Technical Note

Manual corneal thickness measurements of healthy equineeyes using a portable spectral-domain optical coherencetomography deviceC. G. PIRIE*, A. F. ALARIO, C. M BARYSAUSKAS†, C. GRADIL‡ and C. K. URICCHIO§

Department of Clinical Sciences, Tufts Cummings School of Veterinary Medicine (TCSVM), North Grafton, Massachusetts, USA†Department of Biostatistics and Computational Biology, Dana Farber Cancer Institute, Boston, Massachusetts, USA‡Department of Veterinary and Animal Sciences, University of Massachusetts , Amherst, USA§Stockbridge School of Agriculture, University of Massachusetts , Amherst, USA.

*Correspondence email: chris.pirie@tufts.edu; Received: 20.08.13; Accepted: 05.10.13

Summary

Reasons for performing study: Corneal thickness measurements of the equine globe using spectral-domain optical coherence tomography (SD-OCT)have not been reported.Objectives: To determine corneal thickness measurements and the intra- and interoperator reliability of a portable SD-OCT device in equine eyes.Study design: Prospective observational study.Methods: Horses free of ocular disease were used for this study. Gentle manual restraint, in combination with detomidine hydrochloride and a head stand,were employed to ensure proper animal positioning. Corneal pachymetry measurements were obtained from both eyes of each animal 3 times by 2operators in succession. A 6 mm corneal pachymetry protocol was performed using a portable SD-OCT device. All measurements were obtained manuallyby one operator (C.G.P.) using the integrated calliper function. Measurements included epithelial thickness, stromal thickness, Descemet’s membranethickness and total corneal thickness. All recorded measurements were analysed to determine both intra- and interoperator reliability.Results: Thirty horses with a mean age of 10.6 ± 6.4 years were examined. Mean epithelial, stromal, Descemet’s membrane and total corneal thicknessvalues obtained were, respectively, 174.7 ± 13.6, 599.2 ± 45.4, 38.4 ± 15.3 and 812.0 ± 44.1 μm for operator A and 175.9 ± 12.9, 599.2 ± 44.9, 38.4 ± 15.0and 812.9 ± 42.9 μm for operator B. A positive correlation was found between Descemet’s membrane thickness and age, whereby Descemet’s membranethickness increased by 2 μm/year (P<0.0001). The coefficients of variation for both operators were <4% for all measurements. Intraclass correlations rangedfrom 0.92 to 0.98.Conclusions: Manual corneal thickness measurement using a portable SD-OCT device provides epithelial, stromal, Descemet’s membrane and totalcorneal thickness measurements with clinically acceptable intra- and interoperator reliability in healthy equine eyes.

Keywords: horse; optical coherence tomography; cornea; thickness; pachymetry

Introduction

Corneal pachymetry is the process of measuring corneal thickness andserves as an important diagnostic tool in the assessment of corneal health.Furthermore, corneal pachymetry may be useful in surgical planning andmonitoring of disease progression and/or response to therapy. Historically,various modalities have been employed to determine the cornealthickness of horses, including histological sectioning, optical andultrasound pachymetry and specular and confocal microscopy [1–6].Ultrasound pachymetry has been most commonly used, with previousreports citing a mean central corneal thickness of 770–893 μm [1–3,5,6].However, although easy to use, objective and repeatable, ultrasoundpachymetry does not allow for detailed assessment and/or measurementof specific corneal layers.

Relatively recently, a new imaging modality, optical coherencetomography (OCT), has been developed. Optical coherence tomography isan interferometric technique which provides depth-resolved cross-sectional structural information [7]. Optical coherence tomographymeasures the echo time delay and magnitude of back-reflected light by theanalysis of the interference fringe pattern generated [8]. Two approacheshave been developed, namely time-domain OCT and spectral-domain OCT(SD-OCT), with the latter offering more rapid scan acquisition times andhigher signal-to-noise ratios [9]. Comparable to ultrasound, OCT usesback-scattering or back-reflections of light within the structure of interest(i.e. cornea), as opposed to sound waves [10]. However, unlike ultrasound,OCT is a noncontact imaging modality. Optical coherence tomography hasbecome the predominant method by which corneal pachymetry isconducted in physician-based medicine and has been validated to bea highly reliable and repeatable modality for assessing both normaland diseased corneas in human subjects [11–16]. Optical coherence

tomography is gaining increasing popularity in veterinary medicine;however, its use to assess and measure the equine cornea has not yetbeen reported.

