Purchase this article with an account.
Victor A. Sicam, Harry de Vries, Maarten Huijbregtse, Michiel Mensink; Image Processing Of Irregular Corneas In Color-coded Multiple-point-source Corneal Topography (cmct). Invest. Ophthalmol. Vis. Sci. 2011;52(14):4189.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Color-coded Multiple-point-source Corneal Topography (CMCT) is a new technique which is validated to be superior to Placido based corneal topography in reconstructing irregular corneal surfaces (Sicam et. al. OVS 2006 & Snellenburg et. al. Optics Express 2010) . In extreme cases, image processing of CMCT taken for irregular corneas present some difficulties because of smearing and deformation of reflection patterns. In this investigation, an image processing algorithm is developed to enable a reasonable detection of feature points in the reflection pattern of irregular corneas.
The prototype topographer uses 672 LEDs as stimulator for corneal reflection. The LED pattern has a color code involving the variation in arrangement of three LED colors: RED, YELLOW and GREEN LEDs. This ensures that there is no mismatch in source and image points. A photo of an octafoil artificial surface made from PMMA (SUMIPRO BV, Almelo, The Netherlands) is obtained. To deal with the smearing and deformation of points in the reflection pattern an appropriate algorithm involving filtering techniques was applied to locate feature points in the corneal reflection. Surface reconstruction which is described in literature (Sicam et. al. JOSA A 2004) was applied and the amplitude of the peripheral modulations of the octafoil is measured and compared with manufacturer's specifications. A preliminary test on the robustness of the algorithms were applied on three normal eyes. The resulting corneal curvature were compared for surface reconstruction with and without the application ofthe filtering algorithm.
The reconstructed octafoil surface has an amplitude of 10.4 ± 0.5 micrometer compared to the manufacturer's specification (tolerance based on Talysurf measurements) of 10.0 ± 0.3 micrometer. There was a slight decrease (s.d. of 0.02 mm ) in the corneal curvature measured when the algorithm was applied for normal eyes.
The developed image processing algorithm succeeded in detecting feature points on an irregular artificial corneal surface. This enabled surface reconstruction of the surface with submicron elevation height accuracy. The algorithm systematically reduced the measured curvature of the cornea. This can be easily corrected with proper calibration.
This PDF is available to Subscribers Only