1040-5488/14/9107-0793/0 VOL. 91, NO. 7, PP. 793Y802 OPTOMETRY AND VISION SCIENCE Copyright * 2014 American Academy of Optometry
ORIGINAL ARTICLE
Measurement of Anterior Scleral Curvature Using Anterior Segment OCT Hyuk Jin Choi*, Sang-Mok Lee*, Ja Young Lee†, Seung Yong Lee†, Mee Kum Kim*, and Won Ryang Wee*
ABSTRACT Purpose. To investigate and validate methods for measuring the radius of anterior scleral curvature using anterior segment optical coherence tomography images. Methods. Twenty-four volunteers were enrolled in this study. Anterior segment optical coherence tomography images, centered on horizontal/vertical limbus, including adjacent anterior sclera, were obtained in addition to conventional images centered on the optical axis. Central horizontal, nasal, and temporal optical coherence tomography images were consolidated to a new image for subsequent analyses. The reference points of limbal surface and three scleral points were marked nasally and temporally. The radius of a best-fit circle to the six scleral points was derived (the BFC [best-fit circle] method) and the radii of two circles, the centers of which are on the optical axis and pass through the points of the scleral surface at 2 mm from the limbus nasally and temporally, were calculated (the axial method). To assess the reliability and accuracy of each method, intraobserver and interobserver agreements were analyzed and the radii of contact lenses with known curvatures were measured. Results. The mean (TSD) radius of a BFC was 13.12 (T0.80) mm. The mean (TSD) radius of nasal anterior scleral curvature (13.33 T 1.12 mm) was significantly greater than that of temporal anterior scleral curvature (12.32 T 0.77 mm) (paired samples t test, p G 0.001). The BFC and axial methods showed excellent intraobserver and interobserver agreements for measurements (intraclass correlation coefficient 9 0.75, p G 0.001), whereas both methods showed a tendency to slightly underestimate the actual curvature of a rigid contact lens of known dimensions (j0.07 T 0.13 mm [the BFC method] and j0.19 T 0.07 mm [the axial method], Wilcoxon signed rank test, p = 0.173 and p = 0.028, respectively). Conclusions. Anterior segment optical coherence tomography is a valuable tool for measuring the radii of anterior scleral curvatures by image processing and mathematical calculation and can provide useful information in specific clinical situations such as designing scleral lenses. (Optom Vis Sci 2014;91:793Y802) Key Words: anterior segment optical coherence tomography, radius of scleral curvature, sclera, scleral lens, LASIK surgery
T
he human sclera is a roughly spherical, relatively avascular, white, rigid, and dense connective tissue that covers the globe posterior to the cornea.1 Although the sclera covers about 90% of the surface area of the eye, little is known about the mathematical parameters related to the shape beyond the limbus. Generally, it has been known that the sclera is an incomplete
*MD, PhD † MD Department of Ophthalmology, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, Republic of Korea (HJC); Department of Ophthalmology, Seoul National University College of Medicine, Seoul, Republic of Korea (HJC, S-ML, MKK, WRW); Department of Ophthalmology, Incheon Medical Center, Incheon, Republic of Korea (JYL); and Department of Ophthalmology, Eulji University Hospital, Daejeon, Republic of Korea (SYL). Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.optvissci.com).
sphere with an average outer diameter of 24 mm and an average inner diameter of 23 mm (mean radius of curvature, 11.5 mm).1,2 Measurements of scleral curvature can be applied to several clinical situations. The ability to image the contour of the anterior sclera would be important for designing scleral lenses that are supported entirely by the sclera and completely vault the cornea and the limbus. It could also provide a guideline for selecting proper suction rings during LASIK surgery. To analyze the shape of the anterior segment beyond the corneoscleral junction, high-quality shape information about this area should be provided by sophisticated devices such as anterior segment optical coherence tomography (AS-OCT), ultrasound biomicroscopy (UBM), and Scheimpflug camera system.3 The versatility, accuracy, and high resolution of AS-OCT have widened its application to surgical planning and follow-up tools such as corneal refractive surgery and corneal transplantation, anterior
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
794 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
chamber biometry measurement, angle assessment, and diagnosis of various anterior segment diseases including keratoconus and tumors.4 Although AS-OCT is a useful tool that can visualize the shape beyond the limbus, to the best of our knowledge, there are only a few studies that analyze anterior scleral curvature in the anterior segment images obtained using AS-OCT.5Y9 These studies gave clinicians new information, that is, that the peripheral cornea often continues to the sclera in a straight line (tangential).8,9 Furthermore, interestingly, the anterior scleral shape is also tangential in many cases and one or more sections of the sclera are often steeper or flatter than the rest (nonrotationally symmetric in nature).5Y9 The authors were interested in assessing the shape of the sclera and devised methods for analyzing anterior scleral curvature using AS-OCT images, with easily available software, and simple mathematical concepts. In these methods, points at approximately 2 mm from the limbus were thought to be clinically important because largediameter contact lenses settle down, and suction rings in the LASIK procedure come in contact with the eyeball at these points. Particularly, the authors have focused on the axial curvature, based on the fact that hard contact lenses are made by lathe cutting, which puts the lens material on a rotating mount, whereas machine cutting instruments sculpt away excess lens material to carve a precision-cut lens. Therefore, the aims of this study were to introduce new methods for analyzing anterior scleral curvature using AS-OCT images and to evaluate repeatability and reproducibility of the methods by applying them to images from volunteers with normal ocular health.
METHODS Subjects The research was in accordance with the tenets of the Declaration of Helsinki, and the protocol of this study was approved by the institutional review board of Seoul National University Hospital. Informed consent was obtained from the subjects after providing explanation regarding the nature and possible consequences of the study. A total of 24 Korean volunteers were recruited into this study. Volunteers who had previous conjunctival or scleral diseases, ocular surgeries involving the conjunctiva or sclera, and conjunctival degenerative changes such as pterygium and pinguecula were excluded. The mean (TSD) age of the participants was 31.3 (T6.5) years (range, 24 to 54 years). Of the 24 subjects, 12 were men and 12 were women.
Image Acquisition and Processing Data and observations were collected during one study visit. All AS-OCT measurements were performed by one operator. In the anterior segment mode, a total of six images (16 mm wide) for each eye were obtained using the Visante OCT instrument (Carl Zeiss Meditec, Dublin, CA). Two images were obtained horizontally and vertically while the subject fixed an eye on the focusing spot inside the instrument. When the OCT scan was centered on the corneal vertex reflex, the operator could see a bright vertical flare line crossing the corneal vertex. During image acquisition, the operator removed the tilt from the eye image by adjusting the internal fixation target to compensate for the angle between the visual axis and the optical axis (angle alpha), which implied that the bright vertical flare line corresponded approximately
to the optical axis. Then, perpendicular images in the four cardinal directions of gaze were also captured while the subject fixed an eye on the upper, lower, nasal, and temporal external targets, which were located at 15 degrees from the primary position. All images were corrected for distortion using the built-in proprietary image-correction algorithm (dewarping; Software Version 3.0.1.8). With built-in calipers, a reference value of 1.00 mm was marked on each image. Images of the right eye of each participant were used for the subsequent analyses. To obtain an image for analyses, three AS-OCT images of each patient (central horizontal, nasal, and temporal) were composed to a new one by image rotation and movement without resizing using Microsoft PowerPoint (version 2007, Redmond, WA). In brief, nasal and temporal images were rotated and overlapped on the central horizontal image using the scleral spur as a reference point of superimposition. The best composite image, in which the corneal and scleral surface texture of the three images was best matched, was obtained and saved as a new image file. The composite procedure showed excellent repeatability as assessed by one author (S-ML) and excellent reproducibility between two different authors (S-ML and HJC) (Fig. A1 and Table A1Vsee Appendix, available at http://links.lww.com/OPX/A177). The composite images obtained by one of the authors (S-ML) were used for subsequent analyses. Meanwhile, vertical images could not be analyzed because the texture of superior and inferior images was often too distorted to be able to compose them, which was mainly caused by excessive up/down gaze or the operator’s digital pressure for adequate visualization of the sclera.
