Comparison of optical coherence tomographyand fundus photography for measuring theoptic disc size

Aljoscha S. Neubauer, Tina R. Krieglstein, Christos Chryssafis, Martin Thieland Anselm Kampik

Department of Ophthalmology, Ludwig-Maximilians-Universitat, Mathildenstr. 8, 80336 Munich,

Germany

Abstract

Purpose: To assess the agreement and repeatability of optic nerve head (ONH) size measurements

by optical coherence tomography (OCT) as compared to conventional planimetry of fundus

photographs in normal eyes.

Methods: For comparison with planimetry the absolute size of the ONH of 25 eyes from 25 normal

subjects were measured by both OCT and digital fundus photography (Zeiss FF camera 450).

Repeatability of automated Stratus OCT measurements were investigated by repeatedly measuring

the optic disc in five normal subjects.

Results: Mean disc size was 1763 ± 186 vertically and 1632 ± 160 lm horizontally on planimetry.

On OCT, values of 1772 ± 317 lm vertically (p ¼ 0.82) and a significantly smaller horizontal

diameter of 1492 ± 302 lm (p ¼ 0.04) were obtained. The 95% limits of agreement were ()546 lm;

+527 lm) for vertical and ()502 lm; +782 lm) for horizontal planimetric compared to OCT

measurements. In some cases large discrepancies existed. Repeatability of automatic measure-

ments of the optic disc by OCT was moderately good with intra-class correlation coefficients (ICC) of

0.78 horizontally and 0.83 vertically. The coefficient of repeatability indicating instrument precision

was 80 lm for horizontal and 168 lm for vertical measurements.

Conclusions: OCT can be used to determine optic disc margins in moderate agreement with

planimetry in normal subjects. However, in some cases significant disagreement with photographic

assessment may occur making manual inspection advisable. Automatic disc detection by OCT is

moderately repeatable.

Keywords: glaucoma, imaging, optical coherence tomography, optic disc, optic nerve head

Introduction

To date changes of certain criteria of the optic nervehead (ONH) such as cup/disc ratio, corrected for discsize, neuroretinal rim or peri-papillary atrophy, are vitalfor the diagnosis of glaucoma (Jonas et al., 1999). Toobjectively document and follow changes currently the

standard method is still stereo fundus photography ofthe ONH. The main disadvantage of this method is thatit is highly subjective, time-consuming and has a broadinter-observer variability (Garway-Heath et al., 1999).Therefore other methods for quantitative assessment ofthe disc gain increasing importance such as the Heidel-berg Retina Tomograph (HRT) and the Retinal Thick-ness Analyser (RTA). However, those methods requireuser intervention to manually determine the border ofthe optic disc leaving some inter-observer variability(Garway-Heath et al., 1999; Itai et al., 2003).

Recently, with the Stratus Optical Coherence Tomog-raph (OCT; Carl Zeiss Meditec Inc., Dublin, CA, USA)optical coherence tomography has been adopted in acommercial machine to also measure the optic disc. This

Received: 22 November 2004

Revised form: 3 June 2005

Accepted: 20 June 2005

Correspondence and reprint requests to: Tina R. Krieglstein.

Tel.: +49-89-5160-3811; Fax: +49-89-5160-5160.

E-mail address: Tina.Krieglstein@med.uni-muenchen.de

Ophthal. Physiol. Opt. 2006 26: 13–18

ª 2006 The College of Optometrists 13

method offers the advantage of being capable ofdetermining optic disc size independently from any userintervention. Since the method was first published in1991 (Huang et al., 1991) OCT has been applied inophthalmology for numerous diagnostic applications,mainly regarding the retina (Hee et al., 1995) and hasbeen shown to measure accurately and reproducibly forthis purpose (Muscat et al., 2002). In a first studyfocussing on the optic disc, Schuman et al. (2003)clinically compared OCT with confocal HRT scanningand found a good correlation of those two methods.However, little is known about the OCT propertiescompared to the current standard method of photogra-phic planimetry. Therefore the purpose of this study wasto assess the agreement of measurements compared tomanual planimetry on fundus photographs and todetermine repeatability of automatic disc border detec-tion by OCT in normal eyes.

