Investigative Ophthalmology & Visual Science Cover Image for Volume 53, Issue 1
January 2012
Volume 53, Issue 1
Free
Glaucoma  |   January 2012
2-D Pattern of Nerve Fiber Bundles in Glaucoma Emerging from Spectral-Domain Optical Coherence Tomography
Author Affiliations & Notes
  • Mona K. Garvin
    From the Center for the Prevention and Treatment of Visual Loss, Iowa City VA Health Care System, Iowa City, Iowa; and
    the Departments of Electrical and Computer Engineering,
  • Michael D. Abràmoff
    From the Center for the Prevention and Treatment of Visual Loss, Iowa City VA Health Care System, Iowa City, Iowa; and
    the Departments of Electrical and Computer Engineering,
    Ophthalmology and Visual Sciences, and
    Biomedical Engineering, The University of Iowa, Iowa City, Iowa.
  • Kyungmoo Lee
    the Departments of Electrical and Computer Engineering,
  • Meindert Niemeijer
    the Departments of Electrical and Computer Engineering,
  • Milan Sonka
    the Departments of Electrical and Computer Engineering,
    Ophthalmology and Visual Sciences, and
  • Young H. Kwon
    Ophthalmology and Visual Sciences, and
  • Corresponding author: Mona K. Garvin, 4318 Seamans Center for the Engineering Arts and Sciences, The University of Iowa, Iowa City, IA 52242; [email protected]
Investigative Ophthalmology & Visual Science January 2012, Vol.53, 483-489. doi:https://doi.org/10.1167/iovs.11-8349
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Mona K. Garvin, Michael D. Abràmoff, Kyungmoo Lee, Meindert Niemeijer, Milan Sonka, Young H. Kwon; 2-D Pattern of Nerve Fiber Bundles in Glaucoma Emerging from Spectral-Domain Optical Coherence Tomography. Invest. Ophthalmol. Vis. Sci. 2012;53(1):483-489. https://doi.org/10.1167/iovs.11-8349.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose.: To correlate the thicknesses of focal regions of the macular ganglion cell layer with those of the peripapillary nerve fiber layer using spectral-domain optical coherence tomography (SD-OCT) in glaucoma subjects.

Methods.: Macula and optic nerve head SD-OCT volumes were obtained in 57 eyes of 57 subjects with open-angle glaucoma or glaucoma suspicion. Using a custom automated computer algorithm, the thickness of 66 macular ganglion cell layer regions and the thickness of 12 peripapillary nerve fiber layer regions were measured from registered SD-OCT volumes. The mean thickness of each ganglion cell layer region was correlated to the mean thickness of each peripapillary nerve fiber layer region across subjects. Each ganglion cell layer region was labeled with the peripapillary nerve fiber layer region with the highest correlation using a color-coded map.

Results.: The resulting color-coded correlation map closely resembled the nerve fiber bundle (NFB) pattern of retinal ganglion cells. The mean r2 value across all local macular-peripapillary correlations was 0.49 (± 0.11). When separately analyzing the 30 glaucoma subjects from the 27 glaucoma-suspect subjects, the mean r2 value across all local macular-peripapillary correlations was significantly larger in the glaucoma group (0.56 ± 0.13 vs. 0.37 ± 0.11; P < 0.001).

Conclusions.: A two-dimensional (2-D) spatial NFB map of the retina can be developed using structure-structure relationships from SD-OCT. Such SD-OCT-based NFB maps may enhance glaucoma detection and contribute to monitoring change in the future.