Analysis of Radius of a Best-Fit Circle to the Anterior Sclera To evaluate radius of a best-fit circle (BFCR) to the anterior sclera, each composed image was analyzed by authors using Image-Pro Plus 4.5 software (Media Cybernetics, Inc, Rockville, MD; Fig. 1A). Briefly, a reference value of 1.00 mm was converted to the corresponding number of pixels. As described in a previous report,10 the reference points of the limbal surface (NL in the nasal surface and TL in the temporal surface), at which the line passing the scleral spur met perpendicularly, and the scleral surface points N2/T2 (2 mm posterior to NL/TL) were marked. Then, the authors marked the scleral surface points 1 mm apart from N2 (N1 and N3) and T2 (T1 and T3) based on the assumption that N1 to N3 (T1 to T3) were on the same circle and the normal to the line joining N1 and N3 (T1 and T3) was the same normal to the scleral surface at N2 (T2). The first high reflective tissue signal from the episclera was considered as the scleral surface. Finally, the radius of the circle, which best fits the six scleral surface points (N1, N2, N3, T1, T2, and T3), was derived. Each value was converted to the metric unit (millimeter) by dividing it by reference pixels corresponding to 1.00 mm.
Analysis of Axial Radius of Scleral Curvature To measure axial radius of scleral curvature (AR), composite images were analyzed by authors using Image J software version 1.44d (National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov/ij/index.html, Fig. 1B). Firstly, the authors decided the x, y coordinates of two points on the anterior corneal
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
795
FIGURE 1. Measurements of the radius of a best-fit circle to the anterior sclera (BFCR), the axial radii of anterior scleral curvature (AR), and the radii of contact lenses. To measure BFCR and AR, a composite AS-OCT image was used. (A) The radius of a best-fit circle to six scleral surface points (N1, N2, N3, T1, T2, and T3) was regarded as BFCR. (B) The distance from the point at which the T line and the A line meet to T2 was regarded as the axial radius of temporal scleral curvature. (C) The front curve radius of contact lenses was derived using the same BFC and axial methods. NL, the reference point of the nasal limbal surface that corresponds to the scleral spur; TL, the reference point of the temporal limbal surface that corresponds to the scleral spur; N1 to N3, three consecutive surface points from the NL; T1 to T3, three consecutive surface points from the TL; T line, the line that passes through T2 and bisects the lines passing through T1 and T3 at a right angle; A line, the axis line. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
796 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
surface and the anterior lens surface, respectively, which were lying on the axis line (A line). The bright vertical flare line (the optical axis) was used as the A line in most cases, whereas the geometric axis line (defined as the line bisecting the bases of the anterior chamber, which could be drawn between the scleral spurs on both sides) was used as the A line when the bright flare line was hard to visualize. Then, the authors determined the x, y coordinates of the points (N1, N2, N3, T1, T2, and T3 as described in the measurement of BFCR) on the scleral surface. With the data of these points, the AR was calculated mathematically using Microsoft
Excel spreadsheet (version 2007, Redmond, WA) according to the concept described below. For calculating the radius of temporal scleral curvature (ARt), the authors drew a line (T line) that passed through T2 and bisected the lines passing through T1 and T3 at a right angle. The authors then calculated the point where the T line met the A line. The distance from this point to T2 was regarded as ARt, and the same concept was applied for calculating the radius of nasal scleral curvature (ARn). Each value was converted to the metric unit (millimeter) by dividing it by reference pixels corresponding to 1.00 mm.
FIGURE 2. Radii of the anterior scleral curvature. (A) The mean values of BFCR, ARn, and ARt (*paired samples t test, p G 0.001). (B) Correlation between ARn and ARt (Pearson correlation coefficient, r = 0.515, p = 0.01). (C) Correlation between BFCR and ARn (Pearson correlation coefficient, r = 0.757, p G 0.001). (D) Correlation between BFCR and ARt (Pearson correlation coefficient, r = 0.856, p G 0.001). BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal scleral curvature; ARt, axial radius of temporal scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
Validation of Accuracy To validate the accuracy of the abovementioned techniques in calculating radius of curvature, six custom-made spherical rigid gas-permeable (RGP) contact lenses of the same material (Boston XO; hexafocon A), refractive indices (1.415), overall diameters (11.0 mm), and uniform thicknesses (0.5 mm) but with front surface radii between 7.60 and 8.60 mm in 0.2-mm steps were tested. The authors measured only front surface radii because refractive indices of the contact lens material and the cornea were different, which caused a more prominent distortion in the back surface image of the contact lenses after dewarping. Commercially available spherical RGP contact lenses (Lucid Korea Co Ltd) were used for the validation. The manufacturing tolerance for the base curve radius tested by microspherometer or radius gauge was T0.05 mm. In the high-resolution mode of the AS-OCT, a 10-mm-wide horizontal image centered on the vertex reflex was obtained from each contact lens after being dewarped by the same algorithm used in the human eye. Only central high-resolution images were used for analyses because coordinate errors gradually increase toward the periphery even in the dewarped image.11 Like the analysis of human anterior segment, the central bright vertical flare line was considered as the A line and two imaginary landmark points located at 1 mm from the A line were considered as NL and TL, respectively, because there were no landmarks like scleral spurs in the human eye. Then, six surface points corresponding to N1 to N3 and T1 to T3 were marked and the analyses were performed using the above techniques (Fig. 1C). Results were presented as the mean values of three different measurements calculated by one author (HJC).
Statistical Analysis To evaluate intraobserver reliability, BFCR and ARs were evaluated twice by the same author (HJC) at a 1-week interval. To assess interobserver reliability, BFCR and ARs were also measured in the same AS-OCT images by different authors (BFCR by JYL and ARs by SYL). Both authors were blinded to the measurements calculated by the other author. Statistical analysis was performed using SPSS 18.0 (Chicago, IL) software. The intraobserver difference in the radii was analyzed using the paired samples t test. Pearson correlation coefficient and
797
intraclass correlation coefficient (ICC) were used to assess intraobserver and interobserver reliability. The guidelines used for the interpretation of the ICC were as follows: an ICC of less than 0.40 indicated poor reliability, an ICC of 0.40 to 0.75 indicated fair to good reliability, and an ICC of greater than 0.75 indicated excellent reliability.12 p G 0.05 was considered statistically significant. The Bland-Altman plots,13 graphic representations illustrating repeatability of observations, were also used to show the relationship between the two results obtained by the same author or two different authors.
RESULTS The Shape of the Anterior Sclera First measurements of BFCR, ARn, and ARt calculated by one author (HJC) were analyzed to define the shape of the anterior scleral surface. The mean (TSD) BFCR was 13.12 (T0.80) mm (range, 11.69 to 14.63 mm). The mean (TSD) ARn (13.33 T 1.12 mm; range, 11.11 to 16.41 mm) was significantly greater than the mean (TSD) ARt (12.32 T 0.77 mm; range, 11.03 to 13.63 mm) (Fig. 2A, paired samples t test, p G 0.001), whereas ARn and ARt showed a significant positive correlation (Fig. 2B, Pearson correlation coefficient, r = 0.515, p = 0.01). Radius of a best-fit circle was positively correlated with ARn and ARt (Fig. 2C, D, Pearson correlation coefficient, r = 0.757 and 0.856, respectively, each p G 0.001).
Reliability of the Measurements The repeatability of two consecutive measurements (at a 1-week interval) of scleral curvature calculated by one author showed an excellent level of agreement in measurements of BFCR (ICC = 0.981), ARn (ICC = 0.951), and ARt (ICC = 0.837). The mean (TSD) absolute difference between two consecutive measurements was smaller in BFCR (0.13 T 0.09 mm; range, 0.01 to 0.29 mm) than in ARn (0.36 T 0.24 mm; range, 0.12 to 0.92 mm) or ARt (0.34 T 0.26 mm; range, 0 to 1.10 mm). The Bland-Altman plots showed 24 of 24 (100%) in BFCR and 23 of 24 (96%) in ARn/ ARt measurement differences were within the 95% confidence interval of the mean difference (Table 1 and Fig. 3). Excellent interobserver reproducibility for measurements of BFCR (ICC = 0.974), ARn (ICC = 0.941), and ARt (ICC = 0.841)
TABLE 1.
