A Novel Method for Determining Choroidal Thickness in Humans With Partial Coherence Interferometry Axial Length Data

Abstract

Abstract: : Purpose: To evaluate the feasibility and precision of a method to determine choroidal thickness in humans using partial coherence interferometry (PCI). Background: There are currently no reliable methods to measure the thickness of the choroid in human eyes with a precision to assess short term fluctuations in choroidal thickness in vivo. Animal studies suggest that such fluctuations are important in visually guided eye growth. In PCI axial length (AL) data in humans, there is a peak some 250–400µm after the Bruch’s membrane/RPE location (P3) that likely represents a choroidal/scleral signal (P4). However, the presence and morphology of P4 varies widely both between and within subjects. Methods: We developed an algorithm to determine the position and variability of P4 in subjects with a robust P4 signal on PCI. Our typical PCI dataset for AL measurement consists of 80 individual tracings. Unlike other retinal PCI peaks that are often present in individual tracings, P4 is usually evident only in signal averaged data. A bootstrapping method determined P4 variability. Nonlinear Gaussian curve fitting estimated the location of P4 and, with signal averaging and filtering, determined the distance from P4 to P3 (a surrogate for choroidal thickness) and the variability of the P4 location. This method was initially developed in a dataset of 64 subjects and validated on a second set of 40 subjects. Results: In the initial 64 subjects, 18 (28%) had a P4 signal; and in the second 40 subjects, the algorithm identified 11 (28%) with P4s. The average P4–P3 distance was 307µm (std error of measurement: 16µm) for the initial study and 293µm (std error of measurement: 19µm) for the second study. Conclusions: We developed a method to estimate choroidal thickness in vivo from PCI AL measurements, based on signal processing, curve fitting and bootstrapping. Using this method, about 30% of subjects demonstrate a reproducible choroidal signal that can be used to determine whether human choroidal thickness undergoes diurnal oscillations as described in animals.

Keywords: choroid • imaging/image analysis: clinical