repeatability and reproducibility of corneal thickness using soct copernicus hr

RESEARCH PAPER

Repeatability and reproducibility of corneal thickness usingSOCT Copernicus HR

Clin Exp Optom 2013; 96: 278–285 DOI:10.1111/cxo.12002

Silvia Vidal MScValentín Viqueira PhDDavid Mas PhDBegoña Domenech PhDDepartment de Optica, Universidad de Alicante, SanVicente del Raspeig, Alicante, SpainE-mail: [email protected]

Purpose: The aim of this study is to determine the reliability of corneal thickness measure-ments derived from SOCT Copernicus HR (Fourier domain OCT).Methods: Thirty healthy eyes of 30 subjects were evaluated. One eye of each patient waschosen randomly. Images were obtained of the central (up to 2.0 mm from the corneal apex)and paracentral (2.0 to 4.0 mm) cornea. We assessed corneal thickness (central andparacentral) and epithelium thickness. The intra-observer repeatability data were analysedusing the intra-class correlation coefficient (ICC) for a range of 95 per cent within-subjectstandard deviation (SW) and the within-subject coefficient of variation (CW). The level ofagreement by Bland–Altman analysis was also represented for the study of the reproducibil-ity between observers and agreement between methods of measurement (automatic versusmanual).Results: The mean value of the central corneal thickness (CCT) was 542.4 � 30.1 mm (SD).There was a high intra-observer agreement, finding the best result in the central sector withan intra-class correlation coefficient of 0.99, 95 per cent CI (0.989 to 0.997) and the worst, inthe minimum corneal thickness, with an intra-class correlation coefficient of 0.672, 95 percent CI (0.417 to 0.829). Reproducibility between observers was very high. The best result wasfound in the central sector thickness obtained both manually and automatically with anintra-class correlation coefficient of 0.990 in both cases and the worst result in the maximumcorneal thickness with an intra-class correlation coefficient of 0.827. The agreement betweenmeasurement methods was also very high with intra-class correlation coefficient greater than0.91. On the other hand the repeatability and reproducibility for epithelial measurementswas poor.Conclusion: Pachymetric mapping with SOCT Copernicus HR was found to be highlyrepeatable and reproducible. We found that the device lacks an appropriate ergonomicdesign as proper focusing of the laser beam onto the cornea for anterior segment scanningrequired that patients were positioned slightly farther away from the machine head-rest thanin the setup for retinal imaging.

Submitted: 2 March 2012Revised: 28 May 2012Accepted for publication: 14 August 2012

Key words: cornea, corneal thickness, Fourier domain, optical coherence tomography, pachymetry

The measurement of central (CCT) andparacentral (PCT) corneal thickness hasbecome a crucial test in ophthalmologicalpractice. Pre-operative planning for cornealrefractive surgery requires accurate estima-tion of the corneal thickness.1–5 Likewise,assessing the risk of glaucoma requires anaccurate measurement of CCT.1–7 Moreover,the analysis of corneal thickness in certaincorneal diseases4,5 in contact lens wearers isessential for monitoring any changes in thecornea.2

The most commonly used technique andcurrent ‘gold standard’ for CCT assessmentis a spot measurement by ultrasonic pachym-etry (USP).4,5,8 Although USP has advantages

such as low cost, simple use and portability,this method is quite operator-dependent.Corneal indentation, misalignment and vari-ations in placing the probe all influence themeasurement accuracy. Furthermore, ultra-sonic pachymetry is a contact method, whichcarries the risk for corneal epithelial damageand transmission of infections.3

Optical coherence tomography (OCT) isa non-contact imaging technique based onthe principles of low-coherence interferom-etry.1,9 Although the technique was initiallydesigned to study the posterior segment,it can produce images of the anteriorsegment with some minor modifications inthe system.10,11 Fourier domain OCT (FD-

OCT) has a higher speed, lower acquisitiontime and a higher signal-to-noise ratio com-pared to conventional time domain OCT(TD-OCT).1,8

Several authors have confirmed thesefindings with precision studies of OCT withdifferent devices used in clinical practiceand the results show an excellent reliabilityfor pachymetric measurements.2,3,8,12,13

The purpose of this study was to evaluatethe precision (repeatability and reproduc-ibility) of SOCT Copernicus HR (FD-OCT)system of pachymetric mapping over a4.0 mm diameter.

As the OCT permits visualisation of differ-ent layers in the cornea, in addition to

C L I N I C A L A N D E X P E R I M E N T A L

OPTOMETRY

Clinical and Experimental Optometry 96.3 May 2013 © 2013 The Authors

278 Clinical and Experimental Optometry © 2013 Optometrists Association Australia

analysis of total corneal thickness, we alsodecided to include in this study the reliabilityof corneal epithelial thickness measure-ments. Regarding this variable, we foundonly one paper in the literature about thisissue4 and the study shows poor repeatability.Therefore, we will try to confirm or discardthis result.

