Experimental Eye Research 83 (2006) 1096e1101www.elsevier.com/locate/yexer

Quantifying retinal nerve fiber layer thickness in whole-mounted retina

Xiang-Run Huang*, Robert W. Knighton, Valery Shestopalov

Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, 1638 N.W. 10th Ave., Miami, FL 33136, USA

Received 15 December 2005; accepted in revised form 21 May 2006

Available online 7 July 2006

Abstract

In order to relate optical measurement of the retinal nerve fiber layer (RNFL) to the underlying structure, one must have accurate values forRNFL thickness at the locations measured optically. The purpose of this study was to develop a method for measuring RNFL thickness atany location on retinal tissue previously studied by other optical imaging. The method developed used confocal laser scanning microscopy(cLSM) to provide both en face and cross-sectional images of a whole-mounted retina. Isolated rat retina was fixed with 3% glutaraldehyde.Nerve fiber bundles were identified by using phalloidin to label F-actin and ganglion cell bodies were identified by DAPI fluorescent counterstainof nuclei. The flat-mounted retina was examined by cLSM. 2-D images were collected through the retina to a depth at least covering the ganglioncell layer. The images were stacked to reconstruct cross-sectional images of the measured retina. Thickness of nerve fiber bundles was measuredon these synthesized cross sections and compared with the measurement from conventional histologic sections. The en face image displayedindividual nerve fiber bundles and ganglion cells between bundles as different colors. Blood vessels, which also bound phalloidin, were easilydistinguished from nerve fiber bundles. The en face image displayed the same pattern of nerve fiber bundles as seen in imaging measurementsand simplified the identification of corresponding areas in the two modalities. The cross-sectional images provided thickness measurements ofthe RNFL over the entire field-of-view, not just at the points represented by the conventional histologic section, resulting in a large increase inavailable data.� 2006 Elsevier Ltd. All rights reserved.

Keywords: retinal nerve fiber layer; whole-mounted retina; thickness; confocal laser scanning microscopy; phalloidin staining; histologic section

1. Introduction

The retinal nerve fiber layer (RNFL) in humans consists ofbundles of unmyelinated axons of retinal ganglion cells run-ning just under the surface of the retina. The RNFL is dam-aged in glaucoma and other diseases of the optic nerve. Tohelp diagnose and manage these diseases, various opticalmethods have been developed to assess the RNFL and severalare moving toward routine clinical use. Quantitative measure-ments, however, need a comprehensive understanding of thebasic optical properties of the RNFL and their underlyinganatomic basis. Optical properties of the RNFL have been

* Corresponding author. Tel.: þ1 305 482 4115; fax: þ1 305 326 6306.

E-mail addresses: xhuang3@med.miami.edu (X.-R. Huang), rknighton@

med.miami.edu (R.W. Knighton), vshestopalov@med.miami.edu (V.

Shestopalov).

0014-4835/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.

doi:10.1016/j.exer.2006.05.020

studied in vitro by various imaging methods (Dreher et al.,1992; Huang and Knighton, 2002; Huang and Knighton,2003; Knighton et al., 1992, 1998; Knighton and Huang,1999a; Knighton and Zhou, 1995; Weinreb et al., 1990). Inorder to relate an optical measurement to the RNFL structure,one must be able to retrieve the imaged tissue for histologicstudy. In particular, it is necessary to have accurate valuesfor RNFL thickness at the locations measured optically. Forexample, Knighton and Zhou (1995) studied the proportional-ity of the RNFL reflectance and its thickness; Morgan et al.(1998), Weinreb et al. (1990) and Huang and Knighton(2002) calculated the birefringence of the RNFL from RNFLretardance and thickness, and Toth et al. (1997), Huanget al. (1998) and Gloesmann et al. (2003) compared optical co-herence tomography (OCT) and histologic measurements ofRNFL thickness. Although standard histologic measurementsof RNFL thickness in tissue sections are widely used, the

1097X.-R. Huang et al. / Experimental Eye Research 83 (2006) 1096e1101

method has at least three major drawbacks: it is hard to locatethe exact positions measured optically, limited numbers ofnerve fiber bundles can be measured, and the process is verytime consuming.

