Abstract
Purpose :
Corneal surface elevation can be obtained three-dimensionally with anterior segment Optical Coherence Tomography (OCT). However, its accuracy is subject to fixational eye movements during the acquisition. We developed a model to simulate corneal elevation measurements to study the quality of the reconstruction achieved with different scanning patterns.
Methods :
A custom spectral domain OCT (25K A-scans/s) was used to obtain videos (N=10, 22 Hz) of the corneal central meridian in 4 eyes from 2 subjects. The anterior corneal surface in each frame was segmented and fitted by conics. The apex movement was compared with simulations of fixational lateral/axial eye movements from MATLAB Image System Engineering Toolbox for Biology (ISETBIO) fixational eye movement generator with/without saccades. The results were used to simulate 3D-OCT acquisition of a cornea using either a raster or a meridional scan pattern with 300x50 A-scans. The 3-mm central data were fitted by a sphere (radius R) and the residuals fitted with up to 6th order Zernike polynomials (zC). Standard deviations (std) of the fitting parameters of repeated simulated adquisitions for different scan patterns were compared.
Results :
The mean amplitude of the cornea movement found experimentally was 119 in the lateral and 142 μm in the axial direction. The mean amplitude of the eye movement generator was 88 and 73 μm with and without microsaccades. Amplitude factors 1.32 and 1.95 were then applied to match lateral and axial movements respectively, which reduced the rms error between simulated and average experimental zC std from 3.7 to 0.9 μm. For raster scan, R std was 0.13 mm and mean zC std was 1.3 μm being highest for astigmatism and coma oriented with the slow scan direction. For meridional scan, R std was 0.01 mm and mean zC std was 2.0 μm being highest for astigmatism and trefoil terms. When all meridians were registered in the center, R std was 0.01 and mean zC std was reduced to 0.3 μm.
Conclusions :
Simulations of OCT corneal topography measurements with fixational eye movements predict a high influence of the scanning pattern on the measurement variability, particularly on specific Zernike modes. While eye rotations or scan position inaccuracies were not considered, these simulations can serve to determine the variability with different scan patterns and to evaluate the effectivity of motion correction algorithms.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.