To the best of our knowledge, no study hasperformed the repeatability and reproduc-ibility of this system on complete cornealthickness evaluation.

MATERIALS AND METHODS

The device used in this study was the SOCTCopernicus HR (Optopol SA, Poland),designed and built at the Physics Instituteof University of Nicolaus Copernicus inTorun.14 It is a non-invasive tool that uses a‘super luminescent diode’ (SLED) with acentral wavelength of 855 nm. The gener-ated interference signal is detected by a spec-trometer that provides an axial resolution of3.0 mm in tissue, with a scan speed of 52,000A-scans per second and a transversal resolu-tion of 12 to 18 mm. The instrument has anA-Scan resolution of 1,024 points and aB-Scan resolution of 20,000 A-scans. The

focus and alignment of the instrument’sscanning head is adjusted with an auto-mated, motorised system controlled fromthe computer screen.15

SubjectsA total of 30 patients participated in thisstudy, of whom eight were men. One eye ofeach patient was randomly selected. Thesubjects were volunteers from the studentpopulation at the University of Alicante.Excluded from the study were those whohad a history of corneal refractive surgery,corneal abnormalities or at the time ofexploration, were wearing their contactlenses. After being informed of their inclu-sion in the study, all patients signed aninformed consent document in accordancewith the Helsinki Declaration. All measure-ments were carried out without applicationof lubricating eye drops or pupil dilation.

Setup analysisFigure 1 shows the pachymetric map gener-ated over a 4.0 mm diameter circle by SOCTCopernicus HR. In it we can distinguish:1. a preview of the eye examination, which

selects the directional image, where thecorneal reflex appears brighter

2. a pachymetric map of sectors, generatedautomatically and divided into two zones:central (zero to 2.0 mm) and paracentral(2.0 to 4.0 mm), which were subdividedinto eight equally spaced radial sectors.

The pachymetric map appears doubledas shown in Figure 1. The one shown tothe left of the output screen is overlaidwith a pseudo-coloured map for fasterinterpretation.

Another application of interest to ourstudy is the possibility of manual measure-ment of corneal thickness and corneal epi-thelial thickness on the tomographic imageusing the SOCT Copernicus HR instru-ment’s software (Figure 1).

ProcedureFirst, the patient’s head was well positionedon the chin and forehead rest and thepatient was asked to fixate on the centrallight. The anterior segment imaging attach-ment was used for the scans in order forthe instrument to focus effectively on thecornea. To focus the laser beam onto thecornea, we found that patients had to bepositioned slightly farther away from themachine head than in the setup for retinalimaging.

The instrument has two fixation modes,internal and external. Internal fixation isthe most reproducible and therefore is themethod most used. External fixation is indi-cated when the visual acuity of the eye to beexamined is too poor to provide stable fixa-tion (for example, eccentric fixation). Inthe present study, all patients were exam-ined with internal fixation. The anteriorsegment was scanned using the instru-ment’s asterisk anterior scanning mode.The default pattern has 15 lines with alength of 7.0 mm each (Figure 2). Thecross-sections were centred by the examinerto maximise the corneal vertex reflection.The corneal reflex aids in obtaining consist-ent alignment between scans.1,3,8,16 Thisposition ensures that the scans are perpen-dicular to the corneal vertex and determinethe tomographic section where layers arebetter defined.

We assessed the central corneal thicknessautomatically using the pachymetric map,which is divided in a central area andeight paracentral sectors (Figure 1). Cor-neal thickness for each sector was generatedautomatically by the instrument’s softwareby interpolation and presented values ofmaximum, minimum and average thickness.

Figure 1. Detail of the output window of the corneal analysis software SOCT CopernicusHR. Tomographic image shows corneal layers (deriving corneal epithelial thickness andcentral corneal thickness) and vertex reflection

Corneal thickness using SOCT Copernicus HR Vidal, Viqueira, Mas and Domenech

© 2013 The Authors Clinical and Experimental Optometry 96.3 May 2013

Clinical and Experimental Optometry © 2013 Optometrists Association Australia 279

Corneal thickness was also measuredmanually, using the distance measurementtool on the scan section of the tomo-gram, which includes the vertex reflection(Figure 1). The corneal epithelial thicknesswas assessed solely in manual mode and onthis image.

To determine repeatability and reproduc-ibility, three scans were performed by exam-iner A and one further scan by examiner B atthe same session.