In glaucoma, the retinal ganglion cell (RGC) axons that make up the nerve fiber bundles (NFB) in the retina and optic nerve degenerate. Eventually, the RGC soma also degenerates as the cell undergoes apoptotic cell death. 1 An understanding of the retinal trajectory pattern of the RGC NFB is important for diagnosis and monitoring glaucomatous damage. Several attempts have been made to describe NFB trajectories by visually examining histology of stained retina 2 4 and by examining patterns of visual field defects. 5,6 More recently, Jansonius et al. 7 modeled the trajectories mathematically using manual tracings of NFB in fundus photographs of human glaucoma subjects. 
The recent availability of spectral-domain optical coherence tomography (SD-OCT) allows unprecedented 3-D spatial resolution of the retinal nerve fiber layer (RNFL) and other retinal layers including the ganglion cell layer (GCL). 8 11 The peripapillary RNFL thinning in glaucoma has been well documented using optical coherence tomography (OCT). 12 14 GCL thinning in the macula of glaucoma patients has also been documented using OCT. 15,16 Both of these findings are not surprising given the anatomic pathway of the RGC in the retina and optic disc. 
The purpose of the present study is to correlate the thickness of focal regions of the macular GCL with the thickness of focal regions of the peripapillary RNFL in patients with glaucoma (or suspicion of glaucoma). The results show that mapping the peripapillary RNFL region of highest correlation for each local macular region results in a pattern resembling NFB trajectories. 
Methods
Data Acquisition
Patient data were obtained prospectively in both eyes. The study subjects were recruited consecutively from the outpatient Glaucoma Service at the University of Iowa. Open-angle glaucoma and glaucoma suspects were included in the study; angle-closure and combined-mechanism glaucoma were excluded. Glaucoma was defined as optic disc cupping consistent with glaucoma (either diffuse or focal thinning of the neuroretinal rim or nerve fiber layer defects) and visual field defects consistent with optic disc cupping, with or without elevated intraocular pressure. Primary as well as secondary open-angle glaucoma (e.g., pigmentary or exfoliative) was included. Glaucoma suspect was defined as ocular hypertension (>21 mm Hg) without evidence of glaucomatous optic neuropathy or suspicious optic discs (vertical cup-to-disc ratio >0.7 or >0.2 asymmetry between fellow eyes) with normal visual fields. 17 Macula- and optic nerve head-centered SD-OCT volumes (Cirrus, Carl Zeiss Meditec, Inc., Dublin, CA) were acquired. Each SD-OCT volume was 200 × 200 × 1024 voxels, corresponding to physical dimensions of 6 × 6 × 2 mm3. Humphrey visual fields (Carl Zeiss Meditec, Inc.) and simultaneous stereo fundus photographs (3Dx, Nidek Inc., Freemont, CA) were obtained on the same day as OCT imaging. In the present study, only subjects with acceptable quality of both macular and peripapillary OCT images (with signal strength ≥ 5) were included. One eye of each subject was randomly chosen for analysis. The study was approved by the Institutional Review Board of the University of Iowa, adhered to the tenets of the Declaration of Helsinki, and all subjects gave written informed consent. 
Automated Local Thickness Measurements
Within each of the macular and peripapillary volumes, 11 surfaces were segmented in three dimensions using our previously published graph-theoretic approach. 18,19 Five of these surfaces, as labeled in Figure 1, were used to define the retinal nerve fiber layer (between surfaces 1 and 2), the retinal ganglion cell layer (between surfaces 2 and 3), and the photoreceptor/retinal-pigment-epithelium complex (between surfaces 4 and 5). In the peripapillary volume, the neural canal opening (NCO) was also automatically segmented 20 and the layers were considered undefined within the neural canal region (Fig. 1E). 
Figure 1.
 
The segmentation of structures in each macular and peripapillary SD-OCT volume. (A) Example slice from a macular SD-OCT volume. (B) Five labeled surfaces (out of the 11 segmented) illustrated on the macular slice. These five surfaces enabled computation of the retinal nerve fiber layer thickness (between surface 1 and surface 2), the ganglion cell layer thickness (between surface 2 and surface 3), and a projection image of photoreceptor and retinal pigment epithelium complex (between surface 4 and surface 5). (C) Volumetric visualization of five surfaces from the macular volume. (D) Example slice from a peripapillary SD-OCT volume. (E) Five surfaces illustrated on the peripapillary slice. The red column shows the area of neural canal opening in which the layers were not defined. (F) Volumetric visualization of five surfaces from the peripapillary volume.
Figure 1.
 
The segmentation of structures in each macular and peripapillary SD-OCT volume. (A) Example slice from a macular SD-OCT volume. (B) Five labeled surfaces (out of the 11 segmented) illustrated on the macular slice. These five surfaces enabled computation of the retinal nerve fiber layer thickness (between surface 1 and surface 2), the ganglion cell layer thickness (between surface 2 and surface 3), and a projection image of photoreceptor and retinal pigment epithelium complex (between surface 4 and surface 5). (C) Volumetric visualization of five surfaces from the macular volume. (D) Example slice from a peripapillary SD-OCT volume. (E) Five surfaces illustrated on the peripapillary slice. The red column shows the area of neural canal opening in which the layers were not defined. (F) Volumetric visualization of five surfaces from the peripapillary volume.
Using projection images created by computing the mean intensity of voxels within the photoreceptor/retinal-pigment-epithelium complex, the volumetric scans were then automatically registered (i.e., aligned) by first segmenting the vessels using our previously published algorithm 21 and then using a standard registration method (with a similarity transformation and cross-correlation metric) from open-source software (Insight Segmentation and Registration Toolkit 22 ) to align the vessel segmentation maps. An example registration result is presented in Figure 2
Figure 2.
 
Registration of macular and peripapillary SD-OCT volumes. (A) Example projected image from a macular SD-OCT volume (obtained by computing the projected mean intensity of voxels in the segmented photoreceptor/retinal-pigment-epithelium complex). (B) Example projected image from a peripapillary SD-OCT volume. (C) Illustration of automated registration result.
Figure 2.
 
