Purchase this article with an account.
M.–E. Saint–Amour, J.–F. Laliberté, I. Brunette, J. Meunier; 3D Numerical Atlases of Human Cornea . Invest. Ophthalmol. Vis. Sci. 2004;45(13):2855.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: To create, analyse and validate three numerical atlases of corneal topography, i.e. (1) normal, (2) after LASIK to correct myopia, and (3) after LASIK to correct hyperopia. Methods: The data sample included 1084 corneal topographies (Orbscan II, Bausch & Lomb): 797 from normal corneas, 171 from corneas corrected for myopia, and 116 from corneas corrected for hyperopia. The atlases were created from 75% of the topographies in each group. The 25% left were used for the validation of the atlases. All topographies were normalized to a Best–Fit Sphere of 8 mm radius and the (0, 0, –6) point was arbitrarily defined as being the center. The atlases were built as weighted means of the different points constituting the topographies. Results:The means of the anterior surface elevation for the central point of the atlases were 2.0067±0.0055 mm for the normal corneas, 1.9921±0.0106 mm for the corneas corrected for myopia, and 2.0205±0.0119 mm for corneas corrected for hyperopia. The values were, as expected, lower for myopic surgeries and higher for hyperopia surgeries, compared to the normal unoperated cornea. Note that the standard deviations were small relative to the order of magnitude of the means. An analysis of variance over all points determined that the atlases were statistically different (ANOVA ; p = 0.001). Calculation of the cartesian distance allowed a correct classification in 78.5% of the normal topographies, 85.7% following refractive surgery for myopia and 57.1% following refractive surgery for hyperopia. Deeper testing methods, such as calculation of the Cartesian distance over different regions of the corneal topographies, improved the results to 93%, 92.9% and 71.4%, respectively. Conclusions:Three atlases of the corneal topography are proposed. Numerical atlases constitute an excellent source of information about the variability between corneal topographies from the same group and/or between groups.
This PDF is available to Subscribers Only