Statistical analysisAll corneal thickness data were statisticallyanalysed using SPSS (version 18, SPSSInc, Chicago, IL, USA) and Med Calc (ver-sion 11.6.1. MedCalc Software, Mariakerke,Belgium). The normality of all distributionsof parametric data was confirmed with theKolmogorov–Smirnov test. All values pre-sented with p-values greater than 0.05, indi-cating that data are normally distributed andtherefore it is appropriate to use parametricstatistical tests.

The first part of the study investigates theintra-observer reproducibility of cornealthickness measurements obtained with theSOCT Copernicus HR. All tests were per-formed by the same examiner. Althoughthree measurements of each subject wereinitially taken, only the first two wereevaluated for the repeatability study, whilethe third was used for a later experiment.The mean and standard deviation (SD),mean difference and SD, within-subject SD(Sw) of two consecutive measurements andwithin-subject coefficient of variation (CVW)

defined as 100 ¥ Sw / overall mean, wereobtained. We also examined the intra-classcorrelation coefficient (ICC) for a confi-dence interval (CI) of 95 per cent, based onthe analysis of the variance of a randomeffects model of two factors. The intra-classcorrelation coefficient will approach 1.0,when there is no variance within repeatedmeasurements, indicating that the total vari-ation in measurements is solely the result ofthe variability in the parameter being meas-ured.17 The maximum value of the intra-classcorrelation coefficient is 1 and the minimumvalue is zero and according to the classifica-tion proposed by Fermanian, concordance isexcellent for ICC greater than 0.91, good forICC ranging between 0.90 to 0.71, moderatefor ICC ranging between 0.70 to 0.51, fair forICC between 0.50-0.31 and bad for ICC lessthan 0.30.

In the second part of the study, we inves-tigate inter-observer reproducibility. Weused the first measurement of examiner Aand the measurement made by examiner B.Mean and SD, mean difference and SD, Sw,CVW and ICC were obtained. The differencesbetween the two observers and limits ofagreement were represented by the Bland–Altman method.

Finally, in the latter part of the study, weassessed the correlation between differentmethods of measurement of the cornealthickness in the central sector. We calcu-lated the same statistical parameters in thecases of repeatability and reproducibility.This is the only parameter that has beenmeasured both automatically and manually

and thus is the only one that we show in thisanalysis. We also represented the differencesbetween the two measurement methods andlimits of agreement by the Bland–Altmanmethod.

RESULTS

Central corneal thicknessThe study evaluated 30 eyes of 30 patients, ofwhom eight were men. The mean age of themale patients was 21.50 � 2.07 years (SD)(range: 19 to 25 years) and of the femalepatients was 24.77 � 5.27 years (range: 19 to38 years). The mean age of the total 30patients was 24.33 � 5.46 years (range: 18 to38 years). The mean refractive error (spheri-cal equivalent) was -0.88 D � 2.14 (range:-8.00 to +4.00 D). Of the 30 patients studied,12 were emmetropes, 13 myopes (meanrefractive error of -2.56 D � 2.07, range:-8.00 to -0.50 D) and 12 hyperopes (meanrefractive error of +1.40 D � 1.52, range:+0.50 to +4.00 D).

For the 30 eyes, the mean CCT was 542.4 �

30.1 mm.

The repeatability of themeasurement of corneal thicknessTable 1 shows the statistical results obtainedin the intra-observer repeatability of cornealthickness in the nine sectors as well asthe maximum, minimum and manualmeasurement of corneal thickness in thevertex and the epithelial thickness. In allsectors, including manual measurements,within-observer concordance of the meanthickness, according to the classificationproposed by Fermanian18 was noted as verygood (greater than 0.91). The concordancevalues for maximum and minimum thick-nesses are significantly lower with intra-classcorrelation coefficient values of 0.672 and0.715, respectively. On the other hand, thesetwo variables also present higher values forSw and CVW, indicating greater variability, asreflected by the expansion of the confidenceintervals for the ICC. The corneal epithelialthickness ICC was fair according to the samecriterion.

Table 2 shows an analysis of whetherexpanding the number of measurementsto three gives more consistent results formaximum and minimum thicknesses. Therewas little variation in the ICC and CI butvariability was increased when analyzing Sw

and CVW.

Figure 2. Simulation of the scan pattern of the mapping of the cornea




Reproducibility in themeasurement of corneal thicknessTable 3 summarises the results obtained inthe evaluation of agreement between exam-iners. In all sectors, we find a very goodagreement (with ICC values greater than0.91). The best results are found in thecentral (automatic and manual) measure-ment. Although the results obtained inthe maximum and minimum thicknessesare good, when analysing the degree ofagreement using the Bland–Altman plots(Figure 3), we find that in both cases there istoo much variability in practice. Althoughthe measured differences between observersare not too large (5.2 mm and 1.6 mm), whencalculating the range of agreement, values of38 and 53 mm are obtained, which are higher

than five per cent of the average pachymetricreading, thus indicating clinically relevantdifferences. Figure 4 shows the Bland–Altman plot for the epithelium. In this case,although the absolute value of the meandifference is only 0.5 mm, the range of agree-ment is of the order of 10 per cent of theaverage epithelial thickness, which impliesclinically relevant differences.