Registration of macular and peripapillary SD-OCT volumes. (A) Example projected image from a macular SD-OCT volume (obtained by computing the projected mean intensity of voxels in the segmented photoreceptor/retinal-pigment-epithelium complex). (B) Example projected image from a peripapillary SD-OCT volume. (C) Illustration of automated registration result.
After registration, local regions in the macular and peripapillary volumes were automatically defined in each registered pair as follows (Fig. 3). First, the center of the macula was defined as the lowest point of the foveal pit. Next, the center of the optic nerve head was defined from the centroid of the segmented neural canal opening. These two points thus defined a reference line that helped in the placement of the local macular and peripapillary grids. Next, a macular grid consisting of 66 square regions (2° by 2° in size for each square) was rotated so its midline was coincident with this reference line and its center was at the calculated foveal pit. The mean thickness of the RGC layer in each of the 66 regions was computed. Peripapillary regions were defined by evenly dividing the temporal peripapillary retina into 12 wedge-shaped regions (15° each). The inner radial boundary of each wedge was defined by the neural canal opening, and the outer radial boundary was defined by the midpoint of the reference line (which also defined the nasal extent of the macular squares), as presented in Figure 3. We did not analyze the nasal peripapillary retina because we expected only a few macular NFB would reach the disc nasally. The mean RNFL thickness for each of the 12 wedges was computed. Thus, each eye generated 66 grid RGC layer thickness measurements and 12 wedge peripapillary RNFL thickness measurements. 
Figure 3.
 
Local macular and peripapillary grids for each registered macula and peripapillary SD-OCT volumetric pair, on underlying projection image (A) and without projection image (B). Based on the center of the foveal pit and the center of the neural canal opening (each indicated with a plus), a reference line can be drawn between them and bisected. The macular grid of 66 squares (each 2° by 2°) is then placed on the macular volume so as to align the center with the center of the foveal pit and to rotate the grid so that the horizontal grid lines are parallel to the reference line. The square grid ends nasally at the midpoint of the reference line. The 12 peripapillary wedge-shaped regions are segmented in 15° increments. The neural canal opening defines the inner radial boundary and the midpoint of the reference line defines the outer radial boundary.
Figure 3.
 
Local macular and peripapillary grids for each registered macula and peripapillary SD-OCT volumetric pair, on underlying projection image (A) and without projection image (B). Based on the center of the foveal pit and the center of the neural canal opening (each indicated with a plus), a reference line can be drawn between them and bisected. The macular grid of 66 squares (each 2° by 2°) is then placed on the macular volume so as to align the center with the center of the foveal pit and to rotate the grid so that the horizontal grid lines are parallel to the reference line. The square grid ends nasally at the midpoint of the reference line. The 12 peripapillary wedge-shaped regions are segmented in 15° increments. The neural canal opening defines the inner radial boundary and the midpoint of the reference line defines the outer radial boundary.
All maps were represented as right eye configuration (i.e., left eye maps were flipped to right eye configuration). Results are shown as mean (± SD), unless otherwise noted. 
Results
We analyzed the OCT data from 57 eyes of 57 subjects (30 glaucoma subjects and 27 glaucoma-suspect subjects). The mean age was 63.1 (±14.9) years. There were 38 females. The study cohort included 53 Caucasian, 3 African-American, and 1 Hispanic patient. The mean RNFL thickness from the analysis provided by the manufacturer (Cirrus version 5.1; Carl Zeiss Meditec, Inc., Dublin, CA) was 77.8 (±14.1) microns. The mean rim area and average cup-to-disc ratio, as provided by the software (Cirrus), were 0.99 (±0.31) mm2 and 0.64 (±0.14), respectively. The Humphrey 24-2 mean deviation (MD) was −1.74 (±3.32) dB and mean pattern SD (PSD) was 3.19 (±3.22) dB. 
Linear correlation between the macular RGC layer thickness and peripapillary RNFL thickness were examined across individual eyes. In particular, for each of the 66 local macular regions, a squared Pearson's correlation coefficient (r2 ) was computed for each of the 12 local peripapillary regions. The peripapillary wedge representing the highest correlation was identified for display. For the correlation map, each of the 12 peripapillary wedge regions was assigned a separate color and each of the 66 macular regions was assigned the color of the peripapillary wedge representing the highest correlation. The resulting color-coded structure-structure correlation map and corresponding r 2 values for each macular region are presented in Figures 4A and 4B, respectively. The mean r2 value across all local macular-peripapillary relationships was 0.49 (±0.11). The corresponding scatterplots, best-fit lines, and r 2 values for all macular regions are presented in Figure 5. In addition, Figure 6 contains an enhanced color-coded map that illustrates, for each macular region, the set of peripapillary wedges with a Pearson's correlation coefficient not significantly different from that of the peripapillary wedge with the highest correlation. 
Figure 4.
 
