means or boxplot. The decentration distance is the distance between the pupil center and the center of ellipse. Finally, we proposed a classification method based on the ratio between the decentration distance and major axis of the ellipse for the evaluation of lens decentration. The ratios obtained by the two methods are mapped to the level of decentration magnitude respectively. Moreover, Bland-Altman plot analysis is applied to analyze the consistency of the method with K-means and traditional one.
Results :
75 patients (150 eyes) were enrolled this study. The proposed methods have visualization results as shown in Figure 1 and Figure 2. The traditional results were obtained from experienced ophthalmologists. 86.67% eyes have the same results for the method with K-means and the traditional one. 81.33% samples have the same results in boxplot’s and the traditional method. Thus, the K-means method has better results comparing with boxplot method. The Bland-Altman plot results show that the K-means method is consistent with the traditional one (P<0.05).
Conclusions :
Our results demonstrate the effectiveness of the K-means method that it is consistent with the traditional one. Moreover, the proposed evaluation scheme is non-contact, objective and quantitative for the evaluation of lens fitting.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.