Agreement between methods ofmeasurement of CCTTable 4 shows the results obtained by analys-ing the agreement between methods ofmeasurement (automatic versus manual)obtaining a very good agreement with anICC of 0.987, 95 per cent CI (0.972 to 0.994).The Bland–Altman plot (Figure 5) shows

good agreement between both methods.The automatic measurement underesti-mates CCT by 7.2 mm compared to the valueobtained with the manual method. It alsogives a range of 9.3 mm agreement (two percent on average), indicating that they arenot clinically relevant differences.

DISCUSSION

Any study should guarantee the quality ofthe measurements, not only because itlargely determines the validity of the conclu-sions but also the importance of clinical deci-sions that are based on that research.19 Thereare numerous studies that evaluate therepeatability and reproducibility of OCT inthe posterior segment but fewer in the ante-rior segment. To our knowledge, this is thefirst study that evaluates the repeatabilityand reproducibility of corneal thicknessmeasurements with the SOCT CopernicusHR.

The CCT measured with OCT is compara-ble to results obtained with ultrasonicpachymetry, as shown by the study of Muscatand colleagues.2 They found that measure-ments obtained with OCT were lower thanthose obtained by ultrasonic pachymetry (toa value of approximately 50 mm) but theICC between the two methods was very high.Li and colleagues20 also compared the time

Mean (mm) � SD Mean difference (mm) � SD Sw (mm) CVw (%) ICC (95 per cent CI)

Sector S 546.0 � 31.4 3.3 � 8.0 4.4 0.82 0.968 (0.934–0.985)Sector ST 545.6 � 31.3 1.8 � 9.6 4.6 0.82 0.954 (0.905–0.978)Sector T 541.1 � 31.5 2.6 � 8.3 4.3 0.80 0.966 (0.930–0.984)Sector IT 538.9 � 30.5 0.0 � 6.7 3.4 0.62 0.976 (0.950–0.989)Sector I 537.4 � 30.6 0.6 � 8.1 3.9 0.71 0.942 (0.882–0.972)Sector IN 535.2 � 30.7 1.1 � 9.1 4.1 0.76 0.958 (0.914–0.980)Sector N 536.9 � 31.9 3.7 � 12.2 4.1 0.77 0.929 (0.857–0.966)Sector SN 541.7 � 30.5 2.7 � 8.8 4.1 0.75 0.959 (0.916–0.980)Central 543.5 � 29.9 0.5 � 3.1 1.7 0.33 0.995 (0.989–0.997)Vertex manual 550.5 � 29.1 1.0 � 4.7 2.7 0.49 0.973 (0.973–0.994)Minimum 503.5 � 28.0 6.6 � 24.8 12.0 2.40 0.672 (0.417–0.829)Maximum 590.5 � 41.5 4.3 � 23.3 13.3 2.23 0.854 (0.715–0.928)Epithelium 50.0 � 1.2 0.4 � 1.6 0.8 1.65 0.379 (0.028–0.647)

S: superior, ST: supero-temporal, T: temporal, IT: infero-temporal, I: inferior, IN: infero-nasal, N: nasal, SN: supero-nasal,Sw: within-subject SD, CVw: within-subject coefficient of variation, ICC: intra-class correlation coefficient, CI: confidence interval

Table 1. Summary of the results obtained by analyzing the intra-observer repeatability of corneal thicknessmeasurement with SOCT Copernicus HR (n = 30) taking the first and second measurement obtained by the examiner

Mean (mm) � SD Sw (mm) CVw (%) ICC (95 per cent CI)

Minimum 501.2 � 31.1 16.4 3.34 0.600 (0.400–0.765)Maximum 590.1 � 39.1 14.1 2.37 0.829 (0.713–0.908)

Sw: within-subject SD, CVw: within-subject coefficient of variation, ICC: intra-class correlationcoefficient, CI: confidence interval

Table 2. Summary of the results obtained by analysing the intra-observer repeatability ofcorneal thickness measurement with SOCT Copernicus HR (n = 30) taking the threemeasurements obtained by the examiner




domain Visante OCT (Carl Zeiss Meditec)with ultrasonic pachymetry getting anunder-estimation of the CCT of 14.6 mm.Muscat and colleagues2 also analysed the

correlation in a group of patients withcorneal oedema, obtaining a high correla-tion by OCT (Humphrey Zeiss, timedomain). On the other hand, Kim and

colleagues4 obtained good reproducibilitybetween the measurements obtained by theSL-OCT (Heidelberg Engineering, timedomain) and ultrasound but it is important