Reliability of the measurements
First measurement, mm Intraobserver reliability BFCR 13.12 T 0.80 ARn 13.33 T 1.12 Art 12.32 T 0.77 Interobserver reliability BFCR 13.12 T 0.80 ARn 13.33 T 1.12 Art 12.32 T 0.77
Second measurement, mm
Absolute difference, mm
13.11 T 0.87 13.60 T 1.07 12.27 T 0.74 13.15 T 0.87 13.01 T 1.16 12.33 T 0.78
Pearson correlation coefficient
Intraclass correlation coefficient
r
p
ICC
p
0.13 T 0.09 0.36 T 0.24 0.34 T 0.26
0.984 0.952 0.838
G0.001 G0.001 G0.001
0.981 0.951 0.837
G0.001 G0.001 G0.001
0.15 T 0.12 0.40 T 0.30 0.34 T 0.26
0.977 0.942 0.841
G0.001 G0.001 G0.001
0.974 0.941 0.841
G0.001 G0.001 G0.001
BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal anterior scleral curvature; ARt, axial radius of temporal anterior scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
798 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
FIGURE 3. Intraobserver reliability presented by Bland-Altman plots. The same author measured BFCR (A), ARn (B), and ARt (C) twice at a 1-week interval. Solid lines, the mean difference between two measurements; dotted lines, the 95% confidence interval of the mean difference; BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal scleral curvature; ARt, axial radius of temporal scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
was also evident. The mean (TSD) absolute difference between two measurements calculated by two different authors was also smaller in BFCR (0.15 T 0.12 mm; range, 0 to 0.48 mm) than in ARn (0.40 T 0.30 mm; range, 0.01 to 1.07 mm) or ARt (0.34 T 0.26 mm; range, 0.01 to 1.14 mm), whereas the Bland-Altman plots showed 23 of 24 (96%) in BFCR/ARn/ARt measurement differences were within the 95% confidence interval of the mean difference (Table 1 and Fig. 4).
Accuracy of the Measurements Table 2 shows the results of accuracy testing of both BFC and axial methods using contact lenses with different front curve radii. There were small steeper deviations from the actual values in both methods, although only the axial method showed a significant underestimation of the actual curvature (Wilcoxon signed rank test, p = 0.028). The mean deviation of the BFC method (j0.07 T 0.13 mm) was smaller than that of the axial method (j0.19 T 0.07 mm).
DISCUSSION In this study, the authors validated the methods for measuring the radius of a ‘‘best-fit circle’’ to the anterior sclera and ‘‘axial’’ radius of anterior scleral curvature at the points of nasal and temporal scleral surfaces 2 mm from the limbus, using a composite image derived from three horizontal AS-OCT images. The BFCR of the anterior sclera in 24 Korean volunteers was 13.12 mm. The radius of anterior scleral curvature derived from the ‘‘axial’’ method was significantly greater in the nasal sclera (13.33 mm) than in the temporal sclera (12.32 mm). In terms of repeatability and reproducibility, both ‘‘BFC’’ and ‘‘axial’’ methods showed excellent intraobserver and interobserver agreement. Moreover, both the ‘‘BFC’’ and ‘‘axial’’ methods were found to only slightly underestimate the actual anterior curvature values of RGP contact lenses of known dimensions. In this study, the authors used AS-OCT despite the availability of UBM because AS-OCT has several advantages over UBM for objective analysis of anterior scleral imaging. Image acquisition by AS-OCT is a noncontact technique and a less time-consuming procedure, which increases the participant’s comfort level as well as safety. Unlike in AS-OCT, supine positioning and inadvertent pressure on the eyecup may alter the natural shape of the anterior segment during UBM examination.14Y16 This alteration may be more prominent during the participant’s lateral gaze because more pressure needs to be applied to the eyecup. The scanning resolution with standard software is higher in AS-OCT (60 Km of lateral resolution and 18 Km of axial resolution) than in UBM (50 Km of lateral resolution and 25 Km of axial resolution).4,15,17 Furthermore, the scleral spur, which was an especially important landmark for image processing in this study, is more distinct on AS-OCT images.16 Using AS-OCT images, the authors introduced two different ways for evaluating anterior scleral curvature (designated as the ‘‘BFC’’ method and the ‘‘axial’’ method, respectively) and obtained valuable results as described above. The BFCR of the anterior sclera was 13.12 mm, which is comparable to the average anterior scleral radius (12.40 mm; range, 10.10 to 16.60 mm) derived by manually drawing a forced circle through the anterior sclera in a previous report.6 The temporal curvature of the anterior
799
sclera was significantly steeper than the nasal curvature, which confirms the fact that the human eyeball is not a perfect sphere.1,2 Actually, the readers can recognize this fact at a glance in the composite AS-OCT images (Fig. 1A, B), and this result corresponds well with those of previous reports, which demonstrated that the mean scleral curvature is steepest in the temporal sclera but similar in the nasal, superior, and inferior scleral planes8,9 and the ocular surface beyond the cornea is not rotationally symmetric.6 This could be partly explained by the fact that the medial rectus muscle inserts most anteriorly into the sclera and the visual axis forms an angle of 23 degrees with the orbital axis in the primary position, which means that the medial rectus muscle adducts the eyeball with subsequent flattening of the nasal anterior sclera. Regarding intraobserver repeatability and interobserver reproducibility, both BFC and axial methods showed excellent level of agreements, which were higher compared with the method (intrasession ICC = 0.83) using built-in calipers and protractor tools introduced by Hall et al.8 In a subsequent study involving 204 subjects,9 Hall et al. reported that scleral radii of curvature ranged from j57.4 to 312.5 mm (mean, 35.5 mm) in the nasal sclera and from 3.1 to 100.0 mm (mean, 22.4 mm) in the temporal sclera, which are quite different from the results of this study and imply that the radii represent instantaneous curvature that has a higher degree of variability from point to point like that in the tangential map of corneal topographer.18 Unfortunately, the authors could not find any detailed descriptions about how scleral radii were derived and at which point scleral radii were analyzed in the previous reports.8,9 Moreover, the authors could not compare the methods used in this study with those used in other studies because curvature measurement tools could not be provided by a local distributor of AS-OCT. Assignment of multiple points to fit in the BFC method and introduction of the reference plane (A line) into the axial method may be the key factors for decreasing the variations in measurements between sessions and subjects. In terms of accuracy of the measurements, both BFC and axial methods slightly underestimated the actual front curve radius of custom-made RGP contact lenses. The mean deviations of BFC (j0.07 mm) and axial (j0.19 mm) methods were comparable with that (j0.11 mm) in a previous report,3 in which base curve radii of RGP contact lenses were calculated using sag and chord measurements. These results seem to be attributed not only to an error during image processing because refractive indices of contact lens material (1.415) and corneal tissue (1.376) are different but also to a tendency toward underestimation of corneal surface curvature in measurements even after correction with built-in AS-OCT software.11 Measurements of scleral curvature would be of great value in designing scleral lenses. As the currently available corneal topographers do not acquire imaging data beyond the limbus, topographic indices cannot be used to accurately predict the base curve of scleral lenses.19 Neither the central radius of the cornea nor the corneal sagittal depth is reported to be highly predictive of the scleral shape, which is indicated by nasal and temporal angles at the 15-mm chord.20 Therefore, current scleral lens fitting is mainly based on an iterative system of modifications to trial lenses, a rather cumbersome and time-consuming process. Introduction of fitting paradigms based on the sagittal depth and information on scleral contour, which could be simultaneously obtained with
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
800 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
FIGURE 4. Interobserver reliability presented by Bland-Altman plots. Two different authors measured BFCR (A), ARn (B), and ARt (C), while both authors were blinded to the measurements calculated by the other author. Solid lines, the mean difference between two measurements; dotted lines, the 95% confidence interval of the mean difference; BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal scleral curvature; ARt, axial radius of temporal scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
801
TABLE 2.
Accuracy of the measurements Actual FCR, mm 7.60 7.80 8.00 8.20 8.40 8.60 Mean T SD p*
BFCR method
AR method
Calculated FCR, mm
Difference, mm
Calculated FCR, mm
Difference, mm
7.72 7.75 7.98 8.03 8.14 8.54
0.12 j0.05 j0.02 j0.17 j0.26 j0.06 j0.07 T 0.13 0.173
7.28 7.65 7.80 8.02 8.26 8.44
j0.32 j0.15 j0.20 j0.18 j0.14 j0.16 j0.19 T 0.07 0.028
Results were presented as the mean values of three different measurements calculated by one author (HJC). *Wilcoxon signed rank test. FCR, front curve radius; BFCR, the radius of a best-fit circle to the lens; AR, axial radius of curvature.
