November 2012
Volume 53, Issue 12
Free
Glaucoma  |   November 2012
Improved Reproducibility in Measuring the Laminar Thickness on Enhanced Depth Imaging SD-OCT Images Using Maximum Intensity Projection
Author Affiliations & Notes
  • Eun Ji Lee
    From the Department of Ophthalmology, Seoul National University College of Medicine, Seoul National University Bundang Hospital, Seongnam, Korea; and the
  • Tae-Woo Kim
    From the Department of Ophthalmology, Seoul National University College of Medicine, Seoul National University Bundang Hospital, Seongnam, Korea; and the
  • Robert N. Weinreb
    Hamilton Glaucoma Center and Department of Ophthalmology, University of California San Diego, La Jolla, California.
  • Corresponding author: Tae-Woo Kim, Department of Ophthalmology, Seoul National University Bundanga Hospital, Seoul National University College of Medicine, 166 Gumi-dong, Bundang-gu, Seongnam, Gyeonggi-do 463-707, Korea; twkim7@snu.ac.kr
Investigative Ophthalmology & Visual Science November 2012, Vol.53, 7576-7582. doi:10.1167/iovs.12-10305
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Eun Ji Lee, Tae-Woo Kim, Robert N. Weinreb; Improved Reproducibility in Measuring the Laminar Thickness on Enhanced Depth Imaging SD-OCT Images Using Maximum Intensity Projection. Invest. Ophthalmol. Vis. Sci. 2012;53(12):7576-7582. doi: 10.1167/iovs.12-10305.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose.: To investigate whether the thin-slab maximum intensity projection (MIP) image improves the reproducibility of the measurement of spectral domain optical coherence tomography (SD-OCT) based lamina cribrosa thickness (LCT).

Methods.: Optic discs of 63 open angle glaucoma patients or glaucoma suspects were scanned with enhanced depth imaging spectral domain optical coherence tomography (SD-OCT). The B-scan images were constructed three-dimensionally using maximum intensity projection (MIP). The whole MIP volume was then cut into thin slabs of two scan-line distance width (approximately 64 μm) at three 3 locations: mid-horizontal, and superior and inferior mid-peripheral regions of the optic nerve head. The LCT was measured in the thin-slab MIP images (LCT-MIP) and the three B-scans corresponding to each thin-slab MIP image (LCT-Bscan) at the three regions in each eye. The interobserver and intraobserver reproducibility of measuring the LCT-MIP and LCT-Bscan were evaluated by calculating the intraclass correlation coefficient (ICC), coefficient of variation (CV), and coefficient of repeatability (CR). The agreement between the LCT-MIP and the LCT-Bscan was assessed by Bland-Altman analysis.

Results.: The thin-slab MIP images provided better image contrast with more distinct borders of the lamina cribrosa compared to the B-scan images. Both intraobserver and interobserver ICCs were higher for LCT-MIP than for LCT-Bscan in all areas. The 95% limit of agreement of measurement differences between LCT-Bscan and LCT-MIP ranged from −29.43 to 32.55 μm.

Conclusions.: The thin-slab MIP images provided a higher reproducibility in measuring the LCT using SD-OCT than the B-scan images. With thin-slab MIP images, the detection of the laminar borders was more straightforward. This technique should facilitate the evaluation of the lamina cribrosa.

