Principal Component Analysis of Consecutively Measured Wavefront Aberrations

Abstract

Abstract: : Purpose: In order to compress continuous wavefront measurement data, principal component analysis (PCA) was used. Methods: Ocular wavefront aberration measurements were performed every one second for one minute on two eyes of two normal subjects, TT and YH. The subjects were forced to blink every 10 seconds. Data was collected with a Hartmann–Shack wavefront aberrometer (Topcon, Tokyo) of which the lenslet size was 0.22 mm square on the pupil and the focal length was 5 mm. A fixation target was displayed to stabilize eye movement and accommodation. Wavefront aberrations were obtained using Zernike polynomials up to 6th order. PCA based on a correlation matrix was performed on the wavefront measurements for each eye. The wavefront measurement data consisted of 60 sequential measurements of 25 variables (2nd to 6th order Zernike coefficients). Results: 25 principal components were obtained for each eye by PCA. Information of the changes was compressed in several principal components. For TT, the contribution of first five principal components to the wavefront aberration change was more than 90%. For YH, nine principal components contributed 90% of change. The eigenvectors obtained by PCA were a complicated combination of Zernike modes. From visual observation of color coded map of wavefront aberration, the two components with the two largest contributions to the change had advanced wavefront in the upper part of the pupil and delayed wavefront in the lower part for TT. For YH, all components looked like noise. The root means square (RMS) error of wavefront aberrations for TT tended to increase after blink, while the RMS error for YH was stable. TT's principal components could be explained by assuming that the tear film was the cause of the wavefront change. Conclusions: PCA reduced the information of consecutive measurements of wavefront aberrations. It also appears that cause of the change of wavefront aberrations can be analyzed using PCA.

Keywords: cornea: tears/tear film/dry eye • eye movements • topography