This paper presents a method that uses confocal laser scan-ning microscopy (cLSM) to measure RNFL thickness at anylocation on a whole-mounted retina previously studied by im-aging reflectometry or polarimetry.

2. Methods and materials

2.1. Retinal tissue preparation

Rat and pig retinas were chosen for experiments becauseoptical imaging methods have been established to study theoptical properties of RNFL of these retinas and using both tis-sues provided a range of RNFL thickness. The protocol for theuse of animals was approved by the Animal Care and UseCommittee of the University of Miami. Tissue preparation fol-lowed previously developed procedures (Huang and Knighton,2002; Knighton and Huang, 1999b). Briefly, an eye of an anes-thetized animal was removed and the animal was euthanized.For a rat eye, an eye cup of 5 mm diameter that included theoptic nerve was excised with a razor blade. For a pig eye, theeye was hemisected and 4 � 4 mm pieces of tissue wereexcised around the optic nerve head. The excised tissue wasplaced in a dish of warm (33e35 �C) oxygenated physiologicsolution. The retina was dissected free of the retinal pigmentepithelium and choroid with a fine glass probe and then drapedacross a slit in a black membrane with the photoreceptor sideagainst the membrane. A second, thinner membrane with a slitmatched to the black membrane was put on the RNFL surfaceto gently stretch the retina and eliminate wrinkles. Themounted retina was fixed in either 4% paraformaldehyde or3% glutaraldehyde for 25 min at room temperature and rinsedthoroughly in phosphate buffered saline (PBS).

2.2. Optical imaging

Either a reflectometer or a polarimeter set in transmissionmode was used to optically image retinas (Huang andKnighton, 2002; Knighton and Huang, 1999b). The devicesprovided monochromatic illumination and imaged the tissueonto a cooled charge-coupled device (CCD). The CCD pro-vided a pixel size of about 4 mm on a specimen in an aqueousmedium and an en face image of the retinal nerve fiber bundleswith a field of view of 2.0 � 2.0 mm. After the optical imag-ing, the fixed retina was freed from the supporting membranesand preserved in PBS at 4 �C for fluorescence staining.

2.3. Fluorescence staining procedures

Nerve fiber bundles can be identified by antibodies specificto neurofilaments (Villegas-Perez et al., 1998; Wang et al.,2003). The penetration depth of antibodies, however, isgenerally limited to a few microns despite thorough

permeabilization. An adequate immunohistochemical stainingof RNFL required 2e3 days, which is rather time-consuming.

Retinal nerve fiber bundles in whole-mounted retinas areenriched in F-actin and were identified by using phalloidinlabeling specific for F-actin. Ganglion cell bodies in the innerretina were identified by DAPI fluorescent counterstain of nu-clei. To reduce autofluorescent background in the retinas fixedwith glutaraldehyde, the tissue was incubated in freshly pre-pared 1% NaBH4 for 20 min before the staining procedures.The tissue was permeabilized in PBS containing 0.4% Tri-tonX-100 for 30 min followed by incubation in blocking se-rum (5% horse serum and 0.4% TritonX-100) for 60 min atroom temperature. The tissue was washed in PBS (threechanges of ten minutes each) and transferred into a solutionof phalloidin (diluted 1:100 in PBS, Alexa Fluor 488 Phallio-din, Molecular Probes) for 20 min at room temperature. Afterrinsing in PBS containing 0.2% TritonX-100, the tissue wasincubated in DAPI (FluoroPure grade, Molecular Probes) for20 min in subdued lighting. The stained retina was rinsedagain and mounted in an imaging chamber (Coverwell�,Electron Microscopy Sciences) with PBS as a mountingmedium. The retina was preserved at 4 �C until confocalmicroscopy imaging.