The purpose of this study was to obtain corneal thicknessmeasurements and to determine the intra- and interoperator reliability of aportable SD-OCT device in normal equine eyes.

Materials and methods

AnimalsThirty horses were provided by the Hadley Farm, University ofMassachusetts – Amherst for the purpose of this study. The project wasapproved by the University of Massachusetts – Amherst Committee onAnimal Care. Prior to enrolment, informed consent was obtained fromeach animal owner.

All horses were deemed to be free from significant ocular abnormalityand/or disease as assessed by complete ophthalmic examination.Ophthalmic examination included menace response, dazzle and pupillarylight reflexes, fluorescein staining (Ful-Glo)a, slit-lamp biomicroscopy (KowaSL-15 Portable Slit-Lamp Biomicroscope)b, applanation tonometry(Tono-Pen Vet)c and direct ophthalmoscopy (Welch Allyn directophthalmoscope)d.

Study protocolAll horses received detomidine hydrochloride (Dormosedan; 0.01–0.02 mg/kg i.v.)e approximately 5 min prior to imaging. Gentle manualrestraint, in combination with use of a head stand (Hoofjack headstand)f,was employed to ensure proper animal positioning. All horses were

bs_bs_banner

Equine Veterinary Journal ISSN 0425-1644DOI: 10.1111/evj.12198

631Equine Veterinary Journal 46 (2014) 631–634 © 2013 EVJ Ltd

positioned with their heads in a natural resting position, slightly above thelevel of their withers. A 6 mm corneal pachymetry protocol was performedusing a portable SD-OCT device (I-Vue)g and a supplemental Cornea-Anterior Module attachment. This protocol consists of 8 radial scans (1024A scans each), 6 mm in length. All scans were taken by aligning the aimingcircle at the centre of the pupil, ensuring that the SD-OCT device wastangential to the axial corneal surface. Scans were obtained from both eyesof each animal 3 times by the same 2 operators in succession. The SD-OCTunit was repositioned between each replicate performed by each operator.Only scans with an internal quality score of >27, centred on the cornealvertex and free of motion artefact, were accepted. Corneal measurementswere generated manually by one operator (C.G.P.) from each scan,using the calliper function integrated into the OCT software (Fig 1).Measurements included epithelial thickness, stromal thickness,Descemet’s membrane thickness and total corneal thickness. All recordedmeasurements were analysed to determine both inter- and intraoperatorreliability.

Data analysisThe data, including subject details (i.e. age, gender and breed), eye beingimaged, operator and corneal measurement, were entered into MicrosoftExcel 2010. Statistical analysis was performed using commercialsoftware (SAS version 9.3)h. A mixed-effects model was used to examinerelationships between corneal measurements and age, gender, breed,globe, replicate and observer. A P value <0.05 was considered to bestatistically significant. To test reliability, the coefficients of variation × 100(CV) and intraclass correlation coefficients (ICCs) were calculated.Additionally, Bland-Altman plots and the 95% limits of agreement forepithelial, stromal, Descemet’s membrane and total corneal thicknessmeasurements between operators were generated [17].

Results

Thirty horses (60 eyes) with a mean age of 10.6 ± 6.4 years (range 1– 22years) were evaluated in this study. The study population consisted of 3intact males, 16 castrated males and 11 intact females. Various breedswere scanned; however, Morgan horses predominated (n = 17). Theaverage time required to scan each horse, following sedation, wasapproximately 5 min per operator.