Materials and methods

Comparison of OCT and photographic planimetry

A total of 25 eyes from 25 normal Caucasian individualswere included. Mean age was 46 years (±15 S.D.;range: 26–75 years): 10 subjects were female, 15 male.One eye was randomly chosen for further analysisyielding 14 right and 11 left eyes. All participantsunderwent a complete ophthalmic evaluation and afterpupil dilation with 1% tropicamide, fundus photogra-phy and OCT scanning were performed in random orderat the same visit. Inclusion criteria were a non-glauco-matous, normal-appearing ONH on funduscopy (butallowing variants such as tilted discs or myopic peri-papillary atrophy), clear ocular media and good abilityto fixate defined by visual acuity >20/40. Subjects withany fundus abnormalities, refractive errors more than+6.0 or )8.0 D spherical equivalent and contraindica-tions to pupil dilation were excluded. No patients had tobe excluded due to scan quality. All subjects wereexamined after informed consent was obtained, and allresearch was conducted in accordance with institutionalguidelines and conformed to the tenets of the WorldMedical Association Declaration of Helsinki.

The technical details of OCT scanning have beendescribed in detail elsewhere (Huang et al., 1991; Heeet al., 1995). For OCT scanning of the optic disc weapplied the method described by Schuman et al. (2003):six radial scans spaced 30 � apart were performed centredon the optic disc. Technical quality of all OCT scans waschecked by one experienced observer (ASN) and centra-tion assured on the simultaneously recorded fundusimage. No scans had to be excluded for technical reasonsas scans had been appropriately centred and repeatedwhere necessary during OCT recording avoiding any

artefacts from eye movements. The one horizontal andone vertical scan were chosen for comparison to theplanimetric disc size. For manual measurement the opticdisc margin was determined on OCT by the border of theouter highly reflective band representing the retinalpigment epithelium/choriocapillaris layer. A straight lineconnecting the edges was drawn and a parallel line wasconstructed 150 lm anterior representing the referenceplane. Horizontal and vertical diameters were measuredwithin the calibrated OCT software by one examiner(ASN). A total of 19 eyes were scanned with OCT 2000using software A5 and 5.65 mm scan length. Theremaining six eyes were scanned with the Stratus OCTwith radial lines on the optic disc mode when available atour department. In those eyes, measurements of the opticdisc margin were also performed manually as describedabove. For both OCT models no specific data formagnification such as axial length were entered in theinstrument’s software thus yielding a constant value forcorrection of magnification.

The 30 � stereo photographs of the optic disc weretaken digitally utilising a Zeiss FF 450 camera system byan ophthalmic photographer different from the techni-cian performing OCT. Optic disc margins were thenmanually defined on a computer screen by one experi-enced examiner (TRK) masked to the OCT results.Optic disc margin was defined as the entire retinal aspectof the optic nerve as delineated by the inner aspect of thering of Elschnig (Jonas et al., 1988a).

The horizontal and vertical diameters of the optic discwere measured in pixels and converted into absolutedistances using the method described by Bengtsson andKrakau (BK-method; Bengtsson and Krakau, 1992).The best spectacle correction was used in this method tocorrect the optic disc measurements, a method whichgives similar values to the original formula developed byLittmann (1992) without the need for axial lengthmeasurements. When avoiding large deviations fromthe normal axial length the method by Bengtsson andKrakau gives results similar to those obtained byutilising axial length measurements (Garway-Heathet al., 1998). For axial lengths between 20 and 24 mmless than 3% differences result between the BK-methodapplied for planimetry and a constant correction (as forOCT). On the other hand, for axial lengths of more than27 mm the errors between both methods may result inup to 10% larger measurements with the BK-method(Garway-Heath et al., 1998). An appropriate magnifi-cation factor for the Zeiss camera system was obtainedfrom the literature (Rudnicka et al., 2001) and verifiedby a model eye. This custom-made, simple plastic modeleye consisted of an appropriate anterior refractive lensand a round body with 23 mm axial length yieldingrefractive emmetropia. At the inner back part of themodel eye a single fibre of fibreglass with a known outer

14 Ophthal. Physiol. Opt. 2006 26: No. 1

ª 2006 The College of Optometrists

diameter of exactly 1000 lm was imaged to verify themagnification factor of the Zeiss camera system.Statistical analysis was performed using MedCalc