Introduction
Results of recent experimental and computational studies suggest that biomechanical factors may contribute to optic nerve injury in glaucoma. 15 One of the principal elements of optic nerve biomechanics is the translaminar pressure gradient (TLPG). Earlier works suggests that a higher TLPG may promote axonal injury by blockade of the axoplasmic flow across the lamina cribrosa (LC). 69 Recently, it has been demonstrated that a higher translaminar pressure difference (intraocular pressure minus cerebrospinal fluid pressure) correlates with loss of neuroretinal rim area 10 and visual field defect. 10,11  
The TLPG is positively correlated with the translaminar pressure difference and inversely correlated with the lamina cribrosa thickness (LCT). Thus, a higher translaminar pressure difference or a thinner LC may generate a greater TLPG. In this context, LCT has been the focus of several studies that have aimed to elucidate the pathophysiology of the development and progression of glaucomatous optic neuropathy. 1220  
With the emergence of spectral domain optical coherence tomography (SD-OCT), it now is possible to evaluate LC morphology in human patients. Studies using SD-OCT have shown that the LC is thinner in glaucoma patients than in glaucoma suspects or healthy subjects. 16,17 It has also been demonstrated that the LCT is correlated with visual field mean deviation. 18 However, it is often difficult to determine the borders of the LC, especially the posterior border. This is attributable to the low image contrast either due to distant location of the posterior LC border from the zero-decay line in OCT imaging or the interbundle connective tissue, which has high reflectivity similar to the LC. Moreover, vascular shadowing often obscures the LC. 
Recently, Girard et al. 21 demonstrated dramatically improved visibility of the LC by applying shadow removal and a contrast enhancement algorithm. However, they did not report whether such an algorithm improves the delineation of the posterior LC border. We approached this problem differently, using a thin-slab image of the maximum intensity projection (MIP). With the MIP image, only the signal with maximum intensity along the projection path was displayed in a two-dimensional (2D) projection image. When it was viewed from multiple angles, the object could be evaluated three-dimensionally. Previously, we have demonstrated that MIP may be useful for three-dimensional (3D) evaluation of the LC. 22 As described in our previous study, the MIP image may be cut into thin-slab images. We had the impression that it was often easier to delineate the LC borders in the thin-slab MIP images than in the B-scan images. 
The purpose of this study was to investigate whether the reproducibility of SD-OCT–based LCT measurement was improved by this algorithm. 
Methods
This study used the images of open angle glaucoma (OAG) patients and glaucoma suspects who were included in the Lamina Cribrosa Exploration Study. 16,22  
This study was approved by the Seoul National University Bundang Hospital Institutional Review Board. All subjects were treated in accordance with the Declaration of Helsinki. 
Study Subjects
All subjects who were included in the Lamina Cribrosa Exploration Study received comprehensive ophthalmic examinations that included visual acuity measurement, Goldmann applanation tonometry, refraction tests, slit-lamp biomicroscopy, gonioscopy, and dilated stereoscopic examination of the optic disc. They also underwent SD-OCT (Spectralis OCT; Heidelberg Engineering GmbH, Heidelberg, Germany) and standard automated perimetry (Humphrey Field Analyzer II 750; 24-2 Swedish interactive threshold algorithm; Carl-Zeiss Meditec, Dublin, CA). 
To be included, eyes were required to have been diagnosed either as OAG or glaucoma suspect, and to have a best-corrected visual acuity of ≥20/40 with spherical refraction within ± 6.0 diopters (D), and cylinder correction within ± 3.0 D. OAG was defined as having an open angle on gonioscopy, glaucomatous optic nerve damage (i.e., the presence of focal thinning, notching), and associated visual field defect without ocular disease or conditions that may elevate the IOP. A glaucomatous visual field change was defined as (1) outside normal limits on glaucoma hemifield test; or (2) three abnormal points with P < 5% probability of being normal, one with P < 1% by pattern deviation; or (3) pattern standard deviation of <5% if the visual field was otherwise normal, confirmed on two consecutive tests. 
Glaucoma suspect was defined either as ocular hypertension (IOP >21 mm Hg) or the presence of glaucomatous optic nerve damage without visual field defect. 
Those with a history of ocular surgery other than cataract extraction and glaucoma surgery, and intraocular disease (e.g., diabetic retinopathy or retinal vein occlusion) or neurologic disease (e.g., pituitary tumor) that could cause visual field loss, were excluded from the study. Eyes were also excluded when any of the nine selected B-scans had poor quality that did not allow the observers to determine the LC borders. 
Enhanced Depth Imaging OCT of the Optic Disc
The optic nerve head (ONH) was imaged using optical coherence tomography (OCT) (Spectralis OCT; Heidelberg Engineering GmbH) with the enhanced depth imaging technique. Approximately 65 horizontal B-scan section images covering the optic disc, 30 to 34 μm apart (the scan-line distance being determined automatically by the instrument), were obtained from each eye. Each section had 42 OCT frames averaged, which provided the best trade-off between image quality and patient cooperation.16  
3D Construction of the Optic Disc Image and Creation of the Thin-Slab MIP Image
Three-dimensional volumetric rendering was performed using the MIP technique by image-processing software (Amira 5.2.2; Visage Imaging, Berlin, Germany) as described elsewhere. 22 The MIP rendering allowed the visualization of the highest intensity in data volume along the current line of sight. From the 3D volumetric image of the optic nerve, thin-slab images were generated at three regions: superior mid-peripheral, and mid-horizontal and inferior mid-peripheral region (Fig. 1). 
Figure 1. 
 
(A) The MIP volumetric image of an ONH reconstructed by maximum intensity projection. The light-blue line indicates the location where the thin volumetric section (B) was selected. (C, E) The B-scan image at the center of the thin volume section. (D, F) XY view of the thin-slab MIP image. A more clear demarcation of the anterior border (arrows) as well as posterior border (arrowheads) of the lamina cribrosa is noted in the thin-slab MIP image. (E, F) The vertical orange line indicates the reference line, which was set near the center of the Bruch's membrane opening. The anterior and posterior borders were defined as the plane that best represented the margin of the hyper-reflective plate within the 200 μm width area around the reference line (horizontal orange lines).
Figure 1. 
 
