Dual versus single Scheimpflug camera for anterior segment analysis: Precision and agreement

ARTICLE

Dual versus single Scheimpflug camera for anterior segment analysis: Precision and agreement Jaime Aramberri, MD, Luis Araiz, MS, Ane Garcia, OD, Igor Illarramendi, OD, Jaione Olmos, OD, Izaskun Oyanarte, OD, Amaya Romay, OD, Itxaso Vigara, OD

PURPOSE: To assess the repeatability, reproducibility, and agreement of the Pentacam HR singlecamera and Galilei G2 dual-camera Scheimpflug devices in anterior segment analysis. SETTING: Begitek Cl
ınica Oftalmol
ogica, San Sebasti
an, Spain. DESIGN: Prospective randomized observational study. METHODS: Healthy young individuals had 3 consecutive tests by 2 examiners. Analyzed parameters were anterior and posterior cornea simulated keratometry (K), K flat, K steep, astigmatism magnitude and axis, J0 and J45 vectors, asphericity, total corneal higher-order wavefront aberrations (root mean square [RMS], coma, trefoil, spherical aberration), central cornea and thinnest-point thicknesses, and anterior chamber depth. Repeatability and reproducibility were evaluated by calculating the within-subject standard deviation (Sw), some derived coefficients, and the intraclass correlation coefficient. Agreement was assessed with the Bland-Altman method. RESULTS: The single-camera device reproducibility (Sw) was simulated K, 0.04 diopter (D); J0, 0.03 D; J45, 0.04 D; total power, 0.04 D; spherical aberration, 0.02 mm; higher-order aberrations (HOAs), 0.02 mm; central corneal thickness (CCT), 3.39 mm. The dual-camera device Sw was simulated K, 0.07 D; J0, 0.13 D; J45, 0.04 D; total power, 0.08 D; spherical aberration, 0.02 mm; HOAs, 0.11 mm; CCT, 1.36 mm. Agreement was good for most parameters except total corneal power (mean difference 1.58 D G 0.22 (SD) and HOA RMS (mean difference 0.48 G 0.19 mm) (both P<.00). CONCLUSIONS: Repeatability and reproducibility were good for all parameters. The single-camera device was more precise for curvature, astigmatism, and corneal wavefront error measurements and the dual-camera device for pachymetry measurements. Agreement was good with some relevant exceptions. Financial Disclosure: Dr. Aramberri is consultant to Costruzione Strumenti Oftalmici, Firenze, Italy. No other author has a financial or proprietary interest in any material or method mentioned. J Cataract Refract Surg 2012; 38:1934–1949 Q 2012 ASCRS and ESCRS

Scheimpflug tomography has become a valuable tool for corneal and anterior segment analysis. This technology is superior to Placido topography because it can measure beyond the anterior surface; thus, corneal thickness and posterior curvature can be evaluated with high precision.1 Compared with the first commercially available tomography technology (scanning slit; Orbscan, Bausch & Lomb), Scheimpflug tomography gives more precise anterior and posterior corneal measurements.2 Corneal anterior and posterior curvature and elevation measurements are now essential in clinical applications, such as refractive surgery, keratoconus 1934

Q 2012 ASCRS and ESCRS Published by Elsevier Inc.

treatment, and advanced intraocular lens (IOL) power calculations. Therefore, precision must be studied to assess the reliability and to calculate the error contribution of such measurements. Agreement between instruments is also important in establishing whether values obtained from different machines can be accepted as interchangeable. Two of the most popular Scheimpflug cameras, the Pentacam (Oculus Optikger€ate GmbH) and the Galilei (Ziemer Group), have significant differences in hardware configuration. The Pentacam has 1 rotating camera, while the Galilei has 2 rotating cameras combined with a Placido disk. Recently, the hardware of both 0886-3350/$ - see front matter http://dx.doi.org/10.1016/j.jcrs.2012.06.049

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

devices was upgraded (Pentacam HR, software version 1.17; Galilei G2, software version 5.2.1). To our knowledge, there is no published study comparing the precision and agreement between these 2 new devices in the same sample of eyes. There is no gold standard testing method for corneal topography– tomography to check the accuracy of in vivo measurements. Thus, an evaluation of precision in terms of repeatability and reproducibility is a good approach to determine which devices perform better. Agreement is a way of determining the exchangeability of measurements of different instruments and can be an indirect indicator of accuracy. The objective of this study was to compare the repeatability, reproducibility, and agreement of several corneal and anterior segment parameters in normal eyes between the Pentacam HR single-camera device and the Galilei G2 dual-camera device. SUBJECTS AND METHODS This prospective observational study comprised randomly selected eyes of young healthy members of the clinic staff, attending patients, and companions. The volunteers were informed about the objective of the study and signed an ad hoc informed consent document. The study adhered to the principles outlined in the 1964 Declaration of Helsinki. Subjects who had ocular surgery or a history of eye pathology that could affect the measurements were excluded. No eyedrops were instilled 24 hours before the examination. No contact lens use was allowed 7 days before the examination.

