Corneal epithelial thickness mapping by Fourier-domain optical coherence tomography in normal and keratoconic eyes

Abstract

Objective: To map the corneal epithelial thickness with Fourier-domain optical coherence tomography (OCT) and to develop epithelial thickness-based variables for keratoconus detection.

Design: Cross-sectional observational study.

Participants: One hundred forty-five eyes from 76 normal subjects and 35 keratoconic eyes from 22 patients.

Methods: A 26,000-Hz Fourier-domain OCT system with 5-μm axial resolution was used. The cornea was imaged with a Pachymetry + Cpwr scan pattern (6-mm scan diameter, 8 radials, 1024 axial-scans each, repeated 5 times) centered on the pupil. Three scans were obtained at a single visit in a prospective study. A computer algorithm was developed to map the corneal epithelial thickness automatically. Zonal epithelial thicknesses and 5 diagnostic variables, including minimum, superior-inferior (S-I), minimum-maximum (MIN-MAX), map standard deviation (MSD), and pattern standard deviation (PSD), were calculated. Repeatability of the measurements was assessed by the pooled standard deviation. The area under the receiver operating characteristic curve (AUC) was used to evaluate diagnostic accuracy.

Main outcome measures: Descriptive statistics, repeatability, and AUC of the zonal epithelial thickness and diagnostic variables.

Results: The central, superior, and inferior epithelial thickness averages were 52.3 ± 3.6 μm, 49.6 ± 3.5 μm, and 51.2 ± 3.4 μm in normal eyes and 51.9 ± 5.3 μm, 51.2 ± 4.2 μm, and 49.1 ± 4.3 μm in keratoconic eyes. Compared with normal eyes, keratoconic eyes had significantly lower inferior (P = 0.03) and minimum (P<0.0001) corneal epithelial thickness, greater S-I (P = 0.013), more negative MIN-MAX (P<0.0001), greater MSD (P<0.0001), and larger PSD (P<0.0001). The repeatability of the zonal average, minimum, S-I, and MIN-MAX epithelial thickness variables were between 0.7 and 1.9 μm. The repeatability of MSD was better than 0.4 μm. The repeatability of PSD was 0.02 or better. Among all epithelial thickness-based variables investigated, PSD provided the best diagnostic power (AUC = 1.00). Using an PSD cutoff value of 0.057 alone gave 100% specificity and 100% sensitivity.

Conclusions: High-resolution Fourier-domain OCT mapped corneal epithelial thickness with good repeatability in both normal and keratoconic eyes. Keratoconus was characterized by apical epithelial thinning. The resulting deviation from the normal epithelial pattern could be detected with very high accuracy using the PSD variable.

Financial disclosure(s): Proprietary or commercial disclosure may be found after the references.

Conflict of interest statement

Financial interest: Yan Li, Ou Tan, and David Huang have a significant financial interest in Optovue, Inc. (Fremont, CA, USA), a company that may have a commercial interest in the results of this research and technology. These potential conflicts of interest have been reviewed and managed by Oregon Health and Science University. Robert Brass receives speaker honoraria from Optovue, Inc. Jack L. Weiss has no proprietary interest in the topic of this manuscript.

Copyright © 2012 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.

Figures

Figure 1

A: “Pachymetry+Cpwr” scan pattern consisted of 8 radial scans. B: A cross-sectional corneal optical coherence tomography (OCT) image (average of 5 repeated frames). C and D: Magnified sections of the OCT image shown in B.

Figure 2

Corneal epithelial boundaries (delineated by red dots) were identified in all eight meridians.

Figure 3

Average epithelial thickness maps of normal (A–C) and keratoconic eyes (D–F). Left column (A, D): All eyes were included with left eyes mirrored. Middle column (B, E): Right eyes (OD) only. Right column (C, F): Left eyes (OS) only. S = superior, I = inferior, T = temporal, N = nasal. The red circles overlaid on the map had diameters of 2 and 5 mm. The color scale represents the thickness in micrometers.

Figure 4

Examples of typical epithelium segmentation errors. The algorithm mistakenly identified Bowman’s-stroma interface instead of epithelium-Bowman’s interface at locations marked by arrows. (Insets) Magnified sections of the optical coherence tomography images.

Figure 5

The corneal epithelial thickness pattern deviation map (left) was calculated by subtracting the average normal pattern map (right) from the individual pattern map (middle). A: a normal example. B: a keratoconus example. S = superior, I = inferior, T = temporal, N = nasal. The color scale represents pattern deviation with no units.

Figure 6

Average corneal epithelial thickness pattern deviation map of keratoconic eyes. All eyes were included with left eyes mirrored. S = superior, I = inferior, T = temporal, N = nasal. The color scale represents pattern deviation with no units.

Figure 7

Histograms of 4 corneal epithelial thickness-based variables. A: Minimum (MIN). B: Minimum-maximum (MIN-MAX). C: Root-mean-square variation (RMSV). D. Root-mean-square pattern deviation (RMSPD). Dashed lines indicated optimized cut-off values of the variables.

Figure 8

Corneal epithelial thickness, pachymetric, topographic mean power maps, and cross-sectional optical coherence tomography (OCT) images of three cases. S = superior, I = inferior, T = temporal, N = nasal. Case 1 (A) was a randomly chosen normal right eye of a 37-year-old man. The corrected distance visual acuity (CDVA) was 20/20. The topographic simulated keratometry (SimK) readings were 42.2 diopter (D) and 43.0 D. The epithelial thickness was relatively uniform. Case 2 (B) was a 29-year-old woman with mild keratoconus in her right eye. The CDVA was 20/25. The SimKs were 42.9 and 44.7 D. The epithelial thickness map showed inferior epithelial thinning and relatively large epithelial thickness variation with superior-inferior (S-I) = 6.7 μm, root-mean-square variation (RMSV) = 4.3 μm, and root-mean-square pattern deviation (RMSPD) = 0.064. The pachymetry map showed inferior corneal thinning. The topographic mean power map revealed that the cone apex was located approximately 1.5 mm inferior to the pupil center. The epithelial thinning and the corneal thinning were both located over the cone. Case 3 (C) was a 33-year-old man with advanced keratoconus in his right eye. The CDVA was 20/200. The SimKs were 56.1 and 63.6 D. The epithelial thickness map showed thinning approximately 2 mm from the pupil center, and thickening about 1 mm from the pupil center, both inferotemporally. The epithelial thickness variation was large with the minimum = 38.1 μm, S-I = 0.07 μm, minimum-maximum (MIN-MAX) = −32.1 μm, RMSV = 6.1 μm, and RMSPD = 0.112. The pachymetry map showed inferotemporal corneal thinning. The cone apex was 2 mm inferotemporal to the pupil center. The thinnest epithelium was located over the cone apex as defined by point of greatest mean power. The epithelium was actually thicker than the average at the thinnest location of the cornea.