(A) The MIP volumetric image of an ONH reconstructed by maximum intensity projection. The light-blue line indicates the location where the thin volumetric section (B) was selected. (C, E) The B-scan image at the center of the thin volume section. (D, F) XY view of the thin-slab MIP image. A more clear demarcation of the anterior border (arrows) as well as posterior border (arrowheads) of the lamina cribrosa is noted in the thin-slab MIP image. (E, F) The vertical orange line indicates the reference line, which was set near the center of the Bruch's membrane opening. The anterior and posterior borders were defined as the plane that best represented the margin of the hyper-reflective plate within the 200 μm width area around the reference line (horizontal orange lines).
The thin-slab image was obtained by selecting two planes inside the whole data. Then, only the data within these two planes was displayed in the thin-slab MIP image. Therefore, the minimal producible width of the thin slab using image-processing software (Amira 5.2.2; Visage Imaging) was one scan-line width (approximately 32 μm). In the present study, each thin-slab MIP was constructed from three serial B-scan images, thus having two scan-line widths (approximately 64 μm). 
Measurement of the Lamina Cribrosa Thickness
The LCT was measured at the mid-horizontal, and superior and inferior mid-peripheral regions of the optic nerve hypoplasia in the thin-slab MIP images as well as in the corresponding three B-scan images (Fig. 1). The mid-horizontal region was defined as the two scan-line width area located at the center of the optic disc. The superior and inferior mid-peripheral regions were defined as the two scan-line width area located between the center and the superior and inferior borders of the optic disc. 
In each B-scan or thin-slab MIP image, LCT was measured only in the central portion. To do this, a reference line was set near the center of the LC, avoiding the vascular shadow. Then the anterior and posterior borders were determined. The determination of the borders was based on the assumption that the LC was a slightly curved, plate-like structure that was seen as the highly reflective tissue in the SD-OCT images. The LC was measured at the reference line. Thus, one measurement was obtained from each B-scan or MIP image. Because the reference line may locate at the laminar pore in some B-scan images, we considered the 200 μm width area around the reference line, rather than simply selecting the border at the reference line. In such a case, defining the border based on the apparent high reflectivity line (Fig. 2A, inset, black dots) would result in a thinner measurement than the adjacent B-scans. In the MIP image, which overlaps the high reflectivity signal from the three B-scan images, lamina pores at the border were not typically seen. In this circumstance, a difference would result between the LCT B-scan and LCT-MIP. In addition, the reference line may locate at the interbundle connective tissue. In that case, defining the border at the reference line without consideration of the adjacent area would result in overestimation of the LCT. To avoid these matters and based on the idea that the LC is a plate-like structure, we defined the LC border using an imaginary line that represented the surface of the LC plate within the 200 μm width area around the reference line (Fig. 2A, inset, orange dotted line). Then, the distance between the anterior and posterior borders was measured in the direction perpendicular to the anterior LC surface at the measurement point, using a manual caliper tool of image-processing software (Amira 5.2.2; Visage Imaging) and defined as the LCT for each image. The average of the three measurements obtained from the three B-scan images in each region (one from each B-scan) was defined as LCT-Bscan for the region. For the thin-slab MIP images, only one manual measurement was required for each region. The measurement obtained from the thin-slab MIP image was defined as LCT-MIP. 
Figure 2. 
 
Three serial B-scan images (AC) and a thin-slab MIP image (D) of a glaucomatous eye. (A, inset) Black dots indicate the apparent border of the LC at the reference line. Orange dotted lines are imaginary lines that represent the surface of the LC within the 200 μm width area around the reference line (A, vertical orange line). (BD) Orange lines in the inset images indicate the anterior and posterior borders of the LC that were set to determine the LC thickness. (A, C) The borders of the LC are discontinuous either due to laminar pores or poor image contrast (black arrows). (B) The image contrast is poor at the level of the anterior LC border (arrowheads). (B, C) Note the distinct hyper-reflectivity from the retrolaminar interbundle connective tissues (white arrows), which hampers the delineation of the posterior LC borders. (D) The hyper-reflectivity from the retrolaminar interbundle connective tissue is less distinct. The posterior LC border is clearly seen as a continuous curvilinear line. The anterior LC border is also more easily delineable.
Figure 2. 
 