The overall mean (± s.d.) epithelial, stromal, Descemet’s membrane andtotal corneal thickness for all eyes examined was 175.3 ± 13.3, 599.2 ± 45.0,38.4 ± 15.1 and 812.4 ± 43.4 μm, respectively. No significant differencebetween epithelial, stromal and total corneal thickness measurementswere noted with age. However, a positive correlation was found betweenDescemet’s membrane thickness and age, whereby Descemet’smembrane thickness increased by 2 μm/year (P<0.0001). There was nosignificant difference in epithelial, stromal, Descemet’s membrane or totalcorneal thickness based on gender of the animal. Right eyes were found tohave slightly thicker epithelial thickness measurements than left eyes(2 μm; P<0.0001), whereas no significant difference in stromal, Descemet’smembrane or total corneal thickness measurements were noted based onthe eye examined. There was no significant difference in epithelial, stromal,

Descemet’s membrane or total corneal thickness between replicatemeasurements performed by the same operator; however, a small butsignificant difference was noted between operators for epithelial thicknessonly. The mean difference in epithelial thickness between operators was1.2 μm (P = 0.003). The mean epithelial, stromal, Descemet’s membraneand total corneal thickness for each operator, including the 95% confidenceinterval and CV, are listed in Table 1.

The ICC and CV comparing operators are presented in Table 2. TheBland-Altman plot and the limits of agreement for each cornealmeasurement were constructed using the collected data (Table 2 andFig 2). The constructed Bland-Altman plots demonstrate a randomdistribution of observations, with the majority falling between the limits ofagreement.

Discussion

This study provides normative data regarding the epithelial, stromal,Descemet’s membrane and total corneal thickness of horse eyes usinga portable SD-OCT device. Furthermore, the SD-OCT device used andthe manual measuring technique employed provide excellent intra-and interobserver reliability for stromal, Descemet’s membrane andtotal corneal thickness measurements. Variability of epithelialthickness measurements was noted between operators; however,clinically acceptable intra- and interobserver reliability was obtained.

The present study demonstrates a mean overall epithelial, stromal,Descemet’s membrane and total corneal thickness of 175.3, 599.2, 38.4and 812.4 μm, respectively. Accordingly, the epithelium, stroma andDescemet’s membrane account for approximately 21.6, 73.7 and 4.7%,respectively, of the total thickness of the equine cornea. The meanepithelial thickness measurements obtained herein were noted to be

A

B

C

D

Fig 1: Optical coherence tomography scan of a normal equine cornea. Note thedemarcation between the epithelium, stroma and Descemet’s membrane/endothelium. Calliper measurements are shown, where A = epithelial thickness (ET), B= stromal thickness (ST), C = Descemet’s membrane thickness (DT) and D = totalcorneal thickness (TCT).

TABLE 1: Mean corneal thickness measurements by operator; the 95%confidence interval (CI) and coefficient of variation (CV) for eachoperator are also shown

OperatorMean thickness(μm) ± s.d. 95% CI CV %

ET

Operator A 174.7 ± 13.6 172.6–176.7 1.3

Operator B 175.9 ± 12.9 174.0–177.8 1.3

ST

Operator A 599.2 ± 45.4 592.4–605.9 0.56

Operator B 599.2 ± 44.9 592.5–605.9 0.66

DT

Operator A 38.4 ± 15.3 36.1–40.6 3.2

Operator B 38.4 ± 15.0 36.1–40.6 3.1

TCT

Operator A 812.0 ± 44.1 805.4–818.6 0.4

Operator B 812.8 ± 42.9 806.5–819.2 0.4

Abbreviations: DT = Descemet’s membrane thickness; ET = epithelialthickness; ST = stromal thickness; TCT = total corneal thickness.

TABLE 2: Mean difference in corneal thickness measurementsbetween operators; the intraclass correlation coefficient (ICC),coefficient of variation (CV) and limits of agreement betweenoperators are also shown

Mean difference(μm) 1–2 ± s.d. ICC CV %

Limits of agreementWidth (lower–upper; μm)

ET -1.2 ± 4.4 0.92 4.9 17.2 (-9.8 to 7.4)

ST -0.1 ± 6.3 0.97 2.0 24.6 (-12.4 to 12.2)

DT -0.0 ± 1.6 0.95 8.0 6.1 (-3.1 to 3.1)

TCT -0.9 ± 4.7 0.98 1.1 18.3 (-10.0 to 8.3)

Abbreviations: DT = Descemet’s membrane thickness; ET = epithelialthickness; ST = stromal thickness; TCT = total corneal thickness.