Correlation map between macular RGC layer and peripapillary RNFL (A) and corresponding r 2 values (B), using data from 57 eyes of 57 subjects. The color of each macular region in the correlation map in (A) corresponds to the color of peripapillary wedge representing highest correlation.
Figure 4.
 
Correlation map between macular RGC layer and peripapillary RNFL (A) and corresponding r 2 values (B), using data from 57 eyes of 57 subjects. The color of each macular region in the correlation map in (A) corresponds to the color of peripapillary wedge representing highest correlation.
Figure 5.
 
The scatterplots of all 66 macular regions. For each macular region, the best correlation between the macular region (GCL thickness) and the corresponding peripapillary wedge (RNFL thickness) is shown. Each square box contains the best scatterplot, r 2 value, and corresponding wedge letter.
Figure 5.
 
The scatterplots of all 66 macular regions. For each macular region, the best correlation between the macular region (GCL thickness) and the corresponding peripapillary wedge (RNFL thickness) is shown. Each square box contains the best scatterplot, r 2 value, and corresponding wedge letter.
Figure 6.
 
Color-coded correlation map (from all 57 subjects) with a peripapillary wedges overlay for each macular region. The background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white.
Figure 6.
 
Color-coded correlation map (from all 57 subjects) with a peripapillary wedges overlay for each macular region. The background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white.
The same correlation analysis was repeated for the 30 subjects with glaucoma (i.e., excluding the glaucoma-suspect subjects) and then separately repeated for the 27 subjects with glaucoma suspicion. In the glaucoma group (30 subjects), the mean r 2 value across all local macular-peripapillary correlations was 0.56 ± 0.13, which was significantly larger than the mean r 2 value in the glaucoma-suspect group (0.37 ± 0.11; P < 0.001; paired t-test). The corresponding color-coded correlation maps (along with illustration of peripapillary wedge sets with correlations not significantly different from that of the best wedge) for both cases are presented in Figure 7. Note that in the glaucoma case (Fig. 7A), for most macular regions, a smaller number of peripapillary wedges have a Pearson's correlation coefficient not significantly different from that of the best wedge when compared with the number present in the glaucoma-suspect case (Fig. 7B) and the overall pattern more closely resembles expected nerve fiber bundle trajectories. 
Figure 7.
 
Comparison of resulting color-coded correlation map (with overlays) for 30 subjects with glaucoma (A) and 27 subjects with glaucoma suspicion (B). As in Figure 6 (which shows the correlation map with overlays for all subjects), the background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white. Note that the glaucoma case displays a pattern that more closely resembles expected nerve-fiber-bundle trajectories. In addition, for almost all macular regions, the number of wedges with a correlation not significantly different from that of the best wedge is smaller in the glaucoma case, further indicating a tighter relationship in this case.
Figure 7.
 
Comparison of resulting color-coded correlation map (with overlays) for 30 subjects with glaucoma (A) and 27 subjects with glaucoma suspicion (B). As in Figure 6 (which shows the correlation map with overlays for all subjects), the background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white. Note that the glaucoma case displays a pattern that more closely resembles expected nerve-fiber-bundle trajectories. In addition, for almost all macular regions, the number of wedges with a correlation not significantly different from that of the best wedge is smaller in the glaucoma case, further indicating a tighter relationship in this case.
In addition, the analysis was repeated using both eyes of all subjects, with similar results (mean r2 value of 0.49 ± 0.11, data not shown). We also performed a similar correlation analysis between the macular nerve fiber layer (NFL) thickness and the peripapillary NFL thickness and found the correlations to be worse (mean r2 value of 0.40 ± 0.21, data not shown) than those from the macular GCL versus PP NFL thickness analysis. 
Discussion
In the present study, we studied correlation of regional macular GCL thicknesses with those of peripapillary RNFL thicknesses in glaucoma and glaucoma suspect patients, based on the known anatomy of the RGC soma/axons and NFB trajectories. A spatial pattern consistent with NFB trajectories emerged from the structure-structure correlations. For illustrative purposes, the resulting color-coded correlation map (from all subjects) was overlaid on hand-traced nerve fiber bundles (taken from Fitzgibbon and Taylor, 2 Fig. 8), with the map closely resembling the two-dimensional (2-D) pattern of the NFB in the macula and the temporal peripapillary retina. Furthermore, our results demonstrate that the correlation map from glaucoma subjects better reflects expected NFB trajectories than that from glaucoma-suspect subjects (as illustrated in Fig. 7). 
Figure 8.
 
Qualitative comparison of our structure-structure-derived correlation map and hand-drawn nerve fiber bundles from Fitzgibbon and Taylor. 2 (A) Illustration of nerve fiber bundles taken from Fitzgibbon and Taylor. 2 (B) Enlarged portion of retinal nerve fiber bundles overlaid on our color correlation map (from Fig. 4A).
Figure 8.
 