Three serial B-scan images (AC) and a thin-slab MIP image (D) of a glaucomatous eye. (A, inset) Black dots indicate the apparent border of the LC at the reference line. Orange dotted lines are imaginary lines that represent the surface of the LC within the 200 μm width area around the reference line (A, vertical orange line). (BD) Orange lines in the inset images indicate the anterior and posterior borders of the LC that were set to determine the LC thickness. (A, C) The borders of the LC are discontinuous either due to laminar pores or poor image contrast (black arrows). (B) The image contrast is poor at the level of the anterior LC border (arrowheads). (B, C) Note the distinct hyper-reflectivity from the retrolaminar interbundle connective tissues (white arrows), which hampers the delineation of the posterior LC borders. (D) The hyper-reflectivity from the retrolaminar interbundle connective tissue is less distinct. The posterior LC border is clearly seen as a continuous curvilinear line. The anterior LC border is also more easily delineable.
The LCT was measured by two observers (EJL, T-WK) who were masked to the clinical information of the subjects. The two observers used the same reference line for each eye to ensure that they measured the LCT at the same location in each individual eye and were masked from the LCT values measured from different images (i.e., MIP and B-scan images). 
Statistical Analysis
The intraobserver and interobserver reproducibility of measuring the LCT-Bscan and LCT-MIP were evaluated by calculating the intraclass correlation coefficient (ICC), coefficient of variation (CV) and coefficient of repeatability (CR) (1.96 × SD of difference between measures). The intraobserver reproducibility for the LCT-Bscan and LCT-MIP was assessed on the two independent measurements made at least 1 week apart by one examiner (EJL) who was masked from the patient clinical information. The interobserver reproducibility was based on the evaluations made by two independent examiners. Statistical comparison of ICC and a sample size calculation were performed based on work described by Walter et al. 23 using power analysis and sample size software for Windows (PASS 11.0; NCSS, LLC, Kaysville, Utah). A sample size calculation estimated that 46 subjects would be necessary for 80% power at a 5% two-sided significance level for the statistical test to detect a mean difference in ICC of 0.10 between 0.80 and 0.90, and 63 subjects for 90% power. 
The LCT-Bscan and the LCT-MIP were compared using the two-tailed paired Student's t-test. The agreement between the LCT-Bscan and the LCT-MIP was assessed by Bland-Altman analysis. 
Statistical analyses were performed using predictive analytics software (PASW Statistics 18.0.0; SPSS Inc., Chicago, IL). A P value of less than 0.05 was considered statistically significant. 
Results
This study originally involved images of 66 patients who were randomly selected from the image database of the Lamina Cribrosa Exploration Study. Of these, three patients were excluded due to poor image quality, leaving a final sample of 63 patients (30 OAG patients and 33 glaucoma suspects). 
The patients were aged 51.5 ± 15.0 years (range, 21–76 years), and 26 were women and 37 were men. Their visual acuity ranged from 20/40 to 20/16 and the refractive error (spherical equivalent) was −1.27 ± 2.25 D (range, −6.00 to +2.50 diopters); 3.2% (2/63) were pseudophakic eyes. The mean untreated IOP was 13.9 ± 5.3 mm Hg (range, 7–41 mm Hg). The visual field mean deviation was −6.59 ± 9.00 dB (range, −29.20 to 1.78 dB). 
The intraobserver and interobserver ICCs, CVs, and CRs of measuring the LCT-Bscan and LCT-MIP are shown in Table 1. Both intraobserver and interobserver ICCs were significantly higher when the LCT was measured using the thin-slab MIP at the mid-horizontal, and superior and inferior mid-peripheral areas. Figure 3 shows Bland-Altman plots for LCT measurements to evaluate the interobserver agreements. The 95% limits of agreement in measuring the LCT were narrower for the LCT-MIP than LCT-Bscan in all the measurement areas: the ranges were −32.15 to 32.64 μm vs. −64.96 to 64.46 μm for superior mid-peripheral LCT, −22.87 to 33.48 μm vs. −51.55 to 43.12 μm for midhorizontal LCT and −33.12 to 33.61 μm vs. −49.82 to 49.81 μm for inferior mid-peripheral LCT, respectively. 
Figure 3. 
 
Bland-Altman plots of the LCT-Bscan versus LCT-MIP by two observers. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement. Note that the 95% limits of agreements between the two observers are narrower for the measurement based on thin-slab MIP images than that based on the B-scan images.
Figure 3. 
 
Bland-Altman plots of the LCT-Bscan versus LCT-MIP by two observers. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement. Note that the 95% limits of agreements between the two observers are narrower for the measurement based on thin-slab MIP images than that based on the B-scan images.
Table 1. 
 
Intraobserver and Interobserver Reproducibility of Measuring the Lamina Cribrosa Thickness Using B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
Table 1. 
 
Intraobserver and Interobserver Reproducibility of Measuring the Lamina Cribrosa Thickness Using B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
Intraobserver Reproducibility Interobserver Reproducibility
LCT-Bscan LCT-MIP LCT-Bscan LCT-MIP
ICC (95% CI) CV, % CR, μm ICC (95% CI) CV, % CR, μm ICC (95% CI) CV, % CR, μm ICC (95% CI) CV, % CR, μm
Superior LCT 0.792 (0.678–0.869) 9.05 53.7 0.923* (0.875–0.952) 5.93 35.1 0.703 (0.552–0.809) 10.63 64.7 0.922* (0.875–0.952) 6.08 35.9
Mid-horizontal LCT 0.831 (0.736–0.895) 7.76 50.8 0.927* (0.883–0.955) 5.04 33.5 0.868 (0.791–0.918) 7.23 47.3 0.952* (0.923–0.971) 4.38 28.2
Inferior LCT 0.895 (0.832–0.935) 6.70 38.4 0.945* (0.911–0.966) 4.55 26.8 0.858 (0.775–0.911) 8.21 49.8 0.924* (0.877–0.953) 5.60 33.4
The Bland-Altman plots to evaluate the agreements between the LCT-Bscan and LCT-MIP are shown in Figure 4. The mean intermethod (B-scan versus MIP) difference in LCT measurements and the 95% limits of agreements were 1.56 ± 15.81 μm and −29.43 to 32.55 μm for the superior midperipheral LCT, 2.89 ± 12.32 μm and −21.26 to 27.04 μm for the mid-horizontal LCT, and −0.13 ± 13.17 μm and −25.94 to 25.68 μm for the inferior midperipheral LCT, respectively. There were no significant differences between the LCT-Bscan and the LCT-MIP for all the midhorizontal, and superior and inferior mid-peripheral LCT (Table 2). 
Figure 4. 
 