To confirm identification of nerve fiber bundles by phalloi-din staining, retinal cross-sections were prepared and stainedboth with antibodies specific to neurofilaments and with phal-loidin dye. Retina fixed with paraformaldehyde was embeddedin melted 3.5% Agar in PBS and sliced into 50 mm sectionsusing a Vibratome 1000 (Vibratome Inc., St. Louis, MO).The tissue was washed in PBS containing 0.2% TritonX-100for 1 h and incubated in blocking serum (1 � PBS containing1% horse serum and 1% BSA) for one hour at room tempera-ture. The tissue was then transferred into a solution of mono-clonal mouse primary antibody to neurofilament (200 KDa,RT97, Novocastra Laboratories Ltd.) diluted 1:75 with PBScontaining 1% BSA for one to two hours and stained overnightat 4 �C. After three one hour washes in PBS containing 0.15%Tween 20 at room temperature (the last wash in PBS with 5%goat serum), retina samples were incubated with secondarygoat anti-mouse IgG conjugated to Alexa-568 (1:250) foranother two hours at room temperature and then washed thor-oughly with PBS. The tissue was further stained with phalloi-din and DAPI following the procedures described above.

2.4. Confocal laser scanning imaging

To relate measured optical properties of the RNFL to itsthickness, an imaging method has to provide: 1) an en faceimage of the retina that can be used to identify the locationof individual nerve fiber bundles measured optically; and 2)thickness measurements of RNFL in the selected area.

A cLSM (cLSM510 Carl Zeiss Microimaging, Inc.) wasused to provide both en face and cross sectional images ofthe whole-mounted fluorescence stained retina. A 20� objec-tive was used to locate regions that had been imaged opticallyand en face images were taken. A 40� oil objective was thenswitched in and en face subimages were taken with a full field

1098 X.-R. Huang et al. / Experimental Eye Research 83 (2006) 1096e1101

of view of 230 mm � 230 mm. To measure the thickness ofnerve fiber bundles, Z-stack scan mode was used, with whicha series of 2-D images (en face images at different depths) wascollected through the retina to a depth at least covering theganglion cell layer. The retina was then reconstructed in 3-Dand cross-sectional images were synthesized from the recon-struction. Z-stack scan was applied either to the whole fieldof view of an en face image or to a defined region of interest(ROI). With ROI scan, a thin rectangle (2e3 mm wide) was de-fined with the long side perpendicular to the longitudinal axesof bundles in the en face image. Z-stack scan was applied tothe defined ROI and several such scans were collected atevenly spaced positions along nerve fiber bundles.

Thickness of a nerve fiber bundle was measured from recon-structed cross sections as the length of a line perpendicular tothe retinal surface that passed through the center of the bundle.

To reconcile the magnifications of optical and cLSM mea-surements, a piece of nylon mesh with a known grid size wasmounted in the same way as a retina and imaged by both mo-dalities. The magnification value of the optical imaging wasthen adjusted so that the grid size measured by both modalitieswas identical.

2.5. Histologic measurements

RNFL thickness measured from confocal images was com-pared with conventional histologic measurements of the samebundles. After the cLSM measurement, a small piece of retinacontaining the measured nerve fiber bundles was cut out andpostfixed with osmium, dehydrated, and embedded in Spurr’smedium for sectioning. Two sections of 0.5 mm thickness eachand 50 mm apart were cut across the RNFL and stained withMallory’s methylene blue. Digital micrographs of the retinawere obtained with phase contrast. The location of a histologicsection on its corresponding confocal image was identified bymatching the distances between blood vessels in the micro-graph and in the confocal en face image. To reduce ambiguity,three blood vessels were often used for simultaneous distancematching. The bundles measured by the cLSM were then iden-tified by matching the distances between the bundles and theblood vessels. Because bundle thickness was nearly constantalong bundles within the 40� en face image, the bundle thick-ness was measured in a confocal cross section nearest to theplace where the histologic section was located. In the sameway as in confocal cross sections, the thickness of a nerve fiberbundle was measured on a line through its center.

3. Results

Retinal nerve fiber bundles were identified both by anti-bodies specific to neurofilaments and by phalloidin stainingof actin. Fig. 1AeD shows images of a cross-sectionally pre-pared retina stained with the neurofilament specific antibody,phalloidin and DAPI. Nerve fiber bundles lying just underthe retinal surface were the only structures identified by theantibody (Fig. 1A). The individual bundles were also recog-nized by phalloidin staining (Fig. 1B). The cross sections of

the bundles identified by phalloidin were well overlaid bythe areas stained with the antibody (Fig. 1D) and also wereclearly separated from deeper layers (Fig. 1B). A single layerof ganglion cells lying under and within the RNFL was seenwith DAPI counterstain (Fig. 1C); this layer separating theRNFL and deeper layers gave a further confirmation of theRNFL boundary. Because phalloidin staining could provideidentification of the retinal nerve fiber bundles without theadded complexity of neurofilament specific antibodies, itwas used alone for subsequent measurements of RNFL thick-ness in whole-mounted retinas.