AS-OCT, seem to offer a more valuable fitting method.21 Actually, in a small case series, AS-OCT-based sag and chord measurements were reported to be highly useful fitting tools for scleral lenses that seem to offer more accurate and efficient alternatives to the trial lens method.3 Measurements of anterior scleral curvature as in this study may provide additional useful information especially for designing the landing zone of scleral lenses. Moreover, the nonspherical nature of the sclera confirmed by AS-OCT may help the practitioners select nonrotationally symmetric scleral lenses such as toric and quadrant-specific lenses to minimize conjunctival blanching and to improve comfort.5Y7,22 Anterior segment optical coherence tomography could also serve as a valuable tool before LASIK surgery. Scleral curvature measurement will greatly help the refractive surgeon during the process of selecting the correct suction ring, avoiding deformation of the globe wall that induces a secondary traction in the vitreous cavity and an anterior displacement of the posterior pole.23 Furthermore, better refraction outcomes can be achieved by selecting an appropriate suction ring because we can expect more predictable values for the thickness and diameter of the corneal flap by avoiding deformation of the eyeball during the application of the suction ring. There are some limitations to this study. The sample size of 24 was small. Hence, a larger study is necessary for drawing more solid conclusions. When the device conducted a dewarping process on the perpendicular images beyond the cornea, it drew the front line along the anterior scleral surface and the back line along the posterior aspect of the intraocular muscle and inner scleral lamina. Therefore, the scleral contour beyond the cornea might not be the same as the actual contour because the refractive index of the involved tissue is different from that of the cornea. Nevertheless, the authors tried to reduce image distortion by rotating the eyeball minimally, by locating the sclera near the center within vertical analysis limits, and by confirming the presence of a corneal vertex reflex (central vertical flare), which indicates that the cross section is along the corneal meridian, ensuring accurate dewarping of the image. Regarding the accuracy of the procedure, the validation using larger-diameter contact lenses such as scleral lenses appears to be more relevant, given that the scleral measures on the participants were beyond the cornea and required three images to be stitched together. However, it was impossible because AS-OCT continuously failed to proceed with dewarping of the images from
the scleral lenses. The authors were only able to analyze horizontal scleral curvature. While capturing vertical scleral images, shadowing of the eyelids often obscured anterior segment images because of low palpebral fissure height and greater eyelid tension in Asians. Excessive up/down gaze and digital retraction of the eyelids induced a distortion of the original shape of the anterior segment owing to the pulling effect of extraocular muscles (flattening the anterior scleral curvature) and inadvertent digital pressure on the globe, respectively. Similarly, eye turning during nasal and temporal measurements could also induce a change in the scleral shape. Volunteers who had conjunctival or subconjunctival lesions near the limbus were not good candidates because AS-OCT could not clearly discriminate these lesions from the underlying sclera. Therefore, subjects who had conjunctival pathology such as pterygium or pinguecula were excluded from this study. In conclusion, AS-OCT is a useful tool for measuring the radius of scleral curvature, which can be useful in some optometric and ophthalmological practices such as designing scleral lenses and choosing a proper suction ring for LASIK surgery.
ACKNOWLEDGMENTS This study was supported by a grant from the Korean Health Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (Project Nos. A084496 and A120018 [HI12C0015]). The test contact lenses were kindly provided by Lucid Korea Co. Ltd. The authors declare that they have no competing financial interests. Received August 26, 2013; accepted April 15, 2014.
APPENDIX The appendix, including Table A1 and Fig. A1, demonstrating validation of the image composition procedure, is available at http:// links.lww.com/OPX/A177.
REFERENCES 1. Dawson DG, Ubels JL, Edelhauser HF. Cornea and sclera. In: Levin LA, Nilsson SFE, Hoeve JV, Wu S, Kaufman PL, Alm A, eds. Adler’s Physiology of the Eye, 11th ed. Philadelphia, PA: Saunders; 2011: 71Y130. 2. Watson PG, Young RD. Scleral structure, organisation and disease. A review. Exp Eye Res 2004;78:609Y23.
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
802 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al. 3. Gemoules G. A novel method of fitting scleral lenses using high resolution optical coherence tomography. Eye Contact Lens 2008; 34:80Y3. 4. Ramos JL, Li Y, Huang D. Clinical and research applications of anterior segment optical coherence tomographyVa review. Clin Experiment Ophthalmol 2009;37:81Y9. 5. Van der Worp E. New technology in contact lens practice. Contact Lens Spectrum 2010;25(2):22Y9. Available at: http://www.clspectrum.com/ articleviewer.aspx?articleID=103880. Accessed January 19, 2014. 6. Van der Worp E, Graf T, Caroline PJ. Exploring beyond the corneal borders. Contact Lens Spectrum 2010;25(6):26Y32. Available at: http://www.clspectrum.com/articleviewer.aspx?articleID=104343. Accessed January 19, 2014. 7. Van der Worp E. A Guide to Scleral Lens Fitting [monograph online]. Scleral Lens Education Society. Available at: http://commons.pacificu.edu/ mono/4/. Accessed January 19, 2014. 8. Hall LA, Young G, Wolffsohn JS, Riley C. The influence of cor neoscleral topography on soft contact lens fit. Invest Ophthalmol Vis Sci 2011;52:6801Y6. 9. Hall LA, Hunt C, Young G, Wolffsohn J. Factors affecting corneo scleral topography. Invest Ophthalmol Vis Sci 2013;54:3691Y701. 10. Yoo C, Eom YS, Suh YW, Kim YY. Central corneal thickness and anterior scleral thickness in Korean patients with open-angle glaucoma: an anterior segment optical coherence tomography study. J Glaucoma 2011;20:95Y9. 11. Dunne MC, Davies LN, Wolffsohn JS. Accuracy of cornea and lens biometry using anterior segment optical coherence tomography. J Biomed Opt 2007;12:064023. 12. Liu X, Wang F, Xiao Y, Ye X, Hou L. Measurement of the limbusinsertion distance in adult strabismus patients with anterior segment optical coherence tomography. Invest Ophthalmol Vis Sci 2011;52: 8370Y3. 13. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307Y10. 14. Ishikawa H, Inazumi K, Liebmann JM, Ritch R. Inadvertent corneal indentation can cause artifactitious widening of the iridocorneal
15.
16.
17.
18.
19. 20.
21.
22. 23.
angle on ultrasound biomicroscopy. Ophthalmic Surg Lasers 2000;31: 342Y5. Radhakrishnan S, Goldsmith J, Huang D, Westphal V, Dueker DK, Rollins AM, Izatt JA, Smith SD. Comparison of optical coherence tomography and ultrasound biomicroscopy for detection of narrow anterior chamber angles. Arch Ophthalmol 2005;123:1053Y9. Dada T, Sihota R, Gadia R, Aggarwal A, Mandal S, Gupta V. Comparison of anterior segment optical coherence tomography and ultrasound biomicroscopy for assessment of the anterior segment. J Cataract Refract Surg 2007;33:837Y40. Garcia JP, Jr, Rosen RB. Anterior segment imaging: optical coherence tomography versus ultrasound biomicroscopy. Ophthalmic Surg Lasers Imaging 2008;39:476Y84. Swartz TS, Cohen R, Lin RA, Buliano M, Chu YR. Maps and scales. In: Wang M, Swartz TS, eds. Corneal Topography, 2nd ed. Thorofare, NJ: Slack Incorporated; 2012:21Y31. Schornack MM, Patel SV. Relationship between corneal topographic indices and scleral lens base curve. Eye Contact Lens 2010;36:330Y3. Sorbara L, Maram J, Fonn D, Woods C, Simpson T. Metrics of the normal cornea: anterior segment imaging with the Visante OCT. Clin Exp Optom 2010;93:150Y6. Luo ZK, Jacobs DS. Current and potential applications of anterior segment optical coherence tomography in contact lens fitting. Semin Ophthalmol 2012;27:133Y7. Visser ES, Visser R, Van Lier HJ. Advantages of toric scleral lenses. Optom Vis Sci 2006;83:233Y6. Lavaque AJ, Di Marco S, Yilmaz T, Liggett PE. Scleral curvature and LASIK. Ophthalmology 2006;113:157.