Study Design The definitions of precision, repeatability, and reproducibility used by the International Organization for Standardization3,4 were followed. Precision is expressed in terms of data dispersion and can be studied by analyzing the repeatability and reproducibility of a measurement. To assess repeatability, all factors (operator, instrument, calibration, environment, and time between measurements [shortest possible]) are kept constant. To assess reproducibility, 1 factor (the operator in present study) is changed.

Instruments The Pentacam HR is an eye tomographer that uses a 360degree rotating Scheimpflug camera to obtain cross-sectional images of the anterior segment and is illuminated with a monochromatic slit-light source (blue light–emitting diode

Submitted: April 5, 2012. Final revision submitted: June 26, 2012. Accepted: June 26, 2012. From Begitek Cl
ınica Oftalmol
ogica, San Sebasti
an, Spain. Corresponding author: Jaime Aramberri, MD, Begitek Cl
ınica Oftalmologica, Plaza Teresa de Calcuta 7, San Sebasti
an 20012, Spain. E-mail: [email protected].

1935

[LED] of 475 nm wavelength). This device was introduced in 2007 as an upgrade to the previous model, which was commercialized in 2002. The most significant improvement was an increase in image quality resulting from the higher resolution, 1.45-megapixel camera. The device can obtain up to 50 images in 2 seconds with 138 000 evaluated measuring points. In this study, 25 images per scan setting were used. All maps were calculated from the Scheimpflug images. The software version was 1.17.A The Galilei G2 is a tomographer that uses a combination of 2 rotating Scheimpflug cameras and a Placido disk to measure the anterior segment. Observation illumination is obtained from an 810 nm LED, and the slit-light source is a 470 nm LED. Scheimpflug and Placido raw data are merged into a single dataset with a proprietary algorithm to compute all maps of the anterior cornea. Any structure posterior to the anterior cornea is modeled from the Scheimpflug data. This device was introduced in 2010 as an upgrade to the previous model, which was commercialized in 2007. The main improvements were faster processing hardware, new algorithms for edge detection, and merging of Placido and Scheimpflug data. The device can obtain up to 60 Scheimpflug scans and 2 Placido images in 2 seconds with 122 000 evaluated measuring points. The software version was 5.2.1.B Both instruments were reviewed and calibrated before the measurements were taken.

Measurements Two optometrists with more than 7 years of experience in anterior segment diagnostic instruments performed the examinations sequentially during a single session. The order of the examiner and the order of the device were randomized using a computer-generated random-numbers matrix. The eye to be examined was randomized as well. Each of the examiners performed 3 measurements with each machine. The subject was asked to blink before each capture. The device was defocused, realigned, and focused between 2 measurements. The criterion to accept a measurement as valid was the “quality OK” image expressed by the software.

Parameters The following parameters were evaluated: 1. Anterior and posterior cornea keratometry (K) flat and K steep; that is, the keratometric power of the flat meridian and steep meridian, respectively, 90 degrees apart, with the greatest difference in mean power. This power is calculated from the radius of curvature (r) using the paraxial formula K Z (n2 n1)/r, where K is in diopters (D) and r is in meters. For the anterior cornea, n2 is the standard keratometric index of refraction 1.3375 while n1 is 1 (air). For the posterior cornea, n2 is 1.336 (aqueous) and n1 is 1.376 (cornea). The Galilei manual instructs that this parameter should be calculated 0.5 to 2.0 mm from the central cornea, while the Pentacam manual states it should be calculated for the 3.0 mm central cornea. 2. Anterior and posterior cornea simulated K; that is, the arithmetic mean of K steep and K flat (D). 3. Anterior and posterior cornea astigmatism; that is, the arithmetic difference between K flat and K steep (D). 4. Anterior and posterior cornea axis flat; that is, the axis of K flat (degrees).