Fig. 1E,F illustrates the correspondence between an opticalimage of the RNFL and an en face fluorescence image fromcLSM. In a polarimeter image of a rat retina (Fig. 1E), nervefiber bundles appeared as bright stripes separated by darkergaps of retinal tissue. To locate cLSM scanning regions tothe corresponding areas in optical measurements, en face im-ages with a 20� objective were taken. Fig. 1F shows an enface image of a whole-mounted retina stained with phalloidinand DAPI, which displayed individual nerve fiber bundles andganglion cells between bundles as different colors. Blood ves-sel walls, which bound much more phalloidin, were easily dis-tinguished from nerve fiber bundles. The en face image of theRNFL showed the same pattern of nerve fiber bundles and ma-jor blood vessels as seen optically (dashed-line box in Fig. 1E)and simplified the identification of corresponding areas in thetwo modalities.

To achieve high resolution depth scanning, en face imageswere taken with a 40� oil objective (solid-line box in Fig. 1F).The thickness of retinal nerve fiber bundles over the wholefield was then measured with the Z-stack scan mode, whichcollected a series of en face images through the retina.Fig. 2A demonstrates one such image. By reconstructing thisseries of en face images, a cLSM image analysis tool providedsynthesized cross-sectional images either across bundles(Fig. 2B) or along a bundle (Fig. 2C) at any location in thefield. Bundle thickness could be measured from the synthe-sized cross sectional images (Fig. 2B,C). To achieve high ac-curacy of thickness measurements, 72 en face images witha Z-scan step of 0.36 mm were used. Although the Z-stackmode provided RNFL thickness measurement over the wholefield, it took about 20 min to obtain a full field image witha 40� objective (230 mm � 230 mm).

Continuous thickness measurements along bundles, as pro-vided by the Z-stack scan mode are, in general, not necessary.To reduce the cLSM acquisition time, RNFL thickness canalso be measured at discrete positions along bundles withthe scan positions precisely defined by using a region of inter-est (ROI) scan mode. Fig. 2D demonstrates five defined scanpositions on an en face image; ROIs, each 2 mm wide, were45 mm apart. Fig. 2E shows the corresponding cross-sectionalimages for each ROI. With the same depth resolution asFig. 2B,C, each ROI scan took about 20 s.

Our cLSM measurement method could measure nerve fiberbundles in different species over a wide thickness range.Fig. 3AeD shows pig retinal nerve fiber bundles located about3 mm (Fig. 3A,B) and 1 mm (Fig. 3C,D) from the disc margin.

1099X.-R. Huang et al. / Experimental Eye Research 83 (2006) 1096e1101

A B

C D

E

F

Fig. 1. Phalloidin staining identifies retinal nerve fiber bundles. (AeD) Histo-

logical cross section of a rat retina. A: nerve fiber bundles specifically identi-

fied by neurofilament antibody (red). B: retinal layers revealed by phalloidin

staining of actin. C: nuclear layers with DAPI counterstain. D: a composite im-

age of all fluorescence stains. The nerve fiber bundles stained by phalloidin co-

localize with the areas stained by neurofilament antibody. (EeF) Localization

of a confocal en face image on its corresponding optical image. E: retina im-

aged by an imaging polarimeter. Nerve fiber bundles appear as bright stripes.

F: an en face cLSM image with a 20� objective of the same retina stained with

phalloidin (green) and DAPI (blue). The image reveals the same pattern of

nerve fiber bundles as in the optical image and allows precise localization

Bundles with thickness around 20 mm were well defined(Fig. 3B). Although the fluorescence signal decreased withincreased scan depth, bundles with thickness up to 55 mmcould still be resolved (Fig. 3D). Fig. 3EeG displays bundlesof three different thickness in rat RNFL. Fig. 3E shows 40 mmthick bundles near the optic disc, Fig. 3F shows 15 mm thicknerve fiber bundles located about 3 mm from the disc marginand Fig. 3G demonstrates the capability of measuring thinnerve fiber bundles under pathological conditions. The bundlesin Fig. 3G were located near the same eccentricity as those inFig. 3F but were damaged by elevated intraocular pressure(Levkovitch-Verbin et al., 2002). Although the thickness ofthe nerve fiber bundles was reduced to about 5 mm, theywere still clearly identifiable in the cross-sectional image.