Version 7.2.0.2 (Frank Schoonjans, http://www.med-calc.be) a statistical software package focused onbiomedical research. Besides paired t-testing for differ-ences, analysis by Bland–Altman plots (Bland andAltman, 1986) was performed. Those plots are asuitable tool to visualise and evaluate agreementbetween two methods. In brief, on the x-axis theaverage of the two methods is given, while on the y-axis the differences between the two methods areplotted (see Figures 1 and 2). This allows us to easilysee if the absolute values influence the differencesbetween methods. The absolute difference of the twomeans is indicated by a horizontal line. Additionallylimits of agreement can be estimated from the inves-tigated sample: usually ±1.96 S.D. of the sampleyielding a 95% confidence interval (in the case of anormal distribution) are plotted.Power calculations were also performed using Med-

Calc. An alpha of 0.05 and beta error of 0.1 wasassumed to give sufficient statistical power. Under thoseconditions to detect a difference of 200 lm mean withboth groups having a S.D. of 200 lm a total sample ofn ¼ 22 is required. For a difference of 300 lm n ¼ 10 isrequired and for 100 lm n ¼ 85. Therefore a sample ofn ¼ 25 was considered sufficient to detect significantdifferences of 200–300 lm. Post hoc power analysis withthe observed S.D. of 300 and 200 lm for the two groupsand n ¼ 25 yielded a power (i.e. 1-beta) of 77% for a200 lm difference and a power of 98% for a 300 lmdifference assuming an alpha error of 0.05. Moreillustrative, a non-significant difference with n ¼ 25 as

in this study, implies that both group means lie withinthe 95% confidence limits of ±145 lm.

Repeatability

To assess repeatability of automatic measurements ofrecent generation OCT the optic disc was measuredthree times by Stratus OCT in five eyes of fiveophthalmologically completely normal subjects (threeright, two left eyes) with dilated pupils. Automaticborder detection was not available in the OCT 2000version and was measured manually. Therefore repea-tability was assessed by Stratus OCT and not with theOCT 2000, the technical repeatability of which is knownto be good (Koozekanani et al., 2000; Muscat et al.,2002; Schuman et al., 2003). Repeatability of computer-assisted planimetry is known to be 4.0% to 7.2% for theS.D. of the mean disc area (Garway-Heath et al., 1999).

With the Stratus OCT three different sessions (newseating of the subject and new OCT setup) wereperformed within the same day. The fast optic discmode was applied for scanning and no user correctionwas performed on the disc analysis obtained. Statisticalanalysis consisted of intra-class correlation coefficients(ICC; http://department.obg.cuhk.edu.hk) and a meth-od proposed by Bland and Altman (1986): the differencebetween the first and last measurement was calculated.The mean of those differences should equal zero if nobias exists with repeated measurements. Additionallythe S.D. of those differences was calculated. Multiplyingthis S.D. by 2 gives the coefficient of repeatability, whichcan serve as a limit for real changes in measurements,which are not explained by instrument variation (Blandand Altman, 1986). All values obtained are given inTable 1.

1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100

Horizontal measurementsAverage photo and OCT (micrometer)

1000

800

600

400

200

0

–200

–400

–600

Pho

to-O

CT

(m

icro

met

er)

Mean

140,2

–1.96 SD

–501,7

+1.96 SD

782,1

Figure 1. Bland–Altman plot of manual OCT and planimetric

measurements of horizontal disc size. On paired t-testing the mean

horizontal measurement using planimetry being 140 lm larger as

compared to OCT was statistically significant (p ¼ 0.04).

1400 1500 1600 1700 1800 1900 2000 2100

Vertical measurementsAverage photo and OCT (micrometer)

600

400

200

0

-200

-400

-600

-800

Pho

to-O

CT

(m

icro

met

er)

Mean

-9,2

-1.96 SD

-545,8

+1.96 SD

527,4

Figure 2. Bland–Altman plot of manual OCT and planimetric

measurements of vertical disc size. On paired t-testing the mean

vertical measurement using planimetry being 9 lm smaller as

compared to OCT was not statistically significant (p ¼ 0.82).

OCT vs fundus photography for ONH: A. S. Neubauer et al. 15

ª 2006 The College of Optometrists

Results

Comparison of OCT and photographic planimetry

Mean refractive spherical equivalent of the subjectsincluded in the study was )1.2 D (±2.8 S.D.; range:)7.25 to +6 D). No correlation of refractive error tothe differences between OCT and planimetric measure-ments (r ¼ 0.08, p ¼ 0.7) existed for either horizontal orvertical measurements. Mean ± S.D. of disc size was1772 ± 317 lm vertically and 1492 ± 302 lm horizon-tally on OCT. On planimetry, values of 1763 ± 186 lmvertically and 1632 ± 160 lm horizontally wereobtained. Kolmogorov–Smirnov testing showed a nor-mal distribution for all data. On paired t-testing thedifferences between the two imaging modalities werenon-significant at p ¼ 0.82 vertically and weakly [butsignificantly (p ¼ 0.04)] different for horizontal meas-urements.