460

-15

-10

-5

0

5

10

15

Mean CCT (A and B examiner) (µm)

Inte

rexa

min

er D

iffer

ence

(µm

)

Mean

2,1

-1.96 SD-6,4

+1.96 SD10,6

480 520 540 560 580 600 620500

Figure 3. Bland–Altman charts for inter-observer repeatability ofcorneal thickness in the centre (CCT) and the manual, minimumand maximum. The 95 per cent limits of agreement are shown withdashed lines and the solid line represents the mean of the differ-ences between these measurements (n = 30).

6

4

2

0

-2

-4

-647 48 49 50 51 52 53 54

Mean Corneal Epithelium Thickness (A and B examiner) (µm)

Inte

rexa

min

er D

iffer

ence

(µm

)

+1.96 SD

4.9

-1.96 SD

-3.9

Mean

0.5

Figure 4. Bland–Altman charts for inter-observer repeatability ofcorneal epithelial thickness. The 95 per cent limits of agreementare shown with dashed lines and the solid line represents the meanof the differences between these measurements (n = 30).


Sector S 546.2 � 31.7 2.9 � 12.0 6.0 1.10 0.931 (0.860–0.967)Sector ST 544.6 � 31.3 3.8 � 9.5 5.1 0.94 0.955 (0.907–0.978)Sector T 539.9 � 32.0 5.1 � 6.9 4.6 0.86 0.977 (0.952–0.989)Sector IT 537.5 � 30.7 3.0 � 7.9 4.3 0.78 0.967 (0.932–0.984)Sector I 536.6 � 30.3 2.3 � 9.2 4.7 0.87 0.955 (0.908–0.978)Sector IN 534.9 � 31.3 1.6 � 8.1 4.1 0.75 0.974 (0.945–0.987)Sector N 536.6 � 33.1 4.3 � 6.2 4.0 0.75 0.983 (0.964–0.992)Sector SN 541.2 � 31.2 3.7 � 10.5 5.7 1.06 0.945 (0.887–0.973)Central 542.7 � 29.9 2.1 � 4.3 2.7 0.50 0.990 (0.978–0.995)Vertex manual 551.3 � 29.9 -0.8 � 4.3 2.7 0.49 0.990 (0.978–0.995)Minimum 504.1 � 31.8 5.2 � 19.6 8.8 1.98 0.827 (0.667–0.914)Maximum 593.5 � 41.0 -1.6 � 27.0 15.0 2.52 0.805 (0.629–0.902)Epithelium 50.0 � 1.7 0.5 � 2.2 1.1 2.22 0.391 (0.042–0.655)

S: superior, ST: supero-temporal, T: temporal, IT: infero-temporal, I: inferior, IN: infero-nasal, N: nasal, SN: supero-nasal, Sw:within-subject SD, CVw: within-subject coefficient of variation, ICC: intra-class correlation coefficient, CI: confidence interval

Table 3. Summary of the results obtained by analysing inter-observer reproducibility in measuring corneal thicknesswith the SOCT Copernicus HR (n = 30)




to note that in clinical practice, the measure-ments captured by these two methods arenot directly interchangeable. Huang andcolleagues8 compared the results obtainedby a Scheimpflug imaging system (Penta-cam) with the RTVue Fourier domain OCT(Optovue) showing that CCT measurementsare reliable and interchangeable in corneasafter laser in situ keratomileusis (LASIK),which is not the case with those obtainedwith ultrasonic pachymetry. This impliesthat the OCT can be used immediately aftercorneal surgery and can be a useful tool forevaluating the surgical success.

Fukuda and colleagues16 also expandedtheir study to other anterior segment bio-metric measurements such as anteriorchamber depth (ACD) or angle-angle dis-tance (ATA), finding a good repeatabilityand reproducibility.

Sin and Simpson13 found a very goodrepeatability of the OCT (RTVue) in themeasurement of CCT but not in the meas-

urements of the epithelium, demonstratingthe importance of optimising each scan andthe convenience of taking a high number ofscans to maximise the consistency of themeasurements. Mohamed and colleagues12

found that the pachymetric map obtainedwith the OCT (Visante) for the mean centraland peripheral thicknesses are reproduciblein both healthy subjects and those withkeratoconus.

The comparison between the two OCTsystems, FD-OCT and TD-OCT, has alsobeen studied by Prakash and colleagues1 andHuang and colleagues3 concluding that,although both OCT instruments showedgood reliability, the FD-OCT was better.This may be due to a faster image capture,thus reducing testing time and therefore,decreasing the effect of eye movements.