To compare cLSM and conventional histologic sections,thickness of nineteen bundles in five normal rat retinas wasmeasured by both methods. To locate the bundles in histologicsections on the en face image, it was necessary to expand thelateral distance between blood vessels in histologic section byabout 3%. The mean thickness measured with the confocalimages was 11.4 � 2.9 mm (mean � s.d.) while for histologicsections it was 8.8 � 2.4 mm. A paired t-test showed that themeasured bundle thickness was significantly different( p < 0.001) for the two methods.

4. Discussion

Comparison of RNFL thickness with its correspondingoptical measurement can be required, as in birefringence mea-surement, or can serve as the gold standard for thickness mea-sured by another method, such as OCT. Imaging technologyprovides visualization of RNFL over an extended area of theretina. To correlate or validate the optical measurementswith histologic sections is not practical across the whole field;the process is time consuming, locating sections on opticalimages is difficult, and available data points are limited.

The method described in this paper uses confocal laserscanning microscopy with phalloidin staining to identify reti-nal nerve fiber bundles in a whole-mounted retina. The en faceimages display the same pattern of nerve fiber bundles andblood vessels as seen in optical imaging measurements andsimplify the identification of corresponding areas in the twomodalities. The cross-sectional images provide thickness mea-surements of the RNFL at any point in the field-of-view, notjust at the points represented by a conventional histologicalsection, resulting in a large increase in available data.

A general concern of RNFL thickness measurements is thecorrespondence of a measured value with the previously stud-ied optical properties. En face images from cLSM provide anestimate of the lateral change of the tissue between optical andcLSM measurements. Distances between 12 pairs of bundlesin four retinas were measured in cLSM en face images and

of the area in E (dashed-line box). The solid-line boxes in E and F outline

a full field of view for a 40� objective, with which depth scans were generally

taken. Blood vessels (BV) were identified by orientation and strong phalloidin

staining.

1100 X.-R. Huang et al. / Experimental Eye Research 83 (2006) 1096e1101

the corresponding optical images. The distances measured incLSM images were about 3% higher than that measured inthe optical images. These slightly higher values may be dueto tissue swelling caused by tissue processing after opticalmeasurements.

Lateral distance matching of blood vessels in histologic sec-tions and cLSM en face images suggested that about 3% lateralshrinkage of the retinas occurred in histologic sections. Thick-ness of RNFL measured by the confocal method was comparedwith histologic measurements and found to be about 22%higher. Tissue dehydration in histologic processing is often es-timated to cause 10e15% shrinkage (Ogden, 1983). The highervalue found in this study could be due to methodological differ-ences (Frenkel et al., 2005). The large difference in shrinkagebetween lateral and thickness measurements may reflect tissueanisotropy in the whole mounted retina.

A

B

C

E

D

Fig. 2. Two methods for RNFL thickness measurement demonstrated in rat ret-

ina. (AeC) Measurement in Z-stack image mode. A: an en face image of the

RNFL at the depth shown by the thin lines in B, C. B: reconstructed cross sec-

tion perpendicular to nerve fiber bundles (horizontal dashed line) allows mea-

surements of thickness on multiple bundles. C: reconstructed cross section

along a bundle (vertical dashed line) provides longitudinal thickness measure-

ments on one bundle. BV: blood vessel. Z-scan step: 0.36 mm. (DeE) Mea-

surement from a series of cross sections. D: the en face image used to

define the positions of sectional scans (dashed lines). E: cross sections of

the retina at the corresponding positions in A. BV: blood vessel. Z-scan

step: 0.34 mm.

Advantages and disadvantages of the cLSM method re-ported here should be mentioned. Although not specific toaxons, phalloidin staining of F-actin provides identificationof individual nerve fiber bundles (Fig. 1) and, by eliminatingneurofilament-specific antibodies, greatly shortens the proce-dure of fluorescence staining. Furthermore, deep penetrationof phalloidin allows measurement of thick RNFL (Fig. 3).