For further analysis Bland–Altman plots were pro-duced. Figure 1 shows the horizontal measurements andFigure 2 the vertical measurements. While verticalmeasurements did not differ (Figure 2), the horizontalmeasurements in Figure 1 showed a systematic differ-ence. The variation of differences is indicated in thefigures as ±1.96 S.D. of the measurements. It does notvary much between vertical (Figure 2) and horizontalmeasurements (Figure 1). However, for both directionssome cases with quite large differences exist. Those caseswith large absolute differences were further evaluatedfor possible reasons. In those subjects with too highOCT measurements the border of the optic disc wasfound difficult to determine mainly due to vessel trunks.In contrast, no problems were found in this study due toexisting myopic peripapillary atrophy. For the subjectswith relatively too low OCT measurements as comparedto planimetry, no specific reasons could be identified.

Repeatability

Results of repeat measurements with the Stratus OCTare listed in Table 1. The ICC were moderately goodreaching 0.78 horizontally and yielding 0.83 vertically.The repeated measures mainly fulfil the postulation of

Bland and Altman (1986) that the differences of the firstand last measurement should equal zero: a meandifference of )36 lm horizontally and )72 lm verticallywas observed. When calculating the coefficient ofrepeatability, reference intervals of 80 horizontally and168 lm vertically are calculated.

Discussion

In the current study we compare measurement of theONH diameter obtained by fundus photography tothose obtained by OCT. Repeatability of Stratus OCTautomatic measurements was moderately good. Overall,no significant difference of OCT to planimetric disc sizemeasurements was found for vertical measurements. Forhorizontal manual measurements a significant differenceexisted with planimetry yielding higher diameters thanOCT. In some cases considerable disagreement existed.Previous studies have determined the mean diameter ofthe optic disc histologically as 1.88 mm vertically and1.77 mm horizontally (Quigley et al., 1990). Applyingplanimetry a mean diameter of 1.92 ± 0.29 mm (S.D.)vertically and 1.76 ± 0.31 mm horizontally has beendetermined (Jonas et al., 1988b). Considering the lim-ited number of subjects this is well in agreement with ourresults for planimetric and OCT measurements.

Optical coherence tomography measurements tendedto be somewhat smaller than the results of planimetryfor the horizontal measurements. Of course, the opticdisc margin is defined differently for both methods:retinal pigment epithelium is referred to for OCTmeasurements, while the inner border of Elschnig’s ringis used for planimetry. However, this should theoretic-ally give larger measurements for OCT than for plani-metry as is observed with vertical measurements. Theopposite situation was observed in this study for thehorizontal measurements, which cannot be explained bythe different definition of the optic disc border. Incontrast, in cases with a tilted disc the different angle ofmeasuring may be responsible for higher OCT meas-urements: planimetry measures the observed planediameter, which may be biased and reduced by tiltingof the optic disc. Thus in cases of extreme tilting theplanimetric diameter may be very small. By contrast,

Table 1. Repeat automated measurements of optic disc with the Stratus OCT

Inter-class

correlation

coefficient (ICC)

Mean

size

(mm)

S.D.

(mm)

Mean difference

of measurement

1 vs 3 (mm)

S.D. of differences of

measurements (Bland

and Altman, 1986) (mm)

Reference interval (2 S.D. of

differences of measurements)

(Bland and Altman, 1986) (lm)

Horizontally 0.78 1.456 0.098 )0.036 0.040 80

Vertically 0.83 2.016 0.296 )0.072 0.084 168

Data of five normal subjects whose measurements were repeated three times within 1 day using the fast optic disc mode and dilated pupil. The

reference interval gives the amount of change which cannot be explained by instrument variation, but indicates real change.

