comparison between central corneal thickness measurements by ultrasound pachymetry and optical...
TRANSCRIPT
Clinical and Experimental Ophthalmology 2006; 34: 751–754doi:10.1111/j.1442-9071.2006.01343.x
© 2006 Royal Australian and New Zealand College of Ophthalmologists
� Correspondence: Professor Dennis SC Lam, Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, 3/F., Hong Kong
Eye Hospital, 147K Argyle Street, Kowloon, Hong Kong, China. Email: [email protected]
Received 13 September 2005; accepted 11 July 2006.
Original Article
Comparison between central corneal thickness measurements by ultrasound pachymetry and optical coherence tomographyDexter YL Leung FRCS, Douglas KT Lam MBBS, Barry YM Yeung FRCS and Dennis SC Lam MD FRCOphthDepartment of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong Eye Hospital, Kowloon, Hong Kong, China
Key words: cornea, optical coherence tomography,pachymetry, ultrasound.
INTRODUCTION
Measurement of central corneal thickness (CCT) plays animportant role in both diagnostic and therapeutic assessmentof ocular diseases. An ideal tool for pachymetry should beaccurate, reliable, safe and non-invasive. Various methodshave been described in measuring CCT, including ultrasoundpachymetry (U-PACH),1 Orbscan,2 optical coherencetomography (OCT),3 ultrasound biomicroscopy4 and confo-cal microscopy.5 U-PACH is currently most commonly usedand is regarded as the standard in measuring corneal thick-ness. However, U-PACH requires corneal touch with topicalanaesthesia, and the perpendicularity of the probe withrespect to the cornea is often difficult to ascertain, whichmay mean potential inaccuracies.
As for OCT, it is known for its non-invasive, cross-sectional visualization and measurement of thickness of ret-ina. OCT is also used to image the ocular anterior segmentwith detail. OCT has the ability to make repeated cross-sectional measurements at precisely the central part of thecornea. Perpendicularity onto the central cornea is alsoassured. In addition, corneal touch is not required.
We propose that OCT can be an alternative for pachym-etry with its advantages. This study compares agreement ofCCT measurements by U-PACH and OCT, employing theBland–Altman plot into the analysis.
METHODS
All procedures followed the tenets of the Declaration ofHelsinki. Chinese subjects were recruited from a general
ABSTRACT
Purpose: Measurement of central corneal thickness (CCT)plays an important role in both diagnostic and therapeuticassessment of ocular diseases. Although ultrasound pachym-etry (U-PACH) is regarded as the golden standard formeasurement of CCT, optical coherence tomography(OCT) may offer advantages as it can locate the centralcornea with precision with no corneal touch. Nevertheless,the agreement of OCT with U-PACH has not yet beengauged by Bland–Altman analysis. This study compares CCTmeasurement by OCT with that by U-PACH.
Methods: Healthy subjects without ocular abnormality(except refractive errors less than or equal to −6.0 D),contact lens wear or ocular surgery were recruited. CCTwas measured in one eye of normal subjects using OCTand U-PACH. Results were compared using correlation andBland–Altman plots.
Results: Fifty subjects were recruited. Mean ± SD CCTmeasured by OCT was 565 ± 33 µm. This was highly cor-related (Pearson’s coefficient = 0.934) with the mean thick-ness measured by U-PACH (543 ± 33 µm). The coefficientsof variation were good and comparable at 7.9% for U-PACH and 3.5% for OCT. Compared with U-PACH, OCTconsistently overestimated the CCT by a mean of 23 µmas shown on Bland–Altman plot.
Conclusion: CCT measured by OCT and U-PACH is highlycorrelated. With appropriate adjustment factor, OCTagrees well with U-PACH and is a reliable alternative forCCT measurement.
752 Leung et al.
© 2006 Royal Australian and New Zealand College of Ophthalmologists
outpatient eye clinic of our eye hospital. The subjects werescreened for any glaucomatous, anterior and posterior seg-ment diseases. They were included for study if they have noocular abnormality (except refractive errors less than or equalto −6.0 D), corneal pathologies or opacities, contact lenswear or ocular surgery. Written informed consent wasobtained from all participants. One eye was randomly cho-sen from each subject for measurement. CCT was measuredusing OCT and U-PACH.