Qualitative comparison of our structure-structure-derived correlation map and hand-drawn nerve fiber bundles from Fitzgibbon and Taylor. 2 (A) Illustration of nerve fiber bundles taken from Fitzgibbon and Taylor. 2 (B) Enlarged portion of retinal nerve fiber bundles overlaid on our color correlation map (from Fig. 4A).
To our knowledge, this is the first time a 2-D NFB “map” was generated purely using OCT layer thicknesses. We are able to show visually in 2-D that there are meaningful structural correlations between the regions of macular RGC layer thickness with corresponding regions of peripapillary RNFL thickness, and that the emerging spatial pattern resembles underlying NFB trajectories. It is important to note that this 2-D spatial pattern was observed without assumption for a particular model of NFB anatomy. This is in contrast to prior studies that required manual tracings of NFB from fundus photographs and assumed an underlying spatial relationship. 7 While the emerging pattern is still crude, we feel this type of structural correlation can pave the way for a more refined 2-D spatial pattern of NFB in the future. 
It is also important to note that using glaucoma subjects with focal (as opposed to diffuse loss seen in advanced glaucoma) NFB damage in our study may have been critical in enabling an “NFB map” to be obtained across subjects. In other words, we hypothesize that having relatively early, focal damage (thinning) in a particular peripapillary RNFL corresponding to a particular area of thinning in the macular RGC layer would be necessary in developing this NFB map. Indeed, the Humphrey visual field results suggest a relatively mild glaucomatous damage with a mean deviation of −1.74 (± 3.32) dB and pattern SD of 3.19 (± 3.22) dB across our subjects. If this hypothesis were true, we would expect that a resulting NFB map would be less apparent when examined across normal (nonglaucomatous) or very advanced glaucomatous subjects. In fact, our separate analysis of the 30 glaucoma subjects and 27 glaucoma-suspect subjects supports this hypothesis as the resulting correlation map (Fig. 7A) from the glaucoma subjects more closely resembles an expected nerve-fiber-bundle pattern than that from the glaucoma-suspect subjects (Fig. 7B). A more careful examination of this hypothesis will be a subject of future studies. 
NFB maps may provide enhancement for determining structural damage in glaucoma subjects using SD-OCT, over currently available structural measures such as peripapillary RNFL thickness and optic nerve head parameters (e.g., rim area). 23 A 2-D NFB map of the retina is ideal for assessing glaucoma because the NFB (at the optic nerve head) is believed to be the earliest site of glaucomatous damage. It is not difficult to imagine a 2-D NFB map assessment would correlate well with disease severity as well as functional assessment through perimetry. Such a 2-D NFB map assessment would complement the previous work that has related local structural damage in the GCL and/or peripapillary RNFL to functional measurements. 24 30  
Furthermore, while the use of “average” NFB maps may indeed enable an enhanced assessment of glaucoma, it is perhaps more important to note that the current work may provide a foundation for developing individualized NFB maps from SD-OCT. For example, based on the results of this work which demonstrates the ability to generate such NFB maps using only a minimal set of assumptions, we are currently exploring the use of constrained graph-theoretic approaches to enhance our ability to model and predict nerve fiber bundles across subjects and in individual subjects. Overall, the introduction of higher resolution SD-OCT volumetric data (such as that resulting from adaptive-optics-OCT imaging 31 ) should greatly enhance our ability to detect individual NFB maps. Such individualized maps will allow more precise assessment of the glaucomatous RGC damage as well as more precise tracking of the progression of disease. 
There are several limitations to the present study. The size of each macular region was chosen to match the size of Humphrey 10-2 grid points (however, note that the position of the grid points was not exactly the same because the grid was rotated as presented in Fig. 3A). Similarly, the 12 PP wedges were chosen arbitrarily. Future work is necessary to examine other potential region shapes and sizes. We ignored the nasal peripapillary RNFL in this study. Obviously, any full assessment of the glaucoma damage must take nasal NFB into account as well. Each OCT volume was limited to a projected 6 × 6 mm2 area and thus an automated image registration computer algorithm was required to align them. For the NFB map to be more clinically useful, it is desirable to image a larger area that includes both the macula and the optic nerve head simultaneously, rather than requiring the registration step (which, like any software algorithm, is subject to potential inaccuracies). All volumetric scans were acquired using one commercially available imaging device with consistent volumetric scan sizes. Future work is necessary to ensure that all of our custom image-analysis algorithms can more generally work across all commercially-available OCT devices (and different volumetric scan sizes). 
In summary, we correlated focal regions of macular RGC layer with focal regions of peripapillary RNFL thickness using SD-OCT in glaucoma and glaucoma suspect patients. The resulting correlation map resembles NFB trajectories of the RGC in the retina, and may provide the basis for more refined 2-D NFB maps of the retina. Such maps may further enable a more precise structural assessment of glaucomatous damage. 
Footnotes
 Supported by the National Institutes of Health Grants R01 EY018853 (MS, MDA), R01 EY019112 (MDA, MS), and R01 EB004640 (MS); the Department of Veterans Affairs; Research to Prevent Blindness, New York, NY; an American Glaucoma Society Mid-Career Physician Scientist Award; and the Marlene S. and Leonard A. Hadley Glaucoma Research Fund.
