Background: Corneal surface irregularities may limit the visual outcome after penetrating keratoplasty (PK). Corneal topographers mainly render empirically derived and system-specific statistical indices for characterization of superficial inhomogeneities which may lack clinical evidence and make inter-system comparisons difficult. The purpose of this study was to detect and quantify focal surface irregularities of the cornea after nonmechanical PK by 2-dimensional wavelet decomposition based on corneal topography data.
Patients and methods: Our study included 15 patients with keratoconus and 10 patients with Fuchs' dystrophy with all-sutures-out after penetrating keratoplasty. For trephination we used the excimer laser MEL60 (Aesculap-Meditec, Germany) (7.5/7.6 mm diameter in dystrophies, 8.0/8.1 mm in keratoconus, double-running 10-0 nylon suture). After suture removal a complete ophthalmological examination including OrbScan topography analysis (Orbtec, USA) was performed. The refraction data were extracted via "data recorder" and decomposed using 2-dimensional wavelet analysis methods (Daubechies-4-wavelets on five scales of resolution). Corneal irregularities were quantified (scale 1 = fine details to scale 5 = coarse details).
Results: All detail coefficients (horizontal, vertical and diagonal) correlated statistically significant with the "Irregular Astigmatism" provided by the OrbScan-system (p < 0.05). In scale 3 and 4 a relative maximum of the wavelet detail coefficients occurred, whereas the coefficients at scale 2 and 5 were significantly smaller. The horizontal and vertical detail coefficients correlated significantly inversely with the best-corrected visual acuity (p < 0.04). All detail coefficients were significantly lower in the patient group with keratoconus compared to Fuchs' dystrophy.
Conclusions: Wavelet decomposition of corneal topography refraction data allows an analytical isolation and quantification of focal corneal superficial irregularities. This algorithm is independent of the currently used topography system and allows a smoothing of the raw data set adapted to scale of resolution and data compression.