In this study, we obtained a good agree-ment between observer, intra-observer andbetween methods, of CCT measurement.We demonstrated that a reliable measure-

ment of CCT can be obtained without manyacquisitions as shown in Tables 1 and 2.

The results are better in terms of repeat-ability in the central area (both automaticand the manual measurements), with anICC for both sectors higher than 0.91, indi-cating a very good match. These results arecomparable to those obtained by Prakashand colleagues1, Huang and colleagues3 andMohamed and colleagues.12 In our study,we found that the least repeatable andreproducibile results were obtained for themaximum and minimum corneal thicknessmap values with an ICC ranging from 0.805and 0.854, corresponding to a good match.The exception is found in the repeatabilityof the minimum value with an ICC of 0.672(moderate agreement). This result supportsthe results of Prakash and colleagues,1 inwhich they find that the minimum cornealthickness repeatability was the worst param-eter analysed by the TD-OCT but the best forthe FD-OCT.

The reason for this difference in repeat-ability values may come from the origin ofthe data. While central (automatic) andvertex (manual) thicknesses come froman average calculation, the maximum andminimum values correspond to a singlevalue obtained from the whole area. In addi-tion, sampling becomes less dense as it goesfurther from the centre. This may alsoincrease the variability of the maximum andminimum thicknesses, particularly if thesepoints are outside the central region.

Our average CCT was 542.4 � 30.1 mm,which is comparable to the results obtainedby different authors using FD-OCT andAS-OCT.1,3,11,16,20

Table 5 summarises the most relevantstudies of repeatability and reproducibilityof corneal thickness measured with OCT.

In their study Huang and colleagues3 ana-lysed the central and paracentral cornea(two to five millimetres), which is subdividedinto eight sectors, with two OCT devicesfrom different domains. In both cases, the


Central vertex manual 547.3 � 29.2 -7.2 � 4.8 5.5 1.02 0.987 (0.972–0.994)

Sw: within-subject SD, CVw: within-subject coefficient of variation, ICC: intra-class correlation coefficient, CI: confidence interval

Table 4. Summary of the results obtained by analysing the agreement between methods of measurement (automaticversus manual) in corneal thickness measurements with the Copernicus SOCT HR (n = 30)

460 480 500 520 540 560 580 600 620

-20

-15

-10

-5

0

5

10

Mean CCT (automatic and manual measurements) (µm)

Inte

rmea

sure

men

t Diff

eren

ce (

µm)

Mean

-7,2

-1.96 SD

-16,6

+1.96 SD

2,1

Figure 5. Bland–Altman chart agreement between measurementmethods (manual versus automatic). The 95 per cent limits of agree-ment are shown with dashed lines and the solid line represents themean of the differences between these measurements (n = 30).




References Eyes (patients)[age range, years]

Repeatability Reproducibility Device used

ICC (95 per cent CI) 95 per cent LoA(mm)

ICC(95 per cent CI)

95 per cent LoA(mm)

Sin andSimpson13

32 (18)[15–53]

0.970d (—c) —c —c —c Humphrey Zeiss OCT0.980e (—c)

Fukuda andcolleagues16

85 (85)[22–89]

0.999a -12.0–10.1 0.998b -12.0–10.1 CAS-OCT (3Dimages. prototype)0.997b

0.998a -12.0–10.1 0.987b -12.0–10.1 AS-OCT (Visante.Carl Zeiss Meditec)0.968b

Mohamedandcolleagues12

27 (27)[18–71]

0.998 (0.995–0.999)h

0.996 (0.991–0.998)i

—c 0.995(0.988–0.998)

0.993(0.984–0.997)

—c AS-OCT (Visante.Carl Zeiss Meditec)

Muscat andcolleagues2

(14)[21–58]

0.998f (—c) -3–4 Humphrey Zeiss OCT0.979g (—c)

Prakash andcolleagues1

(100)[23.3 � 2.4]

0.962 (0.945–0.975)j -20.37–17.59 —c —c AS-OCT (Visante.Carl Zeiss Meditec)0.949 (0.926–0.966k -20.55–18.11

0.954 (0.948–0.960)l -23.3–19.900.999 (0.998–0.999)j -8.33–8.15 —c —c RTVue-OCT0.999 (0.998–0.999)k -3.33–3.660.995 (0.994–0.996)l -7.35–6.32

Huang andcolleagues3

72 (72)[44–86]

0.994 (0.991–0.996)j -11.97–9.71 —c —c RTVue-OCT0.973 (0.958–0.983)m -30.86–20.88 —c —c

0.978 (0.965–0.986)n -28.25–22.750.964 (0.943–0.977)o -20.17–20.330.978 (0.965–0.986)p -28.30–15.010.976 (0.963–0.985)q -28.26–11.060.972 (0.956–0.983)r -22.88–11.260.974 (0.959–0.984)s -18.01–8.250.966 (0.947–0.979)t -17.04–13.960.989 (0.982–0.993)j -11.97–9.71 —c —c AS-OCT (Visante.