Footnotes
 Disclosure: M.K. Garvin, P; M.D. Abràmoff, P; K. Lee, None; M. Niemeijer, None; M. Sonka, P; Y.H. Kwon, None
The authors thank Wallace L. M. Alward and John H. Fingert for permission to recruit study patients from their clinics, Marilyn Long for help in acquiring and organizing the image data, Zhihong Hu for helping to segment the neural canal opening, and Randy Kardon for helpful discussions. 
References
Nickells RW . Ganglion cell death in glaucoma: from mice to men. Vet Ophthalmol. 2007;10(suppl 1):88–94. [CrossRef] [PubMed]
Fitzgibbon T Taylor SF . Retinotopy of the human retinal nerve fibre layer and optic nerve head. J Comp Neurol. 1996;375:238–251. [CrossRef] [PubMed]
Ballantyne AJ . The nerve fiber pattern of the human retina. Trans Opthal Soc U K. 1947;66:179–191.
Vrabec F . The temporal raphe of the human retina. Am J Ophthalmol. 1966;62:926–938. [CrossRef] [PubMed]
Weber J Ulrich H . A perimetric nerve fiber bundle map. Int Ophthalmol. 1991;15:193–200. [CrossRef] [PubMed]
Garway-Heath DF Poinoosawmy D Fitzke FW Hitchings RA . Mapping the visual field to the optic disc in normal tension glaucoma eyes. Ophthalmology. 2000;107:1809–1815. [CrossRef] [PubMed]
Jansonius NM Nevalainen J Selig B . A mathematical description of nerve fiber bundle trajectories and their variability in the human retina. Vision Res. 2009;49:2157–2163. [CrossRef] [PubMed]
Wojtkowski M Leitgeb R Kowalczyk A Bajraszewski T Fercher AF . In vivo human retinal imaging by Fourier domain optical coherence tomography. J Biomed Opt. 2002;7:457–463. [CrossRef] [PubMed]
de Boer JF Cense B Park BH . Improved signal-to-noise ratio in spectral-domain compared with time-domain optical coherence tomography. Opt Lett. 2003;28:2067–2069. [CrossRef] [PubMed]
Abràmoff MD Garvin MK Sonka M . Retinal imaging and image analysis. IEEETrans Med Imaging. 2010;3:169–208.
Wojtkowski M Srinivasan V Fujimoto JG . Three-dimensional retinal imaging with high-speed ultrahigh-resolution optical coherence tomography. Ophthalmology. 2005;112:1734–1746. [CrossRef] [PubMed]
Leung CK Chan WM Yung WH . Comparison of macular and peripapillary measurements for the detection of glaucoma: an optical coherence tomography study. Ophthalmology. 2005;112:391–400. [CrossRef] [PubMed]
Hood DC Anderson SC Wall M Kardon RH . Structure versus function in glaucoma: an application of a linear model. Invest Ophthalmol Vis Sci. 2007;48:3662–3668. [CrossRef] [PubMed]
Leite MT Rao HL Zangwill LM Weinreb RN Medeiros FA . Comparison of the diagnostic accuracies of the Spectralis, Cirrus, and RTVue optical coherence tomography devices in glaucoma. Ophthalmology. 2011;118:1334–1339. [PubMed]
Ishikawa H Stein DM Wollstein G . Macular segmentation with optical coherence tomography. Invest Ophthalmol Vis Sci. 2005;46:2012–2017. [CrossRef] [PubMed]
Kotera Y Hangai M Hirose F Mori S Yoshimura N . Three-dimensional imaging of macular inner structures in glaucoma by using spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2011;52:1412–1421. [CrossRef] [PubMed]
Kwon YH Adix M Zimmerman MB . Variance owing to observer, repeat imaging, and fundus camera type on cup-to-disc ratio estimates by stereo planimetry. J Glaucoma. 2009;18:305–310. [CrossRef] [PubMed]
Garvin MK Abràmoff MD Wu X . Automated 3-D intraretinal layer segmentation of macular spectral-domain optical coherence tomography images. IEEE Trans Med Imaging. 2009;28:1436–1447. [CrossRef] [PubMed]
Quellec G Lee K Dolejsi M . Three-dimensional analysis of retinal layer texture: identification of fluid-filled regions in SD-OCT of the macula. IEEE Trans Med Imaging. 2010;29:1321–1330. [CrossRef] [PubMed]
Hu Z Abràmoff MD Kwon YH Lee K Garvin MK . Automated segmentation of neural canal opening and optic cup in 3D spectral optical coherence tomography volumes of the optic nerve head. Invest Ophthalmol Vis Sci. 2010;51:5708–5717. [CrossRef] [PubMed]
Niemeijer M Garvin MK van Ginneken B Sonka M Abràmoff MD . Vessel segmentation in 3D spectral OCT scans of the retina. In: Medical Imaging, Volume 6914 of Proceedings of the SPIE. Bellingham, Washington. 2008.
ITK, Insight Toolkit. Available at: http://www.itk.org . Accessed June 30, 2011.
Sung KR Na JH Lee Y . Glaucoma diagnostic capabilities of optic nerve head parameters as determined by Cirrus HD optical coherence tomography. J Glaucoma. 2001 Jun 1. [Epub ahead of print]
Zhu H Crabb DP Schlottmann PG . Predicting visual function from the measurements of retinal nerve fiber layer structure. Invest Ophthalmol Vis Sci. 2010;51:5657–5666. [CrossRef] [PubMed]
Hood DC Raza AS . Method for comparing visual field defects to local RNFL and RGC damage seen on frequency domain OCT in patients with glaucoma. Biomed Opt Express. 2011;2:1097–1105. [CrossRef] [PubMed]
Zhang X Bregman CJ Raza AS De Moraes G Hood DC . Deriving visual field loss based upon OCT of inner retinal thicknesses of the macula. Biomed Opt Express. 2011;2:1734–1742. [CrossRef] [PubMed]
Nakatani Y Higashide T Ohkubo S Takeda H Sugiyama K . Evaluation of macular thickness and peripapillary retinal nerve fiber layer thickness for detection of early glaucoma using spectral domain optical coherence tomography. J Glaucoma. 2011;20:252–259. [CrossRef] [PubMed]
Cho JW Sung KR Lee S . Relationship between visual field sensitivity and macular ganglion cell complex thickness as measured by spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2010;51:6401–6407. [CrossRef] [PubMed]
Lee JR Jeoung JW Choi J . Structure-function relationships in normal and glaucomatous eyes determined by time- and spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2010;51:6424–6430. [CrossRef] [PubMed]
Rao HL Zangwill LM Weinreb RN . Structure-function relationship in glaucoma using spectral-domain optical coherence tomography. ArchOphthalmol. 2011;129:864–871.
Kocaoglu OP Cense B Jonnal RS . Imaging retinal nerve fiber bundles using optical coherence tomography with adaptive optics. VisionRes. 2011;51:1835–1844.
Figure 1.
 