Optical coherence tomography measurements were per-formed with a commercially available OCT device (Hum-phrey Instruments, Dublin, CA, USA). Calibration had beenperformed prior to all measurements. Measurement of dis-tance in the axial direction was a factory calibration. Allcalibrations were performed by certified engineers from themanufacturer (Carl Zeiss Meditec, Hong Kong, China) usinga calibration test-eye tool that was a glass plate whose thick-ness and index of refraction were accurately known. The x-, y- and z-calibrations were shown to be alright when thecalibration test-eye tool, attached to the ocular lens housing,was orientated such that the centre (cross-hair) of the test-eye scale was overlapping the scan circle displayed on themonitor. Corneal imaging was obtained by defocusing thelens and relaying the OCT beam to the cornea. The headand chin rests for the subject were extended for this purpose.During measurement, the patient was asked to fixate to thefixation beam and the probe beam was positioned to passthrough the cornea bisected the pupil horizontally. A 4-mm-long scan is obtained from the central cornea overlying theentrance pupil. Measurement of tissue thickness is performedwith scan profile display. Computer software-controlled cur-sors are manually placed at the peak of the reflectivity spikescorresponding to the anterior corneal surface or the posteriorcorneal surface. Tissue thickness is calculated between peaksfrom the time-delay of reflected light. Computer cursors areplaced at the peak of the reflectivity spikes because this wasfound to yield better reproducibility values than placing thecursors at the beginning of the rising slopes.6 We have inde-pendently verified that cursors placed at the peak of thespikes yielded better reproducibility. Three consecutive mea-surements of CCT were taken by OCT and averaged. Allmeasurements were taken by a single observer (DYLL).
Ultrasound pachymetry was determined using an A-scanultrasonic pachymeter (Pachette 500, DGH Technology Inc,Frazer, PA, USA). Pachymeter was pre-calibrated for allmeasurements. The ultrasonic velocity was set at 1640 m/s.Topical anaesthesia with Novesin (0.4% benoxinatehydrochloride; Novartis, Basel, Switzerland), the probe washeld perpendicular on the central cornea. Five readings wereobtained and averaged. OCT readings were taken first toavoid any potential disturbance caused by the corneal con-tact involved in U-PACH. All measurements were taken bya single observer (DKTL), who was blinded to the OCTresults.
Statistical analysis was performed using SPSS version 10.0(SPSS Inc., Chicago, IL, USA). Test of normality assumptionwas performed with one-sample Kolmogorov–Smirnov test.
Corneal thickness measurements were compared using Pear-son’s correlation, paired t-test and Bland–Altman plots.7 Thewithin-subject reproducibility for U-PACH and OCT mea-surements was examined by calculating the within-subjectcoefficient of variation using root mean square method aspreviously reported.8
RESULTS
Of the entire subject group (n = 50; only one eye per subjectwas used), the male-to-female ratio was 24:26. The meanage of subjects was 60.1 years (range 26–89 years). Themean intraocular pressure was 15.2 mmHg (range 11.0–20.0 mmHg). Test of distribution revealed the CCT data tofollow Gaussian distribution (Kolmogorov–SmirnovZ = 0.726; P = 0.668 for OCT measurements, and Z = 0.558;P = 0.915 for U-PACH measurements). Mean CCT meas-ured by OCT was 565 ± 33 µm. This was highly correlated(Pearson’s correlation = 0.934) with the mean thicknessmeasured by U-PACH (543 ± 33 µm). Figures 1 and 2 showhistograms of CCT as measured using U-PACH and OCT,respectively. The mean difference between the two methodswas 23 ± 12 µm. The difference was statistically significant(paired t-test, P < 0.0001, two-tailed). Coefficients of varia-tion (COV) were comparable at 7.9% for U-PACH and3.5% for OCT. From the Bland–Altman analysis (Fig. 3),OCT appeared to be consistently overestimating the CCT,compared with U-PACH, as the 95% of differences of read-ings between OCT and U-PACH lies between 46.82 µm and−1.30 µm. Figure 4 shows a scattered plot analysis of theCCT as measured by OCT and U-PACH. A regression equa-tion can be derived as follows: (mean OCT CCT) =53.24 + 0.94 × (mean U-PACH CCT). This model predictswell as the coefficient of determination (R2) was 0.87.
Figure 1. Histogram of central corneal thickness as measured byultrasound pachymetry.
12
10
8
5Fre
quen
cy
2
0
450 500
Mean central corneal thickness measured by ultrasoundpachymetry (µm)
550 600 650
Optical coherence tomographic pachymetry 753
© 2006 Royal Australian and New Zealand College of Ophthalmologists
We have independently verified the suggestion fromanother report that cursors placed at the peak of the spikesyielded better reproducibility than that placed at the begin-ning of the rising slope.6 Our reproducibility data (intraob-server COV) from 12 sets of peak-to-peak and 12 sets ofrising slope-to-slope measurements from five patients (i.e. 60measurements from each method) showed that the COVwith peak-to-peak method =2.5% while the COV with risingslop-to-slope method =7.8%. For this set of analysis, themean of peak-to-peak measurements =555.1 µm and themean of rising slope-to-slope measurements =563.5 µm.