Optical coherence tomography of the equine cornea C. G. Pirie et al.

632 Equine Veterinary Journal 46 (2014) 631–634 © 2013 EVJ Ltd

thicker than those in a recent publication [4]. Using confocal microscopy,Ledbetter and Scarlett reported an overall epithelial thickness of 131 μmin 10 anaesthetised horses [4]. Reported discrepancies may reflectdifferences between the imaging modality employed, the manner ofstabilising horses (i.e. sedation vs. anaesthesia) and/or the population ofhorses evaluated. Data regarding stromal and Descemet’s membranethickness of the equine cornea in vivo have not been previously reported.The relative percentage of the stromal thickness (i.e. 73.7%) wassubstantially lower than that commonly cited in the literature (i.e. 90%) [18].Mean total corneal thickness measurements of 812.4 μm as reported hereare within the range of values obtained from previous studies. Usingultrasound pachymetry, specular microscopy or confocal microscopy,mean corneal thickness values range between 770 and 893 μm [1–3,5,6].Albeit not directly comparable owing to varying modalities and theirrespective principles of establishing corneal thickness measurements,total corneal thickness measurements reported here are closest to thosereported by Ledbetter and Scarlett using a light-based imaging system (i.e.confocal microscopy) [4]. Previous reports employing ultrasoundpachymetry have used a preset velocity of 1640 m/s to determine totalcorneal thickness [1–3,5,6]. However, to our knowledge, there is currentlyno information pertaining to the true speed of sound as it passes throughthe equine cornea. As such, velocities used may result in over- orunderestimation of the true corneal thickness using this modality. Furtherstudies directly comparing these modalities in horses are needed toelucidate their agreement and any potential biases that may exist amongthem.

In the present study, no significant difference in epithelial, stromal ortotal corneal thickness was noted with age. However, a positive correlationwas found between Descemet’s membrane thickness and age, whereby

Descemet’s membrane thickness increased by 2 μm/year (P<0.0001).Regarding epithelial thickness, results of the present study contrast with aprior study, which reported a significant negative linear correlation withepithelial thickness and age [6]. Regarding total corneal thickness, resultsof the present report are consistent with most prior publications [2–6].However, Ramsey et al. reported a positive correlation with increasing ageand corneal thickness in the Rocky Mountain horse [1]. Such discrepanciesmay relate to population differences and/or the sample size of horsesevaluated. In the present report, no significant difference in epithelial,stromal, Descemet’s membrane or total corneal thickness was notedregarding gender. These findings are comparable to previous reportsassessing epithelial and total corneal thickness [1–6]. No significantdifferences were noted in stromal, Descemet’s membrane or total cornealthickness measurements based on the eye examined. However, right eyeswere noted to have slightly thicker epithelial thickness measurements thanleft eyes (2 μm; P<0.0001). Albeit statistically significant, this finding isconsidered by the authors to be clinically insignificant, because it is lessthan a single incremental difference of the calliper tool (4 μm) and theresolution of the SD-OCT imaging device itself (5 μm).

The SD-OCT system employed in this study demonstrated excellentintra- and interobserver reliability for manually measuring stromal,Descemet’s membrane and total corneal thickness. There was nostatistically significant difference between replicates for either operator,and the CV for each operator was below 4%. Comparing interobserverreliability, there was no significant difference between operatorsfor stromal, Descemet’s membrane or total corneal thickness, andmeasurements between operators demonstrated near-perfect agreement(ICC≥0.95). For epithelial thickness measurements, intraobserver reliabilitywas excellent. There was no statistically significant difference between

-20

-8

-6

-4

-2

-15

-10

-20

-30

-10-5

-20

-15

-10

-5

0

5

10

15

20

0

5

10

15

20

0

10

20

30

0

2

4

6

8

740 760 780 800 820 840 860 880 900

740 760 780 800 820 840 860 880 9000 10 20 30 40 50 60 70 80

510 530 550 570 590 610 630 650 670 690

Average TCT Operator 1 and 2 (µm) Average ST Operator 1 and 2 (µm)