The segmentation of structures in each macular and peripapillary SD-OCT volume. (A) Example slice from a macular SD-OCT volume. (B) Five labeled surfaces (out of the 11 segmented) illustrated on the macular slice. These five surfaces enabled computation of the retinal nerve fiber layer thickness (between surface 1 and surface 2), the ganglion cell layer thickness (between surface 2 and surface 3), and a projection image of photoreceptor and retinal pigment epithelium complex (between surface 4 and surface 5). (C) Volumetric visualization of five surfaces from the macular volume. (D) Example slice from a peripapillary SD-OCT volume. (E) Five surfaces illustrated on the peripapillary slice. The red column shows the area of neural canal opening in which the layers were not defined. (F) Volumetric visualization of five surfaces from the peripapillary volume.
Figure 1.
 
The segmentation of structures in each macular and peripapillary SD-OCT volume. (A) Example slice from a macular SD-OCT volume. (B) Five labeled surfaces (out of the 11 segmented) illustrated on the macular slice. These five surfaces enabled computation of the retinal nerve fiber layer thickness (between surface 1 and surface 2), the ganglion cell layer thickness (between surface 2 and surface 3), and a projection image of photoreceptor and retinal pigment epithelium complex (between surface 4 and surface 5). (C) Volumetric visualization of five surfaces from the macular volume. (D) Example slice from a peripapillary SD-OCT volume. (E) Five surfaces illustrated on the peripapillary slice. The red column shows the area of neural canal opening in which the layers were not defined. (F) Volumetric visualization of five surfaces from the peripapillary volume.
Figure 2.
 
Registration of macular and peripapillary SD-OCT volumes. (A) Example projected image from a macular SD-OCT volume (obtained by computing the projected mean intensity of voxels in the segmented photoreceptor/retinal-pigment-epithelium complex). (B) Example projected image from a peripapillary SD-OCT volume. (C) Illustration of automated registration result.
Figure 2.
 
Registration of macular and peripapillary SD-OCT volumes. (A) Example projected image from a macular SD-OCT volume (obtained by computing the projected mean intensity of voxels in the segmented photoreceptor/retinal-pigment-epithelium complex). (B) Example projected image from a peripapillary SD-OCT volume. (C) Illustration of automated registration result.
Figure 3.
 