Bland-Altman plots of the LCT measurement using B-scan images versus thin-slab maximum intensity projection images. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement.
Figure 4. 
 
Bland-Altman plots of the LCT measurement using B-scan images versus thin-slab maximum intensity projection images. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement.
Table 2. 
 
Comparison of the Lamina Cribrosa Thickness Obtained in B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
Table 2. 
 
Comparison of the Lamina Cribrosa Thickness Obtained in B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
LCT-Bscan, μm LCT-MIP, μm P Value
Superior LCT 212.60 ± 40.22 214.16 ± 44.66 0.437
Mid-horizontal LCT 235.16 ± 42.73 238.05 ± 44.04 0.067
Inferior LCT 211.02 ± 41.58 210.88 ± 40.66 0.936
Discussion
We demonstrated that evaluation of LCT with thin-slab MIP images was more reproducible than with B-scan images. Moreover, delineation of both the anterior and posterior LC borders was more straightforward using the thin-slab MIP images. 
In the present study, the definition of the laminar borders was based on the previous work by Strouthidis et al. 24 in which OCT images were compared with histologic sections in a primate eye. They reported that the first horizontally oriented high reflectivity signal below the disc surface, which intersects the vertical high reflectivity signal from the prelaminar glial columns, 24 corresponds to the anterior laminar surface. Later, Yang et al. 25 defined the points where the horizontally oriented signal ended as the posterior laminar surface. However, there is no histological evidence that the posterior border of the high reflectivity signal is actually where the posterior surface of the LC lies. Thus, it should be noted that the LCT values in the present study were only the width of the high reflectivity signal as seen in SD-OCT images. Further histological comparison is needed to verify that the posterior end of the high reflectivity signal corresponds to the posterior border of the LC. 
MIP is relatively a simple technique that most 3D image-processing software provides and has been widely used for several decades in the fields of radiology and nuclear medicine to detect and locate the lesion of interest three-dimensionally. Previously, we have shown that this technique may be useful to evaluate the LC three-dimensionally. 22 We considered that the visualization of more distinct LC borders in the thin-slab MIP images was largely attributable to the overlapping nature of a high reflectivity signal. As LC is a porous structure, the LC surface is often seen as a discontinuous line in a single B-scan image; there might be gaps at the location of laminar pores or due to pronounced shadowing from overlying retinal vessels. However, when it is viewed with MIP slab-image, the gaps may be filled in by the high reflectivity of the adjacent B-scan images, allowing easier tracing of the LC borders. In addition, the augmentation of the high reflectivity from the inter-bundle connective tissues in the thin-slab MIP images was of relatively less extent than that from the LC, resulting in better image contrast across the posterior border (Fig. 2). This may be attributable to the sparse nature of the interbundle connective tissue. The interbundle connective tissue and the LC consist of similar materials: collagen types I, III, and IV. 26 Thus, they were expected to be seen with similar reflectivity in OCT images. Since the interbundle connective tissue was contiguous with the posterior LC, it may have interfered with the delineation of the posterior LC border. The enhanced image contrast between the LC and the interbundle connective tissue in the thin-slab MIP image should be helpful to delineate the posterior LC border. 
The 95% limit of agreement for the LCT-Bscan versus LCT-MIP was from −29.43 to 32.55 μm. The values should not be interpreted as the measurement error of the LCT-MIP because the LCT-Bscan is not necessarily the true value of LCT. As there is no perfect method to measure the precise LCT, it was impossible to know which measurement was closer to the true value. Regardless of this matter, using the thin-slab MIP to measure the LCT had a definite advantage over using the B-scans, in that regional evaluation of the LCT may be achieved more easily. Instead of measuring the LCT on multiple images and averaging them, investigators may obtain the data from one measurement for the region, which should expedite the evaluation of the LCT. Meanwhile, it is notable that the LCT measured using MIP technique was not a real measurement but an approximate of the LCTs in corresponding B-scans. 
