Corneal topography matching by iterative registration.

Area of Science:

• Ophthalmology and Biomedical Engineering
• Computational Geometry and Image Analysis

Background:

• Videokeratography measures corneal topography using overlapping maps.
• Accurate reconstruction requires precise registration of these maps, especially given peripheral measurement uncertainties.

Purpose of the Study:

• To evaluate direct search optimization and iterative closest/corresponding point routines for corneal map registration.
• To assess the performance of these methods with simulated and real clinical data.

Main Methods:

• Application of Nelder-Mead algorithm and iterative closest/corresponding point variants for registration.
• Use of Zernike polynomial interpolation for differing measurement points.
• Comparison of point-to-plane iterative closest/corresponding point with other approaches.

Main Results:

• The point-to-plane iterative closest/corresponding point routine demonstrated robustness to noise and insensitivity to initial parameters.
• This method effectively addressed registration and alignment errors common in other techniques.
• High-quality results were achieved with real clinical videokeratography data.

Conclusions:

• The point-to-plane iterative closest/corresponding point algorithm is a superior method for corneal topography registration.
• This technique enhances the accuracy and reliability of reconstructing overall corneal shape from videokeratography data.