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1936

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

5. Anterior and posterior orthogonal corneal power vectors J0 and J45 for the cardinal (0 to 180 degrees) and oblique (45 to 135 degrees) meridians calculated with the following formulas: J0 Z (C/2) cos(2a) and J45 Z (C/2) sin(2a), where C is the astigmatism magnitude and a is the angle in radians.5 6. Anterior and posterior cornea shape factor (asphericity coefficient); that is, the E shape factor for a central diameter of 8.0 mm averaged over all meridians. This value can be related to other descriptors of the shape of a conic section as follows: E Z e2 Z 1 p Z Q 7. Corneal total power; that is, the central corneal power calculated by ray tracing through the anterior and posterior corneal surfaces applying the Snell law with real refraction index numbers (1 for air, 1.376 for cornea, 1.336 for aqueous). The Pentacam device calls this parameter the total corneal refractive power, while the Galilei device calls it the total corneal power. With the present software version (5.2.1) of the Galilei device, there has been a change in how this parameter is calculated with respect to previous software versions and the new total corneal power is stated to be approximately 3% smaller. The calculated area of analysis was 4.0 mm, and this was set at the IOL Power menu of the Galilei device and the Power Distribution menu of the Pentacam device. The unit of measure is diopters. 8. Total corneal higher-order wavefront aberrations (HOAs). Both devices calculate HOAs with a variable area of analysis up to the 8th radial order Zernike coefficients. The following parameters were recorded: HOA root mean square (RMS), 4th-order spherical aberration Z(4,0), 3rd-order coma Z(3,1) and Z(3, 1), and 3rd-order trefoil Z(3,3) and Z(3, 3). All were calculated for a 6.0 mm diameter area of analysis. The unit of measure is microns. 9. Pachymetry. Corneal thickness was recorded at n 2 points; that is, the central cornea (apex) and the thinnest point. The unit of measure is microns. 10. Anterior chamber depth (ACD); that is, the distance from the corneal endothelium to the anterior surface of the crystalline lens. The unit of measure is millimeters.

Statistical Analysis All data were recorded in an Excel spreadsheet (Microsoft Corp.) by the same person. At the time the data were recorded, the quality of the scans was checked again. Statistical analysis was performed with Medcalc software (version 11.4.2.0, Medcalc Software). Normality of data was first assessed with the Kolmogorov-Smirnov test. Parametric statistics were used in all cases. Repeatability and reproducibility were evaluated with the following methods and parameters: 1. Within-subject standard deviation (Sw), which is derived from the square root of the residual mean square from a repeated-measures analysis of variance (ANOVA). 2. Repeatability coefficient, which is computed as 1.96SwO2 (2.77 Sw). For 95% of all pairs of measurements, the absolute difference between 2 measurements may be as much as this value.6 3. Coefficient of variation (CoV), which is the Sw:mean ratio (%). This value gives an idea of the Sw proportionality with respect to the mean. The lower the CoV, the higher the repeatability.

4. Intraclass correlation coefficient (ICC), which is the ratio of the between-subject variance to the sum of the pooled within-subject variance and the between-subject variance. A 2-way random-effects ICC was used for consistency of single measurements. The value is 1 if there is no variance within repeated measurements; that is, when all variations are a result of the variability of the parameter itself. The ICC is considered to be very good if the value is more than 0.90, moderate if 0.75 to 0.90, and bad if less than 0.75. The coefficient and the 95% confidence interval (CI) are reported. 5. Repeated-measures ANOVA. Single-factor ANOVA was performed to check for differences between measurements; P values less than 0.05 were considered statistically significant. Equality of variances (sphericity) was checked with the Epsilon parameter (Greenhouse-Geisser estimate). When this was more than 0.75, the Huynh-Feldt correction was used for F and P values; when it was less than 0.75, the Greenhouse-Geisser correction was used.7 6. The paired t test was used to test the difference in repeatability variance between devices using measurements from examiner 1. The same test was used to test the difference between reproducibility variance between devices. The significance level was 0.05. 7. Agreement between the 2 devices was analyzed with Bland-Altman plots, where differences between the devices were plotted against mean values.8 The mean difference (bias) and 95% limits of agreement (LoA) were calculated. The LoA were computed as the mean difference G 1.96 standard deviation (SD), and they represent the limits of the range for the 95% of differences between the 2 devices.