There was a positive relationship between the mean LCT measured by the two observers and the difference between the LCT-Bscan for the midhorizontal (P = 0.003) and inferior measurements (P = 0.004) (data not presented). This means that interobserver difference of the LCT measurement was larger in eyes with thick LC. This may be related to the low image contrast at the level of posterior border in eyes with thick LC, where the posterior border is deeper in the optic nerve. In OCT, images placed far away from the zero delay had worse contrast than those near the zero delay. Such a relationship was not observed for the LCT-MIP, which may be related to better image contrast of MIP image at the posterior border in eyes with thick LC. 
Although LCT-MIP provided better reproducibility for measuring LCT, the 95% limit of agreement for the LCT measurement was still ±30 μm, which may be unsatisfactory for clinical use. We consider the variability in measuring LCT was mostly derived from the vague nature of the posterior LC border. Further development of technologies that provide better delineation of the posterior LC border is warranted. 
In the present study, we included glaucoma suspects and glaucoma patients. Based on the theoretical concern that thinner LCT may be related to glaucoma development and progression, glaucoma suspects as well as glaucoma patients were candidates for LCT measurement. Thus, we included both groups in our study. Although we did not perform a statistical comparison for the reproducibility of measuring the LCT between glaucoma suspect and glaucoma patients, it appeared comparable based on the Bland-Altman plots. 
This study had limitations. First, an LCT-Bscan for each region was obtained by averaging the three measurements obtained from the three B-scan images, while LCT-MIP was a single measurement from the corresponding region. This may have generated a bias toward a lower SD for LCT-Bscan because averaging three measurements for a given data set would lower the SD to 1/√3 fold than the single measurement for each data. This may be related to the lower SD for the LCT-Bscan in the superior LCT and mid-horizontal LCT (Table 2). In addition, the SD between two repeated measurements would be lower when the measurement for each occasion was obtained by averaging multiple measurements than obtained by a single measurement in each occasion. Thus, one may expect lower CV for LCT-Bscan than LCT-MIP. Nonetheless, the CV was lower for LCT-MIP than for LCT-Bscan (Table 2). This suggested that the improved reproducibility was not attributable to the statistical influence from the difference in obtained data but from the better image contrast, which allowed more reproducible delineation of the LC borders in MIP images. 
Second, the LC borders were measured only at one point in each B-scan or MIP image. The primary purpose of the present study was to investigate whether MIP enabled better delineation of the LC borders measurement, thereby providing better repeatability in measuring the LCT. For this purpose, measurement at a single point may have provided a straightforward analysis. However, the LC had a large regional variation in thickness. Thus, measurement at multiple locations was needed in research that aims to elucidate the role of LCT in glaucoma pathophysiology. Meanwhile, measuring the data at multiple locations and averaging may increase the repeatability more than one-point measurement. 
In conclusion, the thin volume section images generated by MIP provided a higher reproducibility than the B-scan images in measuring the SD-OCT–based LCT. Using the thin-slab MIP, the detection of both the anterior and posterior LC borders was more straightforward. This technique should facilitate the evaluation of the LC. 
Acknowledgments
The authors thank Hyunjoong Kim of the Department of Applied Statistics, Yonsei University, Seoul, Korea, for statistical advice. 
References
Burgoyne CF Downs JC Bellezza AJ Suh JK Hart RT. The optic nerve head as a biomechanical structure: a new paradigm for understanding the role of IOP-related stress and strain in the pathophysiology of glaucomatous optic nerve head damage. Prog Retin Eye Res . 2005;24:39–73. [CrossRef] [PubMed]
Bellezza AJ Hart RT Burgoyne CF. The optic nerve head as a biomechanical structure: initial finite element modeling. Invest Ophthalmol Vis Sci . 2000;41:2991–3000. [PubMed]
Sigal IA Ethier CR. Biomechanics of the optic nerve head. Exp Eye Res . 2009;88:799–807. [CrossRef] [PubMed]
Sigal IA Flanagan JG Tertinegg I Ethier CR. Modeling individual-specific human optic nerve head biomechanics. Part I: IOP-induced deformations and influence of geometry. Biomech Model Mechanobiol . 2009;8:85–98. [CrossRef] [PubMed]