16 Ophthal. Physiol. Opt. 2006 26: No. 1

ª 2006 The College of Optometrists

OCT determines the direct distance between the retinalpigment epithelium, which might theoretically be lessinfluenced by tilting of the optic disc. Another point tobe considered is the different method for size calculation(see Materials and methods section), which may alsocontribute to the effect of different sizes: in eyes withhigh axial length the error for the planimetric resultsobtained by the Bengtsson and Krakau (1992) spectaclecorrection formula may yield up to 10% larger meas-urements compared to constant correction as appliedhere for OCT (Garway-Heath et al., 1998). However, inour study no correlation was observed between refract-ive error and the differences between the imagingmethods and a different magnification factor due toaxial length cannot explain the observed differencesbetween horizontal and vertical measurements.Still, in several cases significant differences between

the two methods of assessing disc size exist as shownhere for normal subjects. This may question the useful-ness of either method, especially in cases of pathologicconditions such as glaucoma. On the other hand, thedifferences of the two methods represent differentaspects of the optic disc and can partly be correctedby proprietary reference intervals for each instrument.Still, this must be kept in mind when interpreting anymeasurements obtained. Peripapillary atrophy mayadditionally limit the use of the OCT method fordetermining the optic disc border in glaucoma patientsbased on retinal pigment epithelium. While this effect isknown to cause significant discrepancies in somepatients, it does not occur in all and seems to notgenerally impair the results in glaucoma patients (Laiet al., 2003).In normal subjects, our study could show that by

using OCT the optic disc size can be relatively accuratelydetermined manually, based on the posterior highlyreflective band. However, the differences to planimetrymay be relatively large between the two methods.Regarding the different versions of OCT – in particularVersion 2000 and the Stratus OCT – in contrast toretinal thickness measurements no significant differencesare known to exist for the ONH measurements obtained(Schuman et al., 2003). This justifies mixing the distancemeasurements of the two models in our study. However,the faster scanning protocol and improved software ofthe Stratus OCT allow us to theoretically determine thedisc margin without any user intervention. To assess thistheoretical advantage, repeatability was assessed in thisstudy. It showed high ICCs but still leaving considerablereference intervals of approximately 100 lm (Column 6/7 of Table 1), yielding 2.7–4.2% S.D. of the mean. Thismight be caused by a combination of instrumentvariability, software detection and small shifts of theposition of measurement due to fixation, as no manualcorrection to the software algorithm was performed in

this part of the study. The ICCs obtained in this studywere in the same range of 70–90% as obtained by adifferent group studying repeatability of different Stra-tus OCT scanning modes (Paunescu et al., 2004).However, this study found for the same fast-acquisitionscanning protocol much lower inter- and intra-visit S.D.of less than 1 lm [Table 6 of Paunescu et al., (2004)].This difference may only be partly explained by thespecial model they used, applying manual correctionsand age-adjustments. In the more clinical setting used inour study much higher variances were observed, whichare more consistent with those for other instrumentssuch as the HRT (where 3.3–6.0% S.D. of the mean discarea is common) and 4.0–7.2% for planimetry (Garway-Heath et al., 1999). The variance together with the factthat some cases encountered considerable deviationsfrom the planimetric results make it advisable tomanually inspect and correct the disc borders deter-mined by OCT software where necessary, even thoughthis may only in some cases yield significantly differentresults (Schuman et al., 2003), especially in cases withperipapillary atrophy (Lai et al., 2003).

We conclude that in normal subjects OCT can be usedto determine optic disc margins in moderate agreementwith planimetry. However, in some cases significantdisagreement with photographic assessment may occur,therefore manual inspection of all automatic scananalyses is recommended. Automatic disc detection byOCT is moderately repeatable.

Conflict of interest

The authors do not have any commercial interest in anyof the materials and methods used in this study.

Acknowledgements

The authors thank Mrs. Ehsani and Mrs. Guthmann forthe OCT measurements and Mrs. Lenggersdorfer andMrs. Merz for expert photography.

References

Bengtsson, B. and Krakau, C. E. (1992) Correction of opticdisc measurements on fundus photographs. Graefes Arch.

Clin. Exp. Ophthalmol. 230, 24–28.Bland, J. M. and Altman, D. G. (1986) Statistical methods forassessing agreement between two methods of clinical meas-urement. Lancet 1, 307–310.

Garway-Heath, D. F., Poinoosawmy, D., Wollstein, G.,Viswanathan, A., Kamal, D., Fontana, L. and Hitchings,R. A. (1999) Inter- and intra-observer variation in the

analysis of optic disc images: comparison of the Heidelbergretina tomograph and computer assisted planimetry. Br. J.Ophthalmol. 83, 664–669.