Carl Zeiss Meditec)0.936 (0.899–0.959)m -30.86–20.88 —c —c

0.947 (0.917–0.967)n -28.25–22.750.951 (0.922–0.969)o -20.17–20.330.960 (0.936–0.974)p -28.30–15.010.977 (0.964–0.986)q -28.26–11.060.980 (0.968–0.988)r -22.88–11.260.980 (0.968–0.987)s -18.01–8.250.980 (0.968–0.987)t -17.04–13.96

a Repeatability of measurements on the same day and same observer (n = 10); b Repeatability of measurements different day and same observer (n = 30);c The information was not provided within the article; f Inter-observer reproducibility; g Inter-session reproducibility; h Value obtained in CCT 0–2 mm;i Value obtained in CT 2–5 mm; j Value obtained in CCT; k Value obtained in CTMinimum; l Value obtained in CT Maximun; m Value obtained in CT 2–5 mm Supero-nasal;n Value obtained in CT 2–5 mm Superior; o Value obtained in CT 2–5 mm Supero-temporal; p Value obtained in CT 2–5 mm Nasal; q Value obtained in CT 2–5 mm Infero-nasal; r Valueobtained in CT 2–5 mm Inferior; s Value obtained in CT 2–5 mm Infero-temporal; t Value obtained in CT 2–5 mm Temporal; 95 per cent LoA: 95 per cent limits of agreement definedas the mean � 1.96 standard deviation SD, ICC: intra-class correlation coefficient, CI: confidence interval.

Table 5. Summary of previous studies of repeatability and reproducibility in the measurement of corneal thickness and the main resultsobtained with different OCT devices




most repeatable results are in the centralsector, as in our study; however, we found adifference of opinion about which sectorsare the least repeatable. When using theFD-OCT (RTVue), Huang and colleagues3

noticed that the less repeatable are thesuperior-temporal and inferior-temporalsectors, whereas when using the AS-OCT(Visante), the superior-nasal and superior-temporal sectors are the least repeatable.In our study (regardless of the values ofmaximum and minimum corneal thickness,which were analysed by Huang and col-leagues3), it follows that the least repeatableare nasal, inferior and superior-temporalsectors but no relationship between themwas found.

When analysing the intra-observer repro-ducibility, we observed an overestimation ofthe first measurement with respect to thesecond (Table 1). In the minimum andmaximum sectors, we made a comparativestudy between the first and second measure-ment and the average of the three measurestaken by the examiner. This study aimed todemonstrate whether improved concord-ance is obtained for these variables based onthree steps instead of two (Table 2). Theresults show that it is not necessary to takethree steps to improve consistency.

In different studies, Prakash and col-leagues1 and Huang and colleagues3 foundan over-estimation of the corneal thicknessmeasurement taken manually in referenceto the automatic. Our study also reflectedthat over-estimation.

In summary, although the results showgood reproducibility, with ICC greater than0.91 except for the minimum and maxi-mum corneal thickness, the Bland–Altmanmethod in these two sectors (Figure 3)shows 95 per cent limits of agreementranging from -33.2 to 43.7 mm for theminimum value and -54.5 to 51.2 mm forthe maximum value, which makes us thinkthere is too much variability in the clinicalmeasurements.

Axial resolution of the OCT is in partresponsible of the poor reliability obtainedin corneal epithelial thickness measure-ments. The accuracy of 3.0 mm is above thefive per cent of the total value of the epithe-lium thickness (approximately 50 mm). Thislimitation together with the experimentalerrors explains the poor results.

Our study has some limitations. First, allpatients who participated in this study hadhealthy corneas. Additional studies areneeded to see whether the results obtained

in patients with normal corneas can beextended to patients following refractivesurgery, such as LASIK or showing cornealdisorders, such as keratoconus or cornealopacities. The changes of curvature of thecornea in these patients may make it difficultto find the apex corneal reflex as suggestedby Huang and colleagues.3

A second limitation is the difficulty intaking images because the SOCT Coperni-cus HR chinrest did not fit ergonomically tothe facial features of patients. Also, for acorrect approach to the cornea, the patienthad to place the head further away from themachine than the configuration of thedevice for the retina scan. The study byMuscat and colleagues2 also refers to thislimitation.

