Corneal power modelling with OCT data - Thin and thick lens paraxial models versus raytracing

Area of Science:

• Ophthalmology
• Optometry
• Biomedical Optics

Background:

• Keratometric power evaluation involves various methods including Zeiss index (PKZ), paraxial thick cornea power (Gullstrand [PG]), power referenced to front (PFV) and back vertex planes (PBV), and raytracing power (PR).
• Understanding deviations from standard keratometry is crucial for accurate refractive error assessment and intraocular lens power calculations.

Purpose of the Study:

• To evaluate and compare different methods of corneal power calculation, including PKZ, PG, PFV, PBV, and PR.
• To model and predict the deviations between these different corneal power measurements using multivariable linear regression.

Main Methods:

• A large dataset of 4604 Casia2 measurements from a cataractous population was expanded to 30,000 using Copula expansion.
• PKZ was compared with PG, PFV, PBV, and PR across varying pupil sizes (1-6 mm) and apertures.
• Multivariate linear models were developed to predict the differences between PKZ and the other measurement methods.

Main Results:

• Average power values were PKZ: 43.05 D, PG: 42.74 D, PFV: 42.83 D, PBV: 43.59 D, and PR: 43.03 D with measured pupil size.
• Increasing aperture size from 1 to 6 mm increased mean raytracing power (PR).
• Linear models effectively predicted deviations for PG, PFV, and PBV from PKZ (R²=0.93, RMSE=0.01 D), but were less effective for PR (R²=0.79, RMSE=0.18 D).

Conclusions:

• Corneal power calculated using the paraxial thick cornea model (PG) differs from standard keratometric power (PKZ).
• A linear model can effectively predict these differences between PG and PKZ.
• Raytracing power (PR) shows greater deviation from keratometry (PKZ), and linear models are less accurate in predicting these differences.