OCT vs fundus photography for ONH: A. S. Neubauer et al. 17

ª 2006 The College of Optometrists

Garway-Heath, D. F., Rudnicka, A. R., Lowe, T., Foster, P.

J., Fitzke, F. W. and Hitchings, R. A. (1998) Measurementof optic disc size: equivalence of methods to correct forocular magnification. Br. J. Ophthalmol. 82, 643–649.

Hee, M. R., Izatt, J. A., Swanson, E. A., Huang, D., Schuman,J. S., Lin, C. P., Puliafito, C. A. and Fujimoto, J. G. (1995)Optical coherence tomography of the human retina. Arch.

Ophthalmol. 113, 325–332.Huang, D., Swanson, E. A., Lin, C. P., Schuman, J. S.,Stinson, W. G., Chang, W., Hee, M. R., Flotte, T., Gregory,K., Puliafito, C. A. and Fujimoto, J. G. (1991) Optical

coherence tomography. Science 254, 1178–1181.Itai, N., Tanito, M. and Chihara, E. (2003) Comparison ofoptic disc topography measured by retinal thickness ana-

lyzer with measurement by Heidelberg retina tomograph II.Jpn. J. Ophthalmol. 47, 214–220.

Jonas, J. B., Budde, W. M. and Panda-Jonas, S. (1999)

Ophthalmoscopic evaluation of the optic nerve head. Surv.Ophthalmol. 43, 293–320.

Jonas, J. B., Gusek, G. C., Guggenmoos-Holzmann, I. andNaumann, G. O. (1988a) Variability of the real dimensions

of normal human optic discs. Graefes Arch. Clin. Exp.Ophthalmol. 226, 332–336.

Jonas, J. B., Gusek, G. C. and Naumann, G. O. (1988b) Optic

disc, cup and neuroretinal rim size, configuration andcorrelations in normal eyes. Invest. Ophthalmol. Vis. Sci.29, 1151–1158.

Koozekanani, D., Roberts, C., Katz, S. E. and Herderick, E.E. (2000) Intersession repeatability of macular thicknessmeasurements with the Humphrey 2000 OCT. Invest.

Ophthalmol. Vis. Sci. 41, 1486–1491.Lai, E., Wollstein, G., Price, L. L., Paunescu, L. A., Stark, P.C., Fujimoto, J. G. and Schuman, J. S. (2003) Optical

coherence tomography disc assessment in optic nerves with

peripapillary atrophy. Ophthalmic Surg. Lasers Imaging 34,

498–504.Littmann, H. (1992) Determination of the true size of an object

on the fundus of the living eye. By Littmann, H. from theoriginal article, �Zur Bestimmung der wahren Grosse einesObjektes auf dem Hintergrund des lebenden Auges�, whichoriginally appeared in Klinisches Monatsblatter fur Aug-enheilkunde 1982; 180, 286–289. Translated by Williams, T.D. Optom. Vis. Sci. 69, 717–720.

Muscat, S., Parks, S., Kemp, E. and Keating, D. (2002)

Repeatability and reproducibility of macular thicknessmeasurements with the Humphrey OCT system. Invest.Ophthalmol. Vis. Sci. 43, 490–495.

Paunescu, L. A., Schuman, J. S., Price, L. L., Stark, P. C.,Beaton, S., Ishikawa, H., Wollstein, G. and Fujimoto, J. G.(2004) Reproducibility of nerve fiber thickness, macular

thickness, and optic nerve head measurements using StratusOCT. Invest. Ophthalmol. Vis. Sci. 45, 1716–1724.

Quigley, H. A., Brown, A. E., Morrison, J. D. and Drance, S.M. (1990) The size and shape of the optic disc in normal

human eyes. Arch. Ophthalmol. 108, 51–57.Rudnicka, A. R., Frost, C., Owen, C. G. and Edgar, D. F.

(2001) Nonlinear behavior of certain optic nerve head

parameters and their determinants in normal subjects.Ophthalmology 108, 2358–2368.

Schuman, J. S., Wollstein, G., Farra, T., Hertzmark, E.,

Aydin, A., Fujimoto, J. G. and Paunescu, L. A. (2003)Comparison of optic nerve head measurements obtained byoptical coherence tomography and confocal scanning laser

ophthalmoscopy. Am. J. Ophthalmol. 135, 504–512.

18 Ophthal. Physiol. Opt. 2006 26: No. 1

ª 2006 The College of Optometrists