The quality of a measuring instrument isreflected in the reliability and validity of itsmeasurements. In this study, we demon-strate the repeatability and reproducibilityof the SOCT Copernicus HR. Therefore,further analyses of its validity are requiredcomparing the results with those obtainedusing a reference test (gold standard) that isa valid and reliable measurement.

REFERENCES1. Prakash G, Agarwal A, Jacob S, Kumar DA, Agarwal

A, Banerjee R. Comparison of Fourier-domain andtime-domain optical coherence tomography forassessment of corneal thickness and intersessionrepeatability. Am J Ophthalmol 2009; 148: 282–290.

2. Muscat S, Mckay N, Parks S, Kemp E, Keating D.Repeatability and reproducibility of corneal thick-ness measurements by optical coherence tomogra-phy. Invest Ophthalmol Vis Sci 2002; 43: 1791–1795.

3. Huang JY, Pekmezci M, Yaplee S, Li S. Intra-examiner repeatability and agreement of cornealpachymetry map meaurement by time-domain andFourier-domain optical coherence tomography.Graefes Arch Clin Exp Ophthalmol 2010; 248: 1647–1656.

4. Kim HY, Budenz DL, Lee PS, Feuer WJ, Barton K.Comparison of central corneal thickness usinganterior segment optical coherence tomography vsultrasound pachymetry. Ophthalmology 2008; 145:228–232.

5. Doors M, Berendshot T, Brabander J, Webers C,Nuijts R. Value of optical coherence tomographyfor anterior segment surgery. J Cataract Refract Surg2010; 36:1213–1229.

6. Friedman D, Wilson M, Liebmann J, Fechtner R,Weinreb R. An evidence-based assessment of riskfactors for the progression of ocular hypertensionand glaucoma. Am J Ophthalmol 2004; 138: s19-s31.

7. Zheng H, Vingrys AJ, Armitage JA, Bui BV. The roleof blood pressure in glaucoma. Clin Exp Optom2011; 94: 133–149.

8. Huang J, Pesudovs K, Yu A, Wright T, Wen D, Li M,Yu Y et al. A comprehensive compararison ofcentral corneal thickness measurement. Optom VisSci 2011; 88: 940–949.

9. Li Y, Tang M, Zhang X, Salardi CH, Ramos JL,Huang D. Pachymetric mapping with Fourier-

domain optical coherence tomography. J CataractRefract Surg 2010; 36: 826–831.

10. Branco JL, Li Y, Huang D. Clinical and researchapplications of anterior segment optical coherencetomography—a review. Clin Experiment Ophthalmol2009; 37: 81–89.

11. Wylegala E, Teper S, Nowinska AK, Milka M,Dobrowolski D. Anterior segment imaging:Fourier-domain optical coherence tomographyversus time-domain optical coherence tomogra-phy. J Cataract Refract Surg 2009; 35: 1410–1414.

12. Mohamed S, Lee G, Rao S, Wong A, Cheng A, Li E,Chi SC et al. Repeatability and reproducibility ofpachymetric mapping with Visante anteriorsegment-optical coherence tomography. InvestOphthalmol Vis Sci 2007; 48: 5499–5504.

13. Sin S, Simpson TL. The repeatability of corneal andcorneal epithelial thickness measurementes usingoptical coherence tomography. Optom Vis Sci 2006;83: 360–365.

14. Benavente-Pérez A, Hosking SL, Logan NS, BansalD. Reproducibility-repeatability of coroidal thick-ness calculation using optical coherence tomogra-phy. Optom Vis Sci 2010; 87: 867–872.

15. Available from: www.optopol.com. [Accessed 22June 2011]

16. Fukuda S, Kawana K, Yasuno Y, Oshika T. Repeat-ability and reproducibility of anterior ocular bio-metric measurements with 2-dimensional and3-dimensional optical coherence tomography. JCataract Refract Surg 2010; 36: 1867–1873.

17. Milla M, Piñero D, Amparo F, Alió J. Pachymetricmeasurements with a new Scheimplugphotography-based system.Intraobserver repeat-ability and agreement with optical coherence tom-ography pachymetric. J Cataract Refract Surg 2011;37: 310–316.

18. Fermanian J. Mesure de l′accordentre deux juges:cas quantitatif. Rev Epidém Santé Publ 1984; 32: 408–413.

19. Sackett D. A rational clinical examination. Aprimer on the precision and accuracy of the clinicalexamination. JAMA 1992; 267: 2638–2644.

20. Li H, Leung C, Wong L, Cheung C, Pang C,Weinreb R, Lam DS. Comparative study of centralcorneal thickness measurement with slit-lampoptical coherence tomography and visante opticalcoherence tomography. Ophthalmology 2008; 115:796–801.




repeatability and reproducibility of corneal thickness using soct copernicus hr

Documents