Won Ryang Wee Department of Ophthalmology Seoul National University College of Medicine 103 Daehak-ro, Jongno-gu, Seoul 110-799 Republic of Korea e-mail: [email protected]
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
ORIGINAL ARTICLE
Measurement of Anterior Scleral Curvature Using Anterior Segment OCT Hyuk Jin Choi*, Sang-Mok Lee*, Ja Young Lee†, Seung Yong Lee†, Mee Kum Kim*, and Won Ryang Wee*
ABSTRACT Purpose. To investigate and validate methods for measuring the radius of anterior scleral curvature using anterior segment optical coherence tomography images. Methods. Twenty-four volunteers were enrolled in this study. Anterior segment optical coherence tomography images, centered on horizontal/vertical limbus, including adjacent anterior sclera, were obtained in addition to conventional images centered on the optical axis. Central horizontal, nasal, and temporal optical coherence tomography images were consolidated to a new image for subsequent analyses. The reference points of limbal surface and three scleral points were marked nasally and temporally. The radius of a best-fit circle to the six scleral points was derived (the BFC [best-fit circle] method) and the radii of two circles, the centers of which are on the optical axis and pass through the points of the scleral surface at 2 mm from the limbus nasally and temporally, were calculated (the axial method). To assess the reliability and accuracy of each method, intraobserver and interobserver agreements were analyzed and the radii of contact lenses with known curvatures were measured. Results. The mean (TSD) radius of a BFC was 13.12 (T0.80) mm. The mean (TSD) radius of nasal anterior scleral curvature (13.33 T 1.12 mm) was significantly greater than that of temporal anterior scleral curvature (12.32 T 0.77 mm) (paired samples t test, p G 0.001). The BFC and axial methods showed excellent intraobserver and interobserver agreements for measurements (intraclass correlation coefficient 9 0.75, p G 0.001), whereas both methods showed a tendency to slightly underestimate the actual curvature of a rigid contact lens of known dimensions (j0.07 T 0.13 mm [the BFC method] and j0.19 T 0.07 mm [the axial method], Wilcoxon signed rank test, p = 0.173 and p = 0.028, respectively). Conclusions. Anterior segment optical coherence tomography is a valuable tool for measuring the radii of anterior scleral curvatures by image processing and mathematical calculation and can provide useful information in specific clinical situations such as designing scleral lenses. (Optom Vis Sci 2014;91:793Y802) Key Words: anterior segment optical coherence tomography, radius of scleral curvature, sclera, scleral lens, LASIK surgery
T
he human sclera is a roughly spherical, relatively avascular, white, rigid, and dense connective tissue that covers the globe posterior to the cornea.1 Although the sclera covers about 90% of the surface area of the eye, little is known about the mathematical parameters related to the shape beyond the limbus. Generally, it has been known that the sclera is an incomplete
*MD, PhD † MD Department of Ophthalmology, Seoul National University Hospital Healthcare System Gangnam Center, Seoul, Republic of Korea (HJC); Department of Ophthalmology, Seoul National University College of Medicine, Seoul, Republic of Korea (HJC, S-ML, MKK, WRW); Department of Ophthalmology, Incheon Medical Center, Incheon, Republic of Korea (JYL); and Department of Ophthalmology, Eulji University Hospital, Daejeon, Republic of Korea (SYL). Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Web site (www.optvissci.com).
sphere with an average outer diameter of 24 mm and an average inner diameter of 23 mm (mean radius of curvature, 11.5 mm).1,2 Measurements of scleral curvature can be applied to several clinical situations. The ability to image the contour of the anterior sclera would be important for designing scleral lenses that are supported entirely by the sclera and completely vault the cornea and the limbus. It could also provide a guideline for selecting proper suction rings during LASIK surgery. To analyze the shape of the anterior segment beyond the corneoscleral junction, high-quality shape information about this area should be provided by sophisticated devices such as anterior segment optical coherence tomography (AS-OCT), ultrasound biomicroscopy (UBM), and Scheimpflug camera system.3 The versatility, accuracy, and high resolution of AS-OCT have widened its application to surgical planning and follow-up tools such as corneal refractive surgery and corneal transplantation, anterior
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
794 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
chamber biometry measurement, angle assessment, and diagnosis of various anterior segment diseases including keratoconus and tumors.4 Although AS-OCT is a useful tool that can visualize the shape beyond the limbus, to the best of our knowledge, there are only a few studies that analyze anterior scleral curvature in the anterior segment images obtained using AS-OCT.5Y9 These studies gave clinicians new information, that is, that the peripheral cornea often continues to the sclera in a straight line (tangential).8,9 Furthermore, interestingly, the anterior scleral shape is also tangential in many cases and one or more sections of the sclera are often steeper or flatter than the rest (nonrotationally symmetric in nature).5Y9 The authors were interested in assessing the shape of the sclera and devised methods for analyzing anterior scleral curvature using AS-OCT images, with easily available software, and simple mathematical concepts. In these methods, points at approximately 2 mm from the limbus were thought to be clinically important because largediameter contact lenses settle down, and suction rings in the LASIK procedure come in contact with the eyeball at these points. Particularly, the authors have focused on the axial curvature, based on the fact that hard contact lenses are made by lathe cutting, which puts the lens material on a rotating mount, whereas machine cutting instruments sculpt away excess lens material to carve a precision-cut lens. Therefore, the aims of this study were to introduce new methods for analyzing anterior scleral curvature using AS-OCT images and to evaluate repeatability and reproducibility of the methods by applying them to images from volunteers with normal ocular health.
METHODS Subjects The research was in accordance with the tenets of the Declaration of Helsinki, and the protocol of this study was approved by the institutional review board of Seoul National University Hospital. Informed consent was obtained from the subjects after providing explanation regarding the nature and possible consequences of the study. A total of 24 Korean volunteers were recruited into this study. Volunteers who had previous conjunctival or scleral diseases, ocular surgeries involving the conjunctiva or sclera, and conjunctival degenerative changes such as pterygium and pinguecula were excluded. The mean (TSD) age of the participants was 31.3 (T6.5) years (range, 24 to 54 years). Of the 24 subjects, 12 were men and 12 were women.
Image Acquisition and Processing Data and observations were collected during one study visit. All AS-OCT measurements were performed by one operator. In the anterior segment mode, a total of six images (16 mm wide) for each eye were obtained using the Visante OCT instrument (Carl Zeiss Meditec, Dublin, CA). Two images were obtained horizontally and vertically while the subject fixed an eye on the focusing spot inside the instrument. When the OCT scan was centered on the corneal vertex reflex, the operator could see a bright vertical flare line crossing the corneal vertex. During image acquisition, the operator removed the tilt from the eye image by adjusting the internal fixation target to compensate for the angle between the visual axis and the optical axis (angle alpha), which implied that the bright vertical flare line corresponded approximately
to the optical axis. Then, perpendicular images in the four cardinal directions of gaze were also captured while the subject fixed an eye on the upper, lower, nasal, and temporal external targets, which were located at 15 degrees from the primary position. All images were corrected for distortion using the built-in proprietary image-correction algorithm (dewarping; Software Version 3.0.1.8). With built-in calipers, a reference value of 1.00 mm was marked on each image. Images of the right eye of each participant were used for the subsequent analyses. To obtain an image for analyses, three AS-OCT images of each patient (central horizontal, nasal, and temporal) were composed to a new one by image rotation and movement without resizing using Microsoft PowerPoint (version 2007, Redmond, WA). In brief, nasal and temporal images were rotated and overlapped on the central horizontal image using the scleral spur as a reference point of superimposition. The best composite image, in which the corneal and scleral surface texture of the three images was best matched, was obtained and saved as a new image file. The composite procedure showed excellent repeatability as assessed by one author (S-ML) and excellent reproducibility between two different authors (S-ML and HJC) (Fig. A1 and Table A1Vsee Appendix, available at http://links.lww.com/OPX/A177). The composite images obtained by one of the authors (S-ML) were used for subsequent analyses. Meanwhile, vertical images could not be analyzed because the texture of superior and inferior images was often too distorted to be able to compose them, which was mainly caused by excessive up/down gaze or the operator’s digital pressure for adequate visualization of the sclera.