DISCUSSION
Accurate measurement of corneal thickness is vital in allkeratorefractive surgeries. Moreover, variations in CCT
affect the accuracy of measuring intraocular pressure byGoldmann applanation tonometry,9 which was proposed tohave a significant effect on the clinical management ofpatients with glaucoma and glaucoma suspect.10 As many as20.2% of patients may have an outcomes-significant intraoc-ular pressure adjustment (≥3.0 mmHg) when CCT was takeninto account, which had led to a change of decisions regard-ing eyedrop therapy, laser therapy and glaucoma surgery in8.5%, 2.1% and 3.2%, respectively.10 Increased CCT is asso-ciated with ocular hypertension11 and a thin cornea is asso-ciated with normal tension glaucoma.12 Visual fieldprogression in patients with open-angle glaucoma wasreported to be significantly associated with thinner CCT.13
What is more, it was suggested that CCT is an independentrisk factor in development of primary open-angle glaucomafrom ocular hypertension.14
In this present study, measurement of CCT of eyes withOCT showed high correlation to the readings obtained withU-PACH. COV showed comparably low within-subjectvariation between the two techniques. Bland–Altman analy-sis showed that the limit of agreement was good over a widerange of mean CCT (410–610 µm). Of note, the mean CCTby OCT (565 µm ± 33 µm) was 23 µm more than the meanCCT recorded with U-PACH (543 ± 33 µm). Comparedwith U-PACH, OCT overestimates consistently over a rangeof mean CCT, as demonstrated in the Bland–Altman plot.Some other studies appeared to show higher mean valuesobserved with U-PACH than with OCT, which appeared tounderestimate CCT by 24–49 µm.10,13 We are not sure of thereason for the discrepancy. We propose an explanation thatas U-PACH required corneal touch, the applanation force ofthe probe tip could displace tear film (up to 10 µm) and
Figure 2. Histogram of central corneal thickness as measured byoptical coherence tomography (OCT).
20
15
10
Fre
quen
cy
5
0
450 500
Mean central corneal thickness measured by OCT (µm)
550 600 650
Figure 3. A Bland–Altman plot analysis of the central cornealthickness (CCT) as measured by optical coherence tomography(OCT) versus ultrasound pachymetry (U-PACH).
100
50
–50
480 520
Mean of CCT measured byOCT and U-PACH (µm)
Diff
eren
ce b
etw
een
OC
T-
and
U-P
AC
H-
mea
sure
d C
CT
(µm
)
560 600
Mean + 2SD = 46.82 µm
Mean – 2SD = –1.30 µm
Mean = 22.76 µm
0
Figure 4. A scattered plot analysis of the central corneal thick-ness (CCT) as measured by optical coherence tomography(OCT) versus ultrasound pachymetry (U-PACH). (Mean OCTCCT) = 53.24 + 0.94 × (mean U-PACH CCT); R2 = 0.87.
Mean of CCT measured by U-PACH (µm)
Mea
n of
CC
T m
easu
red
by O
CT
(µm
)
620600
580560
540520
500480
460
640
620
600
580
560
540
520
500
480
754 Leung et al.
© 2006 Royal Australian and New Zealand College of Ophthalmologists
compress the epithelium, thereby giving a lower CCT mea-surement.15 Another possible reason was difference in theway in defining CCT over OCT. We have chosen the peakof reflectivity spikes rather than the beginning of risingslopes as in other studies. Other possible reasons were suchas difference in refractive indices of cornea and retina, orissues relating to calibration. In fact, it was even suggestedthere was a large variation in corneal thickness in the samecorneas measured by ultrasonic pachymeters of differentmanufacturers.16 After all, the most important finding inthese studies is that OCT measurements were close to thatof U-PACH, and in our study, utilizing Bland–Altman anal-ysis, it demonstrated a good agreement to U-PACH, afteradjusting with a correction factor (in our case 23 µm).
We believe that our choice of placing cursors at spikesrather than rising slopes for measurements of corneal thick-ness over OCT was justified by our better COV with theformer method.
