Related Experiment Video
Updated: Jan 9, 2026
In vivo Structural Assessments of Ocular Disease in Rodent Models using Optical Coherence Tomography
Published on: July 24, 2020
Corneal power modelling with OCT data - Thin and thick lens paraxial models versus raytracing
Achim Langenbucher1, Nóra Szentmáry1,2, Alan Cayless3
1Department of Experimental Ophthalmology, Saarland University, Homburg (Saar), Germany.
Background:
Evaluating keratometric power with Zeiss index (PKZ), paraxial thick cornea power (Gullstrand [PG]) and power referenced to the front (PFV) and back vertex plane (PBV) and raytracing power (PR), and modelling the deviation from PKZ with a multivariable linear prediction model.
Methods:
A dataset of 4604 Casia2 measurements from a cataractous population was Copula expanded to N = 30 000 maintaining the individual univariate distributions and interactions. PKZ was compared with paraxial thick lens corneal power PG, PFV, PBV, and PR using measured pupil size and variations of apertures from 1 to 6 mm.
Results:
On average, PKZ/PG/PFV/PBV was 43.05/42.74/42.83/43.59 D, and PR was 43.03 D with the measured pupil size. Varying pupil size from 1 (1) 6 mm increased mean PR with aperture size (42.84/42.91/43.02/43.17/43.37/43.62 D). The multivariate linear models predicting the deviation of PG-PKZ, PFV-PKZ, and PBV-PKZ with corneal radii and central thickness performed well, with R2 = 0.93 and a root mean squared prediction error of 0.01 D, whereas the equivalent model for PR-PKZ with corneal radii and asphericities, corneal thickness, and aperture sizes performed less well, with R2 = 0.79 and a root mean squared prediction error of 0.18 D.
Conclusion:
Corneal power derived using the paraxial thick cornea model differs from keratometric power and a linear model performs well in predicting the differences. However, raytracing power differs even more from keratometry with the linear model far less effective.