Sigal IA Flanagan JG Tertinegg I Ethier CR. Modeling individual-specific human optic nerve head biomechanics. Part II: influence of material properties. Biomech Model Mechanobiol . 2009;8:99–109. [CrossRef] [PubMed]
Anderson DR Hendrickson A. Effect of intraocular pressure on rapid axoplasmic transport in monkey optic nerve. Invest Ophthalmol . 1974;13:771–783. [PubMed]
Minckler DS Tso MO Zimmerman LE. A light microscopic, autoradiographic study of axoplasmic transport in the optic nerve head during ocular hypotony, increased intraocular pressure, and papilledema. Am J Ophthalmol . 1976;82:741–757. [CrossRef] [PubMed]
Levy NS. The effects of elevated intraocular pressure on slow axonal protein flow. Invest Ophthalmol . 1974;13:691–695. [PubMed]
Quigley H Anderson DR. The dynamics and location of axonal transport blockade by acute intraocular pressure elevation in primate optic nerve. Invest Ophthalmol . 1976;15:606–616. [PubMed]
Ren R Wang N Zhang X Cui T Jonas JB. Trans-lamina cribrosa pressure difference correlated with neuroretinal rim area in glaucoma. Graefes Arch Clin Exp Ophthalmol . 2011;249:1057–1063. [CrossRef] [PubMed]
Ren R Jonas JB Tian G Cerebrospinal fluid pressure in glaucoma: a prospective study. Ophthalmology . 2010;117:259–266. [CrossRef] [PubMed]
Jonas JB Berenshtein E Holbach L. Anatomic relationship between lamina cribrosa, intraocular space, and cerebrospinal fluid space. Invest Ophthalmol Vis Sci . 2003;44:5189–5195. [CrossRef] [PubMed]
Jonas JB Berenshtein E Holbach L. Lamina cribrosa thickness and spatial relationships between intraocular space and cerebrospinal fluid space in highly myopic eyes. Invest Ophthalmol Vis Sci . 2004;45:2660–2665. [CrossRef] [PubMed]
Ren R Wang N Li B Lamina cribrosa and peripapillary sclera histomorphometry in normal and advanced glaucomatous Chinese eyes with various axial length. Invest Ophthalmol Vis Sci . 2009;50:2175–2184. [CrossRef] [PubMed]
Ren R Li B Gao F Central corneal thickness, lamina cribrosa and peripapillary scleral histomorphometry in non-glaucomatous Chinese eyes. Graefes Arch Clin Exp Ophthalmol . 2010;248:1579–1585. [CrossRef] [PubMed]
Lee EJ Kim TW Weinreb RN Park KH Kim SH Kim DM. Visualization of the lamina cribrosa using enhanced depth imaging spectral-domain optical coherence tomography. Am J Ophthalmol . 2011;152:87–95. e81. [CrossRef] [PubMed]
Park HY Jeon SH Park CK. Enhanced depth imaging detects lamina cribrosa thickness differences in normal tension glaucoma and primary open-angle glaucoma. Ophthalmology . 2012;119:10–20. [CrossRef] [PubMed]
Inoue R Hangai M Kotera Y Three-dimensional high-speed optical coherence tomography imaging of lamina cribrosa in glaucoma. Ophthalmology . 2009;116:214–222. [CrossRef] [PubMed]
Yang H Downs JC Girkin C 3-D histomorphometry of the normal and early glaucomatous monkey optic nerve head: lamina cribrosa and peripapillary scleral position and thickness. Invest Ophthalmol Vis Sci . 2007;48:4597–4607. [CrossRef] [PubMed]
Roberts MD Grau V Grimm J Remodeling of the connective tissue microarchitecture of the lamina cribrosa in early experimental glaucoma. Invest Ophthalmol Vis Sci . 2009;50:681–690. [CrossRef] [PubMed]
Girard MJ Strouthidis NG Ethier CR Mari JM. Shadow removal and contrast enhancement in optical coherence tomography images of the human optic nerve head. Invest Ophthalmol Vis Sci . 2011;52:7738–7748. [CrossRef] [PubMed]
Lee EJ Kim TW Weinreb RN Three-dimensional evaluation of the lamina cribrosa using spectral-domain optical coherence tomography in glaucoma. Invest Ophthalmol Vis Sci . 2012;53:198–204. [CrossRef] [PubMed]
Walter SD Eliasziw M Donner A. Sample size and optimal designs for reliability studies. Stat Med . 1998;17:101–110. [CrossRef] [PubMed]
Strouthidis NG Grimm J Williams GA Cull GA Wilson DJ Burgoyne CF. A comparison of optic nerve head morphology viewed by spectral domain optical coherence tomography and by serial histology. Invest Ophthalmol Vis Sci . 2010;51:1464–1474. [CrossRef] [PubMed]
Yang H Qi J Hardin C Spectral-domain optical coherence tomography enhanced depth imaging of the normal and glaucomatous nonhuman primate optic nerve head. Invest Ophthalmol Vis Sci . 2012;53:394–405. [CrossRef] [PubMed]
Rehnberg M Ammitzboll T Tengroth B. Collagen distribution in the lamina cribrosa and the trabecular meshwork of the human eye. Br J Ophthalmol . 1987;71:886–892. [CrossRef] [PubMed]
Footnotes
 Supported by National Research Foundation of Korea, a grant funded by the Korean Government (2010-0004210) and, in part, by an unrestricted grant from Research to Prevent Blindness (RNW) (New York, NY). The authors alone are responsible for the content and writing of the paper.
Footnotes
 Disclosure: E.J. Lee, None; T.-W. Kim, None; R.N. Weinreb, Heidelberg Engineering (F)
Figure 1. 
 