Average TCT Operator 1 and 2 (µm)Average DT Operator 1 and 2 (µm)

Diff

eren

ce in

DT

Op

erat

or 1

–2 (µ

m)

Diff

eren

ce in

TC

T O

per

ator

1–2

(µm

)

Diff

eren

ce in

ST

Op

erat

or 1

–2 (µ

m)

Diff

eren

ce in

TC

T O

per

ator

1–2

(µm

)

a) b)

d)c)

Fig 2: Bland-Altman plot demonstrating equine epithelial thickness (a), stromal thickness (b), Descemet’s membrane thickness (c) and total corneal thickness (d) measurementdifferences between 2 operators. Mean difference (continuous line) and the upper and lower limits of agreement (dashed lines) are represented.

C. G. Pirie et al. Optical coherence tomography of the equine cornea

633Equine Veterinary Journal 46 (2014) 631–634 © 2013 EVJ Ltd

replicates for either operator, and the CV for each operator was below 2%.The interobserver reliability regarding epithelial thickness demonstrated asmall but statistically significant difference (1.2 μm) between operators;however, the ICC (0.92) was comparable to that obtained for stromal,Descemet’s membrane and total corneal thickness (0.95–0.98), and thenoted difference between operators was considered by the authors to beclinically insignificant. Factors which may have contributed to these resultspertaining to epithelial thickness measurements may have includedvariation in the location of measurements from the true corneal vertex,differences in the subjective delineation between corneal layers, and therelatively large incremental difference by which the calliper tool currentlyfunctions (4 μm). Further advancements in the measurement algorithmmay help to reduce this variability.

In summary, this report provides normative data regarding epithelial,stromal, Descemet’s membrane and total corneal thicknessmeasurements in healthy equine eyes using a portable SD-OCT device. Thedevice used demonstrated excellent intra- and interoperator repeatability,with clinically acceptable differences in measurements between operators.Optical coherence tomography offers cross-sectional detail and analysis ofthe cornea and its specific layers in vivo. It is a noncontact imagingmodality, in comparison to ultrasound pachymetry, and does not needgeneral anaesthesia, which is required for confocal microscopy.

Authors’ declaration of interests

No competing interests have been declared.

Ethical animal research

This study was approved by the University of Massachusetts – AmherstIACUC. Prior to enrolment, informed consent was obtained from eachanimal owner.

Source of funding

None; IVue SD-OCT unit on loan from Optovue.

Acknowledgements

The authors would like to thank the staff at the Hadley Farm, University ofMassachusetts – Amherst for their assistance with this study.

Authorship

C. G. Pirie and A. F. Alario contributed to study design and execution, dataanalysis and interpretation and preparation of the manuscript. C. M.Barysauskas contributed to study design and data analysis andinterpretation. C. Gradil and C. Uricchio contributed to study design andexecution. All authors approved the final version of the manuscript.

Manufacturers’ addressesaAkorn Inc., Buffalo Grove, Illinois, USA.bKowa Co., Ltd, Tokyo, Japan.cReichert, Depew, New York, USA.dWelch Allyn Distributors, Skaneateles Falls, New York, USA.ePfizer Inc., New York, New York, USA.

fSwissvet Veterinary Products, Knoxville, Tennessee, USA.gOptovue Inc., Freemont, California, USA.hSAS Institute, Cary, North Carolina, USA.

References

1. Ramsey, D.T., Hauptman, J.G. and Petersen-Jones, S.M. (1999) Cornealthickness, intraocular pressure, and optical corneal diameter in RockyMountain Horses with cornea globosa or clinically normal corneas. Am. J. Vet.Res. 60, 1317-1321.

2. Plummer, C.E., Ramsey, D.T. and Hauptman, J.G. (2003) Assessment of cornealthickness, intraocular pressure, optical corneal diameter, and axial globedimensions in Miniature Horses. Am. J. Vet. Res. 64, 661-665.

