Comparison of curvature-based and biometry-based methods for in vivo crystalline lens power calculation

Abstract

Purpose : In vivo crystalline lens power is typically estimated from measurements of ocular distances and refraction (Bennett method) since standard commercial biometry devices do not measure lens curvatures. Lens power can also be calculated directly from measurements of lens thickness and curvatures and an estimate of the lens equivalent refractive index. Lens curvatures have been measured using phakometry, Scheimpflug imaging and more recently Optical Coherence Tomography (OCT). The purpose of the present study was to compare lens power calculated using 1) a modified version of the Bennett method and 2) the lens shape obtained from extended depth OCT images.

Methods : Images of the left eye of 8 subjects (25.0 ± 4.2 y/o, spherical equivalent: -5.62 to -0.13 D) focused at distance were obtained using a custom-built extended depth OCT imaging system (Ruggeri et al, Biomed Opt Express, 2012). The images were automatically segmented and corrected for distortions due to refraction to determine the boundaries of the cornea, lens, and retina and calculate central corneal thickness, anterior chamber depth, lens thickness, vitreous chamber depth, and anterior and posterior corneal and lens curvature. For each subject, distortion correction was performed for a range of equivalent refractive indices to find the value of the refractive index, minimizing the error between predicted refraction and measured refraction. Lens power was calculated using two different methods. Method 1 is a modified version of the Bennett method which relies on corneal shape, ocular distances and refraction to predict the lens power (Hernandez et al, Biomed Opt Express, 2015). Method 2 applies the formula for the power of a thick lens using lens curvatures, thickness and equivalent refractive index. The two methods were compared using a Bland-Altman analysis.

Results : The lens power was 24.38 ± 2.02 D for the modified Bennett method (method 1) and 24.41 ± 2.16 D for the curvature-based method (method 2). The mean difference between the two methods was -0.02D, with a 95% confidence interval of ± 3.20 D. The absolute value of the error for the 8 lenses ranged from 0.10 to 2.87 D.

Conclusions : On average, there is no statistically-significant difference between curvature- and biometry-based methods for estimating lens power. However, there are significant intra-individual differences in estimated lens power between the two methods.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.