Precise retinal shape measurement by alignment error and eye model calibration

Objective: To evaluate the repeatability of the optical coherence tomography (OCT) retinal shape measurement with and without alignment correction in children and adults. Methods: 62 eyes of the 31 subjects were examined on the OCT with auto-alignment and alignment correction functions. We performed three measurements on each eye, created 2D retinal height maps, and extracted horizontal and vertical profiles for repeatability analysis and Legendre polynomial representation. Repeatability was determined from the average standard deviation. We estimated the refractive errors produced by the observed alignment errors. We also examined the repeatability improvement of the slopes, curvatures, and higher-order coefficients for the subjects using the Student t test. Results: Repeatability was higher in the data with alignment. We observed the average repeatability in the child group with standard deviation (SD) equal 38.20 µm in raw data, and SD = 6.11 µm in the corrected data, in adults SD = 17.04 µm in raw data, and SD = 5.22 µm in the corrected data. The slope repeatability improved in both child [horizontal t (18) = 4.62, p < 0.001 and vertical t (17) = 4.43, p < 0.001] and adult groups [t (18) = 2.73, p = 0.007 and vertical t (26) = 2.14, p = 0.02], while higher-order coefficients were not affected. Conclusions: Alignment correction improved repeatability of the OCT retinal shape measurements, especially for child subjects. Curvature and higher-order distortions were not affected by the alignment error. The refractive errors produced by alignment errors are low, and the model can be used to estimate the peripheral refraction.