Analysis of Radius of a Best-Fit Circle to the Anterior Sclera To evaluate radius of a best-fit circle (BFCR) to the anterior sclera, each composed image was analyzed by authors using Image-Pro Plus 4.5 software (Media Cybernetics, Inc, Rockville, MD; Fig. 1A). Briefly, a reference value of 1.00 mm was converted to the corresponding number of pixels. As described in a previous report,10 the reference points of the limbal surface (NL in the nasal surface and TL in the temporal surface), at which the line passing the scleral spur met perpendicularly, and the scleral surface points N2/T2 (2 mm posterior to NL/TL) were marked. Then, the authors marked the scleral surface points 1 mm apart from N2 (N1 and N3) and T2 (T1 and T3) based on the assumption that N1 to N3 (T1 to T3) were on the same circle and the normal to the line joining N1 and N3 (T1 and T3) was the same normal to the scleral surface at N2 (T2). The first high reflective tissue signal from the episclera was considered as the scleral surface. Finally, the radius of the circle, which best fits the six scleral surface points (N1, N2, N3, T1, T2, and T3), was derived. Each value was converted to the metric unit (millimeter) by dividing it by reference pixels corresponding to 1.00 mm.
Analysis of Axial Radius of Scleral Curvature To measure axial radius of scleral curvature (AR), composite images were analyzed by authors using Image J software version 1.44d (National Institutes of Health, Bethesda, MD; http://rsb.info.nih.gov/ij/index.html, Fig. 1B). Firstly, the authors decided the x, y coordinates of two points on the anterior corneal
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
795
FIGURE 1. Measurements of the radius of a best-fit circle to the anterior sclera (BFCR), the axial radii of anterior scleral curvature (AR), and the radii of contact lenses. To measure BFCR and AR, a composite AS-OCT image was used. (A) The radius of a best-fit circle to six scleral surface points (N1, N2, N3, T1, T2, and T3) was regarded as BFCR. (B) The distance from the point at which the T line and the A line meet to T2 was regarded as the axial radius of temporal scleral curvature. (C) The front curve radius of contact lenses was derived using the same BFC and axial methods. NL, the reference point of the nasal limbal surface that corresponds to the scleral spur; TL, the reference point of the temporal limbal surface that corresponds to the scleral spur; N1 to N3, three consecutive surface points from the NL; T1 to T3, three consecutive surface points from the TL; T line, the line that passes through T2 and bisects the lines passing through T1 and T3 at a right angle; A line, the axis line. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
796 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
surface and the anterior lens surface, respectively, which were lying on the axis line (A line). The bright vertical flare line (the optical axis) was used as the A line in most cases, whereas the geometric axis line (defined as the line bisecting the bases of the anterior chamber, which could be drawn between the scleral spurs on both sides) was used as the A line when the bright flare line was hard to visualize. Then, the authors determined the x, y coordinates of the points (N1, N2, N3, T1, T2, and T3 as described in the measurement of BFCR) on the scleral surface. With the data of these points, the AR was calculated mathematically using Microsoft
Excel spreadsheet (version 2007, Redmond, WA) according to the concept described below. For calculating the radius of temporal scleral curvature (ARt), the authors drew a line (T line) that passed through T2 and bisected the lines passing through T1 and T3 at a right angle. The authors then calculated the point where the T line met the A line. The distance from this point to T2 was regarded as ARt, and the same concept was applied for calculating the radius of nasal scleral curvature (ARn). Each value was converted to the metric unit (millimeter) by dividing it by reference pixels corresponding to 1.00 mm.
FIGURE 2. Radii of the anterior scleral curvature. (A) The mean values of BFCR, ARn, and ARt (*paired samples t test, p G 0.001). (B) Correlation between ARn and ARt (Pearson correlation coefficient, r = 0.515, p = 0.01). (C) Correlation between BFCR and ARn (Pearson correlation coefficient, r = 0.757, p G 0.001). (D) Correlation between BFCR and ARt (Pearson correlation coefficient, r = 0.856, p G 0.001). BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal scleral curvature; ARt, axial radius of temporal scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
Validation of Accuracy To validate the accuracy of the abovementioned techniques in calculating radius of curvature, six custom-made spherical rigid gas-permeable (RGP) contact lenses of the same material (Boston XO; hexafocon A), refractive indices (1.415), overall diameters (11.0 mm), and uniform thicknesses (0.5 mm) but with front surface radii between 7.60 and 8.60 mm in 0.2-mm steps were tested. The authors measured only front surface radii because refractive indices of the contact lens material and the cornea were different, which caused a more prominent distortion in the back surface image of the contact lenses after dewarping. Commercially available spherical RGP contact lenses (Lucid Korea Co Ltd) were used for the validation. The manufacturing tolerance for the base curve radius tested by microspherometer or radius gauge was T0.05 mm. In the high-resolution mode of the AS-OCT, a 10-mm-wide horizontal image centered on the vertex reflex was obtained from each contact lens after being dewarped by the same algorithm used in the human eye. Only central high-resolution images were used for analyses because coordinate errors gradually increase toward the periphery even in the dewarped image.11 Like the analysis of human anterior segment, the central bright vertical flare line was considered as the A line and two imaginary landmark points located at 1 mm from the A line were considered as NL and TL, respectively, because there were no landmarks like scleral spurs in the human eye. Then, six surface points corresponding to N1 to N3 and T1 to T3 were marked and the analyses were performed using the above techniques (Fig. 1C). Results were presented as the mean values of three different measurements calculated by one author (HJC).
Statistical Analysis To evaluate intraobserver reliability, BFCR and ARs were evaluated twice by the same author (HJC) at a 1-week interval. To assess interobserver reliability, BFCR and ARs were also measured in the same AS-OCT images by different authors (BFCR by JYL and ARs by SYL). Both authors were blinded to the measurements calculated by the other author. Statistical analysis was performed using SPSS 18.0 (Chicago, IL) software. The intraobserver difference in the radii was analyzed using the paired samples t test. Pearson correlation coefficient and
797
intraclass correlation coefficient (ICC) were used to assess intraobserver and interobserver reliability. The guidelines used for the interpretation of the ICC were as follows: an ICC of less than 0.40 indicated poor reliability, an ICC of 0.40 to 0.75 indicated fair to good reliability, and an ICC of greater than 0.75 indicated excellent reliability.12 p G 0.05 was considered statistically significant. The Bland-Altman plots,13 graphic representations illustrating repeatability of observations, were also used to show the relationship between the two results obtained by the same author or two different authors.
RESULTS The Shape of the Anterior Sclera First measurements of BFCR, ARn, and ARt calculated by one author (HJC) were analyzed to define the shape of the anterior scleral surface. The mean (TSD) BFCR was 13.12 (T0.80) mm (range, 11.69 to 14.63 mm). The mean (TSD) ARn (13.33 T 1.12 mm; range, 11.11 to 16.41 mm) was significantly greater than the mean (TSD) ARt (12.32 T 0.77 mm; range, 11.03 to 13.63 mm) (Fig. 2A, paired samples t test, p G 0.001), whereas ARn and ARt showed a significant positive correlation (Fig. 2B, Pearson correlation coefficient, r = 0.515, p = 0.01). Radius of a best-fit circle was positively correlated with ARn and ARt (Fig. 2C, D, Pearson correlation coefficient, r = 0.757 and 0.856, respectively, each p G 0.001).
Reliability of the Measurements The repeatability of two consecutive measurements (at a 1-week interval) of scleral curvature calculated by one author showed an excellent level of agreement in measurements of BFCR (ICC = 0.981), ARn (ICC = 0.951), and ARt (ICC = 0.837). The mean (TSD) absolute difference between two consecutive measurements was smaller in BFCR (0.13 T 0.09 mm; range, 0.01 to 0.29 mm) than in ARn (0.36 T 0.24 mm; range, 0.12 to 0.92 mm) or ARt (0.34 T 0.26 mm; range, 0 to 1.10 mm). The Bland-Altman plots showed 24 of 24 (100%) in BFCR and 23 of 24 (96%) in ARn/ ARt measurement differences were within the 95% confidence interval of the mean difference (Table 1 and Fig. 3). Excellent interobserver reproducibility for measurements of BFCR (ICC = 0.974), ARn (ICC = 0.941), and ARt (ICC = 0.841)
TABLE 1.