A

B

C D

E

F

G

Fig. 3. Cross-sectional method applied in different species and to bundles of

different thickness. (AeD) Thick RNFL of a pig retina. A, C: en face images

of regions located about 3 mm and 1 mm from the disc margin, respectively. B,

D: cross sections at the positions in A and C. Z-scan step: 0.36 mm. (EeG)

Cross-sectional images of rat RNFL of various thickness. E: thick bundles lo-

cated near the optic disc. F: thinner bundles located approximately 3 mm from

the disc margin in a normal retina. G: very thin bundles located as in F but in

an eye with elevated IOP. BV: blood vessel. Z-scan step: 0.36 mm.

1101X.-R. Huang et al. / Experimental Eye Research 83 (2006) 1096e1101

However, lack of specific binding causes a noisy background inconfocal images. To achieve precise RNFL thickness measure-ment, high depth resolution is required, which can be achievedby appropriate selections of the numerical aperture (NA) of anobjective and the confocal aperture size in fluorescence detec-tion channels. In this study, a depth resolution of 0.4 mm wasachieved by using a 40� oil objective with a NA of 1.3 and set-ting the aperture diameter to 55 mm. Depth resolution, however,decreases with the depth of the scan in the tissue (Majlof andForsgren, 2002). Weak signals from deep portions of thick bun-dles also limited the definition of the bundles. Together with thelimited working distance of an objective with high numericalaperture (about 200 mm for the objective used in this study),the method may not be able to measure very thick nerve fiberbundles, such as occur in human RNFL.

Regional damage of optic nerve axons has been found inanimal models of glaucoma by optic nerve axonal counts(Johnson et al., 2000; Mabuchi et al., 2004; Morrison et al.,1997). However, axonal counts have limited ability to identifyfocal damage, a hallmark of glaucoma (Frenkel et al., 2005).Peripapillary RNFL thickness can provide data on such dam-age, because all bundles converge to the ONH. Becausea whole-mounted retina provides a full view of nerve fiberbundle distribution around the ONH, the cLSM method mayprovide a means to study RNFL damage in animal models. Fo-cal changes of never fiber bundles might be revealed on an enface image and degrees of damage could be further assessedon cross-sectional measurements of bundle thickness. Thepower of the confocal method in measuring RNFL damagedby elevated IOP has been demonstrated in this study (Fig. 3G).

In summary, individual retinal nerve fiber bundles can beidentified with phalloidin staining in a whole-mounted retina.Confocal en face images of the retina provide easy and precisematching to corresponding optical images. Bundle thicknesscan be measured across the whole field of view, providinga large amount of available data.

Acknowledgements

This study was supported by NIH grants No. R01-EY008684, R01-EY013516, and center grant P30-EY014801and by an unrestricted grant to the University of Miamifrom Research to Prevent Blindness, Inc.

References

Dreher, A.W., Reiter, K., Weinreb, R.N., 1992. Spatially resolved birefrin-

gence of the retinal nerve-fiber layer assessed with a retinal laser ellipsom-

eter. Appl. Opt. 31, 3730e3735.

Frenkel, S., Morgan, J.E., Blumenthal, E.Z., 2005. Histological measurement

of retinal nerve fibre layer thickness. Eye 19, 491e498.

Gloesmann, M., Hermann, B., Schubert, C., Sattmann, H., Ahnelt, P.K.,

Drexler, W., 2003. Histologic correlation of pig retina radial stratification

with ultrahigh-resolution optical coherence tomography. Invest. Ophthal-

mol. Vis. Sci. 44, 1696e1703.

Huang, X.-R., Knighton, R.W., 2002. Linear birefringence of the retinal nerve

fiber layer measured in vitro with a multispectral imaging micropolarime-

ter. J. Biomed. Opt. 7, 199e204.

Huang, X.-R., Knighton, R.W., 2003. Diattenuation and polarization preserva-

tion of retinal nerve fiber layer. Appl. Opt. 42, 5737e5743.