Optical coherence tomography is a non-invasive andnon-contact method in obtaining images and measurementsof the eye. The probe beam can be placed with precisionover the central cornea with magnified view on the videomonitor. A fixation light from the OCT facilitates patient’sgaze, thereby ensuring perpendicularity. Then, its non-contact method potentially avoids errors during applanation.
Two previous studies had investigated the central cornealmeasurements using U-PACH, OCT and Orbscan.17,18 Ourstudy goes further by first incorporating the Bland–Altmanplot,7 which is the golden standard for comparison of agree-ment among instruments, into our analysis. Second, weavoided the use of both eyes for each subject, which mayintroduce a bias of over-representation of samples. Our studyagrees that OCT produces consistent and reliable centralcorneal measurements; however, a correction factor of whichshould be taken consideration.
In conclusion, CCT measurements by OCT and U-PACH are highly correlated. Mean OCT measurementswere approximately 23 µm (4%) more than the thicknessmeasured by the gold-standard U-PACH, and is consistentlyso over a wide range of CCT, as shown in the Bland–Altmanplot. OCT is a non-contact examination technique, measure-ments are easy to perform, with a suitable correction factor,it agrees well with and can be a reliable alternative to ultra-sound pachymetry.
REFERENCES
1. Marsich MW, Bullimore MA. The repeatability of cornealthickness measures. Cornea 2000; 19: 792–5.
2. Liu Z, Huang AJ, Pfugfelder SC. Evaluation of corneal thick-ness and topography in normal eyes using the Orbscan cornealtopography system. Br J Ophthalmol 1999; 83: 774–8.
3. Bechman M, Thiel MJ, Neubauer AS et al. Central cornealthickness measurement with a retinal optical coherence tomog-raphy device versus standard ultrasonic pachymetry. Cornea2001; 20: 50–4.
4. Avitrabile T, Marano F, Uva MG et al. Evaluation of central andperipheral corneal thickness with ultrasound biomicroscopy innormal and keratoconic eyes. Cornea 1997; 16: 639–44.
5. Masters BR, Kino G. Confocal microscopy of the eye. In:Masters BR, ed. Noninvasive Diagnostic Techniques in Ophthalmology.New York, NY: Springer-Verlag NY Inc, 1990; 152–71.
6. Maldonado MJ, Ruiz-Oblitas L, Munuera JM et al. Opticalcoherence tomography evaluation of the corneal cap and stro-mal bed features after laser in situ keratomileusis for high myo-pia and astigmatism. Ophthalmology 2000; 107: 81–8.
7. Bland JM, Altman DG. Statistical methods for assessing agree-ment between two methods of clinical measurement. Lancet1986; 1: 307–10.
8. Bland JM, Altman DG. Statistics notes: measurement error.BMJ 1996; 313: 744.
9. Ehlers N, Bramsen T, Sperling S. Applanation tonometry andcentral corneal thickness. Acta Ophthalmol (Copenh) 1975; 53:34–43.
10. Shih CY, Graff Zivin JS, Trokel SL, Tsai JC. Clinical signifi-cance of central corneal thickness in the management of glau-coma. Arch Ophthalmol 2004; 122: 1270–5.
11. Herman DC, Hodge DO, Bourne WM et al. Increased cornealthickness in patients with ocular hypertension. Arch Ophthalmol2001; 119: 334–6.
12. Morad Y, Sharon E, Hefetz L et al. Corneal thickness and cur-vature in normal-tension glaucoma. Am J Ophthalmol 1998; 125:164–8.
13. Kim JW, Chen PP. Central corneal pachymetry and visual fieldprogression in patients with open-angle glaucoma. Ophthalmol-ogy 2004; 111: 2126–32.
14. Brandt JD, Beiser JA, Kass MA et al. Central corneal thicknessin the ocular hypertension treatment study (OHTS). Ophthal-mology 2001; 108: 1779–88.
15. Lin MC, Graham AD, Polse KA et al. Measurement of post-lenstear thickness. Invest Ophthalmol Vis Sci 1999; 40: 2833–9.
16. Reader AL, Salz JJ. Differences among ultrasonic pachyme-ters in measuring corneal thickness. J Refract Surg 1987; 3: 7–11.
17. Wong AC, Wong CC, Yuen NS et al. Correlational studyof central corneal thickness measurements on Hong KongChinese using optical coherence tomography, Orbscan andultrasound pachymetry. Eye 2002; 16: 715–21.
18. Fishman GR, Pons ME, Seedor JA et al. Assessment of centralcorneal thickness using optical coherence tomography. J Cat-aract Refract Surg 2005; 31: 707–11.