Local macular and peripapillary grids for each registered macula and peripapillary SD-OCT volumetric pair, on underlying projection image (A) and without projection image (B). Based on the center of the foveal pit and the center of the neural canal opening (each indicated with a plus), a reference line can be drawn between them and bisected. The macular grid of 66 squares (each 2° by 2°) is then placed on the macular volume so as to align the center with the center of the foveal pit and to rotate the grid so that the horizontal grid lines are parallel to the reference line. The square grid ends nasally at the midpoint of the reference line. The 12 peripapillary wedge-shaped regions are segmented in 15° increments. The neural canal opening defines the inner radial boundary and the midpoint of the reference line defines the outer radial boundary.
Figure 3.
 
Local macular and peripapillary grids for each registered macula and peripapillary SD-OCT volumetric pair, on underlying projection image (A) and without projection image (B). Based on the center of the foveal pit and the center of the neural canal opening (each indicated with a plus), a reference line can be drawn between them and bisected. The macular grid of 66 squares (each 2° by 2°) is then placed on the macular volume so as to align the center with the center of the foveal pit and to rotate the grid so that the horizontal grid lines are parallel to the reference line. The square grid ends nasally at the midpoint of the reference line. The 12 peripapillary wedge-shaped regions are segmented in 15° increments. The neural canal opening defines the inner radial boundary and the midpoint of the reference line defines the outer radial boundary.
Figure 4.
 
Correlation map between macular RGC layer and peripapillary RNFL (A) and corresponding r 2 values (B), using data from 57 eyes of 57 subjects. The color of each macular region in the correlation map in (A) corresponds to the color of peripapillary wedge representing highest correlation.
Figure 4.
 
Correlation map between macular RGC layer and peripapillary RNFL (A) and corresponding r 2 values (B), using data from 57 eyes of 57 subjects. The color of each macular region in the correlation map in (A) corresponds to the color of peripapillary wedge representing highest correlation.
Figure 5.
 
The scatterplots of all 66 macular regions. For each macular region, the best correlation between the macular region (GCL thickness) and the corresponding peripapillary wedge (RNFL thickness) is shown. Each square box contains the best scatterplot, r 2 value, and corresponding wedge letter.
Figure 5.
 
The scatterplots of all 66 macular regions. For each macular region, the best correlation between the macular region (GCL thickness) and the corresponding peripapillary wedge (RNFL thickness) is shown. Each square box contains the best scatterplot, r 2 value, and corresponding wedge letter.
Figure 6.
 
Color-coded correlation map (from all 57 subjects) with a peripapillary wedges overlay for each macular region. The background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white.
Figure 6.
 
Color-coded correlation map (from all 57 subjects) with a peripapillary wedges overlay for each macular region. The background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white.
Figure 7.
 
Comparison of resulting color-coded correlation map (with overlays) for 30 subjects with glaucoma (A) and 27 subjects with glaucoma suspicion (B). As in Figure 6 (which shows the correlation map with overlays for all subjects), the background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white. Note that the glaucoma case displays a pattern that more closely resembles expected nerve-fiber-bundle trajectories. In addition, for almost all macular regions, the number of wedges with a correlation not significantly different from that of the best wedge is smaller in the glaucoma case, further indicating a tighter relationship in this case.
Figure 7.
 
Comparison of resulting color-coded correlation map (with overlays) for 30 subjects with glaucoma (A) and 27 subjects with glaucoma suspicion (B). As in Figure 6 (which shows the correlation map with overlays for all subjects), the background color (and wedge letter) of each macular region corresponds to the peripapillary wedge with the largest squared Pearson's correlation coefficient. This best wedge is also colored in the wedge overlay for that region and indicated with an asterisk. Each wedge with a correlation coefficient not significantly different from this best wedge is also colored with its corresponding wedge color, with the remaining wedges in the overlay being indicated in white. Note that the glaucoma case displays a pattern that more closely resembles expected nerve-fiber-bundle trajectories. In addition, for almost all macular regions, the number of wedges with a correlation not significantly different from that of the best wedge is smaller in the glaucoma case, further indicating a tighter relationship in this case.
Figure 8.
 
Qualitative comparison of our structure-structure-derived correlation map and hand-drawn nerve fiber bundles from Fitzgibbon and Taylor. 2 (A) Illustration of nerve fiber bundles taken from Fitzgibbon and Taylor. 2 (B) Enlarged portion of retinal nerve fiber bundles overlaid on our color correlation map (from Fig. 4A).
Figure 8.
 
Qualitative comparison of our structure-structure-derived correlation map and hand-drawn nerve fiber bundles from Fitzgibbon and Taylor. 2 (A) Illustration of nerve fiber bundles taken from Fitzgibbon and Taylor. 2 (B) Enlarged portion of retinal nerve fiber bundles overlaid on our color correlation map (from Fig. 4A).
×
×

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.

×