Reliability of the measurements
First measurement, mm Intraobserver reliability BFCR 13.12 T 0.80 ARn 13.33 T 1.12 Art 12.32 T 0.77 Interobserver reliability BFCR 13.12 T 0.80 ARn 13.33 T 1.12 Art 12.32 T 0.77
Second measurement, mm
Absolute difference, mm
13.11 T 0.87 13.60 T 1.07 12.27 T 0.74 13.15 T 0.87 13.01 T 1.16 12.33 T 0.78
Pearson correlation coefficient
Intraclass correlation coefficient
r
p
ICC
p
0.13 T 0.09 0.36 T 0.24 0.34 T 0.26
0.984 0.952 0.838
G0.001 G0.001 G0.001
0.981 0.951 0.837
G0.001 G0.001 G0.001
0.15 T 0.12 0.40 T 0.30 0.34 T 0.26
0.977 0.942 0.841
G0.001 G0.001 G0.001
0.974 0.941 0.841
G0.001 G0.001 G0.001
BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal anterior scleral curvature; ARt, axial radius of temporal anterior scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
798 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
FIGURE 3. Intraobserver reliability presented by Bland-Altman plots. The same author measured BFCR (A), ARn (B), and ARt (C) twice at a 1-week interval. Solid lines, the mean difference between two measurements; dotted lines, the 95% confidence interval of the mean difference; BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal scleral curvature; ARt, axial radius of temporal scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
was also evident. The mean (TSD) absolute difference between two measurements calculated by two different authors was also smaller in BFCR (0.15 T 0.12 mm; range, 0 to 0.48 mm) than in ARn (0.40 T 0.30 mm; range, 0.01 to 1.07 mm) or ARt (0.34 T 0.26 mm; range, 0.01 to 1.14 mm), whereas the Bland-Altman plots showed 23 of 24 (96%) in BFCR/ARn/ARt measurement differences were within the 95% confidence interval of the mean difference (Table 1 and Fig. 4).
Accuracy of the Measurements Table 2 shows the results of accuracy testing of both BFC and axial methods using contact lenses with different front curve radii. There were small steeper deviations from the actual values in both methods, although only the axial method showed a significant underestimation of the actual curvature (Wilcoxon signed rank test, p = 0.028). The mean deviation of the BFC method (j0.07 T 0.13 mm) was smaller than that of the axial method (j0.19 T 0.07 mm).
DISCUSSION In this study, the authors validated the methods for measuring the radius of a ‘‘best-fit circle’’ to the anterior sclera and ‘‘axial’’ radius of anterior scleral curvature at the points of nasal and temporal scleral surfaces 2 mm from the limbus, using a composite image derived from three horizontal AS-OCT images. The BFCR of the anterior sclera in 24 Korean volunteers was 13.12 mm. The radius of anterior scleral curvature derived from the ‘‘axial’’ method was significantly greater in the nasal sclera (13.33 mm) than in the temporal sclera (12.32 mm). In terms of repeatability and reproducibility, both ‘‘BFC’’ and ‘‘axial’’ methods showed excellent intraobserver and interobserver agreement. Moreover, both the ‘‘BFC’’ and ‘‘axial’’ methods were found to only slightly underestimate the actual anterior curvature values of RGP contact lenses of known dimensions. In this study, the authors used AS-OCT despite the availability of UBM because AS-OCT has several advantages over UBM for objective analysis of anterior scleral imaging. Image acquisition by AS-OCT is a noncontact technique and a less time-consuming procedure, which increases the participant’s comfort level as well as safety. Unlike in AS-OCT, supine positioning and inadvertent pressure on the eyecup may alter the natural shape of the anterior segment during UBM examination.14Y16 This alteration may be more prominent during the participant’s lateral gaze because more pressure needs to be applied to the eyecup. The scanning resolution with standard software is higher in AS-OCT (60 Km of lateral resolution and 18 Km of axial resolution) than in UBM (50 Km of lateral resolution and 25 Km of axial resolution).4,15,17 Furthermore, the scleral spur, which was an especially important landmark for image processing in this study, is more distinct on AS-OCT images.16 Using AS-OCT images, the authors introduced two different ways for evaluating anterior scleral curvature (designated as the ‘‘BFC’’ method and the ‘‘axial’’ method, respectively) and obtained valuable results as described above. The BFCR of the anterior sclera was 13.12 mm, which is comparable to the average anterior scleral radius (12.40 mm; range, 10.10 to 16.60 mm) derived by manually drawing a forced circle through the anterior sclera in a previous report.6 The temporal curvature of the anterior
799
sclera was significantly steeper than the nasal curvature, which confirms the fact that the human eyeball is not a perfect sphere.1,2 Actually, the readers can recognize this fact at a glance in the composite AS-OCT images (Fig. 1A, B), and this result corresponds well with those of previous reports, which demonstrated that the mean scleral curvature is steepest in the temporal sclera but similar in the nasal, superior, and inferior scleral planes8,9 and the ocular surface beyond the cornea is not rotationally symmetric.6 This could be partly explained by the fact that the medial rectus muscle inserts most anteriorly into the sclera and the visual axis forms an angle of 23 degrees with the orbital axis in the primary position, which means that the medial rectus muscle adducts the eyeball with subsequent flattening of the nasal anterior sclera. Regarding intraobserver repeatability and interobserver reproducibility, both BFC and axial methods showed excellent level of agreements, which were higher compared with the method (intrasession ICC = 0.83) using built-in calipers and protractor tools introduced by Hall et al.8 In a subsequent study involving 204 subjects,9 Hall et al. reported that scleral radii of curvature ranged from j57.4 to 312.5 mm (mean, 35.5 mm) in the nasal sclera and from 3.1 to 100.0 mm (mean, 22.4 mm) in the temporal sclera, which are quite different from the results of this study and imply that the radii represent instantaneous curvature that has a higher degree of variability from point to point like that in the tangential map of corneal topographer.18 Unfortunately, the authors could not find any detailed descriptions about how scleral radii were derived and at which point scleral radii were analyzed in the previous reports.8,9 Moreover, the authors could not compare the methods used in this study with those used in other studies because curvature measurement tools could not be provided by a local distributor of AS-OCT. Assignment of multiple points to fit in the BFC method and introduction of the reference plane (A line) into the axial method may be the key factors for decreasing the variations in measurements between sessions and subjects. In terms of accuracy of the measurements, both BFC and axial methods slightly underestimated the actual front curve radius of custom-made RGP contact lenses. The mean deviations of BFC (j0.07 mm) and axial (j0.19 mm) methods were comparable with that (j0.11 mm) in a previous report,3 in which base curve radii of RGP contact lenses were calculated using sag and chord measurements. These results seem to be attributed not only to an error during image processing because refractive indices of contact lens material (1.415) and corneal tissue (1.376) are different but also to a tendency toward underestimation of corneal surface curvature in measurements even after correction with built-in AS-OCT software.11 Measurements of scleral curvature would be of great value in designing scleral lenses. As the currently available corneal topographers do not acquire imaging data beyond the limbus, topographic indices cannot be used to accurately predict the base curve of scleral lenses.19 Neither the central radius of the cornea nor the corneal sagittal depth is reported to be highly predictive of the scleral shape, which is indicated by nasal and temporal angles at the 15-mm chord.20 Therefore, current scleral lens fitting is mainly based on an iterative system of modifications to trial lenses, a rather cumbersome and time-consuming process. Introduction of fitting paradigms based on the sagittal depth and information on scleral contour, which could be simultaneously obtained with
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
800 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
FIGURE 4. Interobserver reliability presented by Bland-Altman plots. Two different authors measured BFCR (A), ARn (B), and ARt (C), while both authors were blinded to the measurements calculated by the other author. Solid lines, the mean difference between two measurements; dotted lines, the 95% confidence interval of the mean difference; BFCR, the radius of a best-fit circle to the anterior sclera; ARn, axial radius of nasal scleral curvature; ARt, axial radius of temporal scleral curvature. Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al.
801
TABLE 2.
Accuracy of the measurements Actual FCR, mm 7.60 7.80 8.00 8.20 8.40 8.60 Mean T SD p*
BFCR method
AR method
Calculated FCR, mm
Difference, mm
Calculated FCR, mm
Difference, mm
7.72 7.75 7.98 8.03 8.14 8.54
0.12 j0.05 j0.02 j0.17 j0.26 j0.06 j0.07 T 0.13 0.173
7.28 7.65 7.80 8.02 8.26 8.44
j0.32 j0.15 j0.20 j0.18 j0.14 j0.16 j0.19 T 0.07 0.028
Results were presented as the mean values of three different measurements calculated by one author (HJC). *Wilcoxon signed rank test. FCR, front curve radius; BFCR, the radius of a best-fit circle to the lens; AR, axial radius of curvature.