3. Andrew, S.E., Ramsey, D.T., Hauptman, J.G. and Brooks, D.E. (2001) Density ofcorneal endothelial cells and corneal thickness in eyes of euthanatized horses.Am. J. Vet. Res. 62, 479-482.

4. Ledbetter, E.C. and Scarlett, J.M. (2009) In vivo confocal microscopy of thenormal equine cornea and limbus. Vet. Ophthalmol. 12, 57-64.

5. Mochal, C.A., Miller, W.W., Cooley, A.J., Linford, R.L., Ryan, P.L. andRashmir-Raven, A.M. (2010) Ocular findings in Quarter Horses with hereditaryequine regional dermal asthenia. J. Am. Vet. Med. Assoc. 237, 304-310.

6. van der Woerdt, A., Gilger, B.C., Wilkie, D.A. and Strauch, S.M. (1995) Effect ofauriculopalpebral nerve block and intravenous administration of xylazine onintraocular pressure and corneal thickness in horses. Am. J. Vet. Res. 56,155-158.

7. Yaqoob, Z., Wu, J. and Yang, C. (2005) Spectral domain optical coherencetomography: a better OCT imaging strategy. BioTechniques 39, S6-S13.

8. Wojtkowski, M. (2010) High-speed optical coherence tomography: basics andapplications. Appl. Opt. 49, D30-D61.

9. Gabriele, M.L., Wollstein, G., Ishikawa, H., Kagemann, L., Xu, J., Folio, L.S. andSchuman, J.S. (2011) Optical coherence tomography: history, current status,and laboratory work. Invest. Ophthalmol. Vis. Sci. 52, 2425-2436.

10. Huang, D., Swanson, E.A., Lin, C.P., Schuman, J.S., Stinson, W.G., Chang, W.,Hee, M.R., Flotte, T., Gregory, K. and Puliafito, C. (1991) Optical coherencetomography. Science 254, 1178-1181.

11. Francoz, M., Karamoko, I., Baudouin, C. and Labbe, A. (2011) Ocular surfaceepithelial thickness evaluation with spectral-domain optical coherencetomography. Invest. Ophthalmol. Vis. Sci. 52, 9116-9123.

12. Prakash, G., Agarwal, A., Mazhari, A.I., Chari, M., Kumar, D.A., Kumar, G. andSingh, B. (2012) Reliability and reproducibility of assessment of cornealepithelial thickness by Fourier domain optical coherence tomography. Invest.Ophthalmol. Vis. Sci. 53, 2580-2585.

13. Li, Y., Tan, O., Brass, R., Weiss, J.L. and Huang, D. (2012) Corneal epithelialthickness mapping by Fourier-domain optical coherence tomography innormal and keratoconic eyes. Ophthalmology 119, 2425-2433.

14. Correa-Pérez, M.E., López-Miguel, A., Miranda-Anta, S., Iglesias-Cortiñas, D.,Alió, J.L. and Maldonado, M.J. (2012) Precision of high definitionspectral-domain optical coherence tomography for measuring central cornealthickness. Invest. Ophthalmol. Vis. Sci. 53, 1752-1757.

15. Haque, S., Jones, L. and Simpson, T. (2008) Thickness mapping of the corneaand epithelium using optical coherence tomography. Optom. Vis. Sci. 85,E963-E976.

16. Sin, S. and Simpson, T.L. (2006) The repeatability of corneal and cornealepithelial thickness measurements using optical coherence tomography.Optom. Vis. Sci. 83, 360-365.

17. Bunce, C. (2009) Correlation, agreement, and Bland–Altman analysis:statistical analysis of method comparison studies. Am. J. Ophthalmol. 148,4-6.

18. Clode, A.B. (2010) Diseases and surgery of the cornea. In: EquineOphthalmology, 2nd edn., Ed: B.C. Gilger, Elsevier Saunders, St Louis,Missouri. pp 181-266.

Optical coherence tomography of the equine cornea C. G. Pirie et al.

634 Equine Veterinary Journal 46 (2014) 631–634 © 2013 EVJ Ltd