(A) The MIP volumetric image of an ONH reconstructed by maximum intensity projection. The light-blue line indicates the location where the thin volumetric section (B) was selected. (C, E) The B-scan image at the center of the thin volume section. (D, F) XY view of the thin-slab MIP image. A more clear demarcation of the anterior border (arrows) as well as posterior border (arrowheads) of the lamina cribrosa is noted in the thin-slab MIP image. (E, F) The vertical orange line indicates the reference line, which was set near the center of the Bruch's membrane opening. The anterior and posterior borders were defined as the plane that best represented the margin of the hyper-reflective plate within the 200 μm width area around the reference line (horizontal orange lines).
Figure 1. 
 
(A) The MIP volumetric image of an ONH reconstructed by maximum intensity projection. The light-blue line indicates the location where the thin volumetric section (B) was selected. (C, E) The B-scan image at the center of the thin volume section. (D, F) XY view of the thin-slab MIP image. A more clear demarcation of the anterior border (arrows) as well as posterior border (arrowheads) of the lamina cribrosa is noted in the thin-slab MIP image. (E, F) The vertical orange line indicates the reference line, which was set near the center of the Bruch's membrane opening. The anterior and posterior borders were defined as the plane that best represented the margin of the hyper-reflective plate within the 200 μm width area around the reference line (horizontal orange lines).
Figure 2. 
 
Three serial B-scan images (AC) and a thin-slab MIP image (D) of a glaucomatous eye. (A, inset) Black dots indicate the apparent border of the LC at the reference line. Orange dotted lines are imaginary lines that represent the surface of the LC within the 200 μm width area around the reference line (A, vertical orange line). (BD) Orange lines in the inset images indicate the anterior and posterior borders of the LC that were set to determine the LC thickness. (A, C) The borders of the LC are discontinuous either due to laminar pores or poor image contrast (black arrows). (B) The image contrast is poor at the level of the anterior LC border (arrowheads). (B, C) Note the distinct hyper-reflectivity from the retrolaminar interbundle connective tissues (white arrows), which hampers the delineation of the posterior LC borders. (D) The hyper-reflectivity from the retrolaminar interbundle connective tissue is less distinct. The posterior LC border is clearly seen as a continuous curvilinear line. The anterior LC border is also more easily delineable.
Figure 2. 
 
Three serial B-scan images (AC) and a thin-slab MIP image (D) of a glaucomatous eye. (A, inset) Black dots indicate the apparent border of the LC at the reference line. Orange dotted lines are imaginary lines that represent the surface of the LC within the 200 μm width area around the reference line (A, vertical orange line). (BD) Orange lines in the inset images indicate the anterior and posterior borders of the LC that were set to determine the LC thickness. (A, C) The borders of the LC are discontinuous either due to laminar pores or poor image contrast (black arrows). (B) The image contrast is poor at the level of the anterior LC border (arrowheads). (B, C) Note the distinct hyper-reflectivity from the retrolaminar interbundle connective tissues (white arrows), which hampers the delineation of the posterior LC borders. (D) The hyper-reflectivity from the retrolaminar interbundle connective tissue is less distinct. The posterior LC border is clearly seen as a continuous curvilinear line. The anterior LC border is also more easily delineable.
Figure 3. 
 
Bland-Altman plots of the LCT-Bscan versus LCT-MIP by two observers. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement. Note that the 95% limits of agreements between the two observers are narrower for the measurement based on thin-slab MIP images than that based on the B-scan images.
Figure 3. 
 
Bland-Altman plots of the LCT-Bscan versus LCT-MIP by two observers. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement. Note that the 95% limits of agreements between the two observers are narrower for the measurement based on thin-slab MIP images than that based on the B-scan images.
Figure 4. 
 
Bland-Altman plots of the LCT measurement using B-scan images versus thin-slab maximum intensity projection images. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement.
Figure 4. 
 
Bland-Altman plots of the LCT measurement using B-scan images versus thin-slab maximum intensity projection images. The plane line represents the mean difference and the two dotted lines represent the 95% limits of agreement.
Table 1. 
 
Intraobserver and Interobserver Reproducibility of Measuring the Lamina Cribrosa Thickness Using B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
Table 1. 
 
Intraobserver and Interobserver Reproducibility of Measuring the Lamina Cribrosa Thickness Using B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
Intraobserver Reproducibility Interobserver Reproducibility
LCT-Bscan LCT-MIP LCT-Bscan LCT-MIP
ICC (95% CI) CV, % CR, μm ICC (95% CI) CV, % CR, μm ICC (95% CI) CV, % CR, μm ICC (95% CI) CV, % CR, μm
Superior LCT 0.792 (0.678–0.869) 9.05 53.7 0.923* (0.875–0.952) 5.93 35.1 0.703 (0.552–0.809) 10.63 64.7 0.922* (0.875–0.952) 6.08 35.9
Mid-horizontal LCT 0.831 (0.736–0.895) 7.76 50.8 0.927* (0.883–0.955) 5.04 33.5 0.868 (0.791–0.918) 7.23 47.3 0.952* (0.923–0.971) 4.38 28.2
Inferior LCT 0.895 (0.832–0.935) 6.70 38.4 0.945* (0.911–0.966) 4.55 26.8 0.858 (0.775–0.911) 8.21 49.8 0.924* (0.877–0.953) 5.60 33.4
Table 2. 
 
Comparison of the Lamina Cribrosa Thickness Obtained in B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
Table 2. 
 
Comparison of the Lamina Cribrosa Thickness Obtained in B-Scan Images (LCT-Bscan) and Thin-Slab Maximum Intensity Projection Images (LCT-MIP)
LCT-Bscan, μm LCT-MIP, μm P Value
Superior LCT 212.60 ± 40.22 214.16 ± 44.66 0.437
Mid-horizontal LCT 235.16 ± 42.73 238.05 ± 44.04 0.067
Inferior LCT 211.02 ± 41.58 210.88 ± 40.66 0.936
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×