Huang, Y.J., Cideciyan, A.V., Papastergiou, G.I., Banin, E., Semple-

Rowland, S.L., Milam, A.H., Jacobson, S.G., 1998. Relation of optical

coherence tomography to microanatomy in normal and rd chickens. Invest.

Ophthalmol. Vis. Sci. 39, 2405e2416.

Johnson, E.C., Deppmeier, L.M.H., Wentzien, S.K.F., Hsu, I., Morrison, J.C.,

2000. Chronology of optic nerve head and retinal responses to elevated

intraocular pressure. Invest. Ophthalmol. Vis. Sci. 41, 431e442.

Knighton, R.W., Baverez, C., Bhattacharya, A., 1992. The directional reflec-

tance of the retinal nerve fiber layer of the toad. Invest. Ophthalmol.

Vis. Sci. 33, 2603e2611.

Knighton, R.W., Huang, X.-R., 1999a. Directional and spectral reflectance of

the rat retinal nerve fiber layer. Invest. Ophthalmol. Vis. Sci. 40, 639e647.

Knighton, R.W., Huang, X.-R., 1999b. Visible and near-infrared imaging of

the nerve fiber layer of the isolated rat retina. J. Glaucoma 8, 31e37.

Knighton, R.W., Huang, X.-R., Zhou, Q., 1998. Microtubule contribution to

the reflectance of the retinal nerve fiber layer. Invest. Ophthalmol. Vis.

Sci. 39, 189e193.

Knighton, R.W., Zhou, Q., 1995. The relation between reflectance and thick-

ness of the retinal nerve fiber layer. J. Glaucoma 4, 117e123.

Levkovitch-Verbin, H., Quigley, H.A., Martin, K.R.G., Valenta, D.,

Baumrind, L.A., Pease, M.E., 2002. Translimbal laser photocoagulation

to the trabecular meshwork as a model of glaucoma in rats. Invest.

Ophthalmol. Vis. Sci. 43, 402e410.

Mabuchi, F., Aihara, M., Mackey, M.R., Lindsey, J.D., Weinreb, R.N., 2004.

Regional optic nerve damage in experimental mouse glaucoma. Invest.

Ophthalmol. Vis. Sci. 45, 4352e4358.

Majlof, L., Forsgren, P.-O., 2002. Confocal microscopy: important consider-

ations for accurate imaging. In: Matsumoto, B. (Ed.), Cell Biological

Applications of Confocal Microscopy. Academic Press, Boston.

Morgan, J.E., Waldock, A., Jeffery, G., Cowey, A., 1998. Retinal nerve fiber

layer polarimetry: histological and clinical comparison. Br. J. Ophthalmol.

82, 684e690.

Morrison, J.C., Moore, C.G., Deppmeier, L.M.H., Gold, B.G., Meshul, C.K.,

Johnson, E.C., 1997. A rat model of chronic pressure-induced optic nerve

damage. Exp. Eye Res. 64, 85e96.

Ogden, T., 1983. Nerve fiber layer of the macaque retina: retinotopic organi-

zation. Invest. Ophthalmol. Vis. Sci. 24, 85e98.

Toth, C.A., Narayan, D.G., Boppart, S.A., Hee, M.R., Fujimoto, J.G.,

Birngruber, R., Cain, C.P., DiCarlo, C.D., Roach, W.P., 1997. A compari-

son of retinal morphology viewed by optical coherence tomography and by

light microscopy. Arch. Ophthalmol. 115, 1425e1428.

Villegas-Perez, M.P., Lawrence, J.M., Vidal-Sanz, M., Lavail, M.M.,

Lund, R.D., 1998. Ganglion cell loss in RCS rat retina: a result of compres-

sion of axons by contracting intraretinal vessels linked to the pigment

epithelium. J. Comparative Neurol. 392, 58e77.

Wang, L., Dong, J., Cull, G., Fortune, B., Cioffi, G.A., 2003. Varicosities of

intraretinal ganglion cell axons in human and non-human primates. Invest.

Ophthalmol. Vis. Sci. 44, 2e9.

Weinreb, R.N., Dreher, A.W., Coleman, A., Quigley, H., Shaw, B., Reiter, K.,

1990. Histopathologic validation of Fourier-ellipsometry measurements of

retinal nerve fiber layer thickness. Arch. Ophthalmol. 108, 557e560.