AS-OCT, seem to offer a more valuable fitting method.21 Actually, in a small case series, AS-OCT-based sag and chord measurements were reported to be highly useful fitting tools for scleral lenses that seem to offer more accurate and efficient alternatives to the trial lens method.3 Measurements of anterior scleral curvature as in this study may provide additional useful information especially for designing the landing zone of scleral lenses. Moreover, the nonspherical nature of the sclera confirmed by AS-OCT may help the practitioners select nonrotationally symmetric scleral lenses such as toric and quadrant-specific lenses to minimize conjunctival blanching and to improve comfort.5Y7,22 Anterior segment optical coherence tomography could also serve as a valuable tool before LASIK surgery. Scleral curvature measurement will greatly help the refractive surgeon during the process of selecting the correct suction ring, avoiding deformation of the globe wall that induces a secondary traction in the vitreous cavity and an anterior displacement of the posterior pole.23 Furthermore, better refraction outcomes can be achieved by selecting an appropriate suction ring because we can expect more predictable values for the thickness and diameter of the corneal flap by avoiding deformation of the eyeball during the application of the suction ring. There are some limitations to this study. The sample size of 24 was small. Hence, a larger study is necessary for drawing more solid conclusions. When the device conducted a dewarping process on the perpendicular images beyond the cornea, it drew the front line along the anterior scleral surface and the back line along the posterior aspect of the intraocular muscle and inner scleral lamina. Therefore, the scleral contour beyond the cornea might not be the same as the actual contour because the refractive index of the involved tissue is different from that of the cornea. Nevertheless, the authors tried to reduce image distortion by rotating the eyeball minimally, by locating the sclera near the center within vertical analysis limits, and by confirming the presence of a corneal vertex reflex (central vertical flare), which indicates that the cross section is along the corneal meridian, ensuring accurate dewarping of the image. Regarding the accuracy of the procedure, the validation using larger-diameter contact lenses such as scleral lenses appears to be more relevant, given that the scleral measures on the participants were beyond the cornea and required three images to be stitched together. However, it was impossible because AS-OCT continuously failed to proceed with dewarping of the images from
the scleral lenses. The authors were only able to analyze horizontal scleral curvature. While capturing vertical scleral images, shadowing of the eyelids often obscured anterior segment images because of low palpebral fissure height and greater eyelid tension in Asians. Excessive up/down gaze and digital retraction of the eyelids induced a distortion of the original shape of the anterior segment owing to the pulling effect of extraocular muscles (flattening the anterior scleral curvature) and inadvertent digital pressure on the globe, respectively. Similarly, eye turning during nasal and temporal measurements could also induce a change in the scleral shape. Volunteers who had conjunctival or subconjunctival lesions near the limbus were not good candidates because AS-OCT could not clearly discriminate these lesions from the underlying sclera. Therefore, subjects who had conjunctival pathology such as pterygium or pinguecula were excluded from this study. In conclusion, AS-OCT is a useful tool for measuring the radius of scleral curvature, which can be useful in some optometric and ophthalmological practices such as designing scleral lenses and choosing a proper suction ring for LASIK surgery.
ACKNOWLEDGMENTS This study was supported by a grant from the Korean Health Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (Project Nos. A084496 and A120018 [HI12C0015]). The test contact lenses were kindly provided by Lucid Korea Co. Ltd. The authors declare that they have no competing financial interests. Received August 26, 2013; accepted April 15, 2014.
APPENDIX The appendix, including Table A1 and Fig. A1, demonstrating validation of the image composition procedure, is available at http:// links.lww.com/OPX/A177.
REFERENCES 1. Dawson DG, Ubels JL, Edelhauser HF. Cornea and sclera. In: Levin LA, Nilsson SFE, Hoeve JV, Wu S, Kaufman PL, Alm A, eds. Adler’s Physiology of the Eye, 11th ed. Philadelphia, PA: Saunders; 2011: 71Y130. 2. Watson PG, Young RD. Scleral structure, organisation and disease. A review. Exp Eye Res 2004;78:609Y23.
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
802 Anterior Scleral Curvature Using Anterior Segment OCTVChoi et al. 3. Gemoules G. A novel method of fitting scleral lenses using high resolution optical coherence tomography. Eye Contact Lens 2008; 34:80Y3. 4. Ramos JL, Li Y, Huang D. Clinical and research applications of anterior segment optical coherence tomographyVa review. Clin Experiment Ophthalmol 2009;37:81Y9. 5. Van der Worp E. New technology in contact lens practice. Contact Lens Spectrum 2010;25(2):22Y9. Available at: http://www.clspectrum.com/ articleviewer.aspx?articleID=103880. Accessed January 19, 2014. 6. Van der Worp E, Graf T, Caroline PJ. Exploring beyond the corneal borders. Contact Lens Spectrum 2010;25(6):26Y32. Available at: http://www.clspectrum.com/articleviewer.aspx?articleID=104343. Accessed January 19, 2014. 7. Van der Worp E. A Guide to Scleral Lens Fitting [monograph online]. Scleral Lens Education Society. Available at: http://commons.pacificu.edu/ mono/4/. Accessed January 19, 2014. 8. Hall LA, Young G, Wolffsohn JS, Riley C. The influence of cor neoscleral topography on soft contact lens fit. Invest Ophthalmol Vis Sci 2011;52:6801Y6. 9. Hall LA, Hunt C, Young G, Wolffsohn J. Factors affecting corneo scleral topography. Invest Ophthalmol Vis Sci 2013;54:3691Y701. 10. Yoo C, Eom YS, Suh YW, Kim YY. Central corneal thickness and anterior scleral thickness in Korean patients with open-angle glaucoma: an anterior segment optical coherence tomography study. J Glaucoma 2011;20:95Y9. 11. Dunne MC, Davies LN, Wolffsohn JS. Accuracy of cornea and lens biometry using anterior segment optical coherence tomography. J Biomed Opt 2007;12:064023. 12. Liu X, Wang F, Xiao Y, Ye X, Hou L. Measurement of the limbusinsertion distance in adult strabismus patients with anterior segment optical coherence tomography. Invest Ophthalmol Vis Sci 2011;52: 8370Y3. 13. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307Y10. 14. Ishikawa H, Inazumi K, Liebmann JM, Ritch R. Inadvertent corneal indentation can cause artifactitious widening of the iridocorneal
15.
16.
17.
18.
19. 20.
21.
22. 23.
angle on ultrasound biomicroscopy. Ophthalmic Surg Lasers 2000;31: 342Y5. Radhakrishnan S, Goldsmith J, Huang D, Westphal V, Dueker DK, Rollins AM, Izatt JA, Smith SD. Comparison of optical coherence tomography and ultrasound biomicroscopy for detection of narrow anterior chamber angles. Arch Ophthalmol 2005;123:1053Y9. Dada T, Sihota R, Gadia R, Aggarwal A, Mandal S, Gupta V. Comparison of anterior segment optical coherence tomography and ultrasound biomicroscopy for assessment of the anterior segment. J Cataract Refract Surg 2007;33:837Y40. Garcia JP, Jr, Rosen RB. Anterior segment imaging: optical coherence tomography versus ultrasound biomicroscopy. Ophthalmic Surg Lasers Imaging 2008;39:476Y84. Swartz TS, Cohen R, Lin RA, Buliano M, Chu YR. Maps and scales. In: Wang M, Swartz TS, eds. Corneal Topography, 2nd ed. Thorofare, NJ: Slack Incorporated; 2012:21Y31. Schornack MM, Patel SV. Relationship between corneal topographic indices and scleral lens base curve. Eye Contact Lens 2010;36:330Y3. Sorbara L, Maram J, Fonn D, Woods C, Simpson T. Metrics of the normal cornea: anterior segment imaging with the Visante OCT. Clin Exp Optom 2010;93:150Y6. Luo ZK, Jacobs DS. Current and potential applications of anterior segment optical coherence tomography in contact lens fitting. Semin Ophthalmol 2012;27:133Y7. Visser ES, Visser R, Van Lier HJ. Advantages of toric scleral lenses. Optom Vis Sci 2006;83:233Y6. Lavaque AJ, Di Marco S, Yilmaz T, Liggett PE. Scleral curvature and LASIK. Ophthalmology 2006;113:157.
Won Ryang Wee Department of Ophthalmology Seoul National University College of Medicine 103 Daehak-ro, Jongno-gu, Seoul 110-799 Republic of Korea e-mail: [email protected]
Optometry and Vision Science, Vol. 91, No. 7, July 2014
Copyright © American Academy of Optometry. Unauthorized reproduction of this article is prohibited.