RESULTS Eighteen right eyes and 17 left eyes of 35 subjects were analyzed. Twenty-four subjects (68.57%) were women, and 11 (31.43%) were men. The mean age was 34.91 years G 5.42 (SD). From the 210 scans per device (2 examiners, 35 eyes, 3 repetitions), 4 (1.90%) had to be retaken with the single-camera device and 6 (2.86%) with the dualcamera device. This was due to the bad quality of the captured images, as reported by the software. After the data collection, it was found that in 3 maps of 3 eyes, the dual-camera device could not calculate the aberrometry coefficients up to 6.0 mm of diameter area, although the quality of the scan had been reported as OK by the software. All 3 cases were eliminated, decreasing the sample size in the corneal aberrometry analysis to 32. Tables 1 to 5 show the repeatability results for all parameters. Only results of the first examiner are shown. All anterior curvature repeatability results were better with the single-camera device (Table 1). For examiner 2, the simulated K Sw value was 0.05 D with the single-camera device and 0.10 D with the dual-camera device (P Z .00). The simulated K ICC was more than 0.99 for both instruments. Within-subject differences were not statistically significant. Repeatability

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1937

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 1. Repeatability of anterior cornea measurements of the first examiner. Parameter/Device Sim K (D) Single camera Dual camera K flat (D) Single camera Dual camera K steep (D) Single camera Dual camera Axis flat (degrees) Single camera Dual camera Astigmatism (D) Single camera Dual camera J0 (D) Single camera Dual camera J45 (D) Single camera Dual camera E Single camera Dual camera

Sw

R

CoV

ICC

ICC 95% CI

0.06 0.10

0.17 0.27

0.14 0.23

0.998 0.994

0.996, 0.999 0.991, 0.997

0.07 0.22

0.20 0.60

0.17 0.51

0.997 0.975

0.995, 0.998 0.956, 0.986

0.07 0.18

0.20 0.51

0.17 0.42

0.997 0.983

0.995, 0.998 0.972, 0.991

15.06 23.82

41.72 65.98

8.14 12.68

0.526 0.274

0.329, 0.702 0.067, 0.497

0.09 0.35

0.25 0.97

10.55 35.72

0.996 0.603

0.942, 0.982 0.420, 0.757

0.04 0.18

0.12 0.51

d d

0.974 0.700

0.955, 0.986 0.543, 0.822

0.07 0.15

0.18 0.43

d d

0.876 0.622

0.796, 0.931 0.443, 0.700

0.03 0.06

0.07 0.16

8.87 30.67

0.936 0.808

0.892, 0.965 0.694, 0.890

P Value* .00

.01

.00

NS

.01

.00

.00

.00

CoV Z coefficient of variation; E Z asphericity coefficient; ICC Z intraclass correlation coefficient; ICC 95% Z intraclass correlation coefficient with 95% confidence interval; J0 Z Jackson cross-cylinder, axes at 0 degrees and 90 degrees; J45Z Jackson cross-cylinder, axes at 45 degrees and 135 degrees; K Z keratometry; R Z repeatability coefficient; Sim Z simulated; Sw Z within-subject standard deviation *t test; comparison variance between the 2 devices for examiner 1

differences between the 2 devices were statistically significant. The posterior cornea repeatability values were similar (Table 2). Posterior astigmatism J0 was more repeatable with the single-camera device than with the dual-camera device, while J45 was equal. The total power was more repeatable with the single-camera device than with the dual-camera device (P Z .01) (Table 3). The reliability of HOA RMS was not very good (Table 4). The dual-camera device had a better ICC and CoV values than the single-camera device. However, the second examiner had a much better score with the single-camera device (ICC, 0.913; CoV, 8.91%). This was the result of outliers resulting from an incorrect peripheral measurement produced by the upper lid in the first examiner’s dataset. The spherical aberration Sw was similar between the 2 devices (Table 4). The pachymetry repeatability figures were similar for central and thinnest-point locations, with the dual-camera device showing more consistency in

both measurements. All ICCs were greater than 0.98 (Table 5). Tables 6 to 10 show the reproducibility results, which were better than the repeatability results because 3 measurements per examiner were averaged. The differences between the 2 devices are similar to those of the repeatability analysis. Tables 11 to 15 show the agreement results; the mean values, mean differences, SD, 95% CI, and P value of each parameter are shown. Figures 1 to 8 show Bland-Altman plots of the most relevant parameters. The agreement of the anterior cornea parameters was very good (Table 11). Only anterior astigmatism showed a slight difference between the 2 devices (P Z .01). The posterior cornea had more significant disagreements, but again with very small clinical relevance (Table 12). There was a very significant disagreement in the mean HOA RMS value between the 2 devices (P Z .00) (Table 13 and Figure 6). There was also a significant difference in the mean trefoil value (P Z .00) (Table 13).

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1938

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 2. Repeatability of posterior cornea measurements of the first examiner. Parameter/Device Sim K (D) Single camera Dual camera K flat (D) Single camera Dual camera K steep (D) Single camera Dual camera Axis flat (degrees) Single camera Dual camera Astigmatism (D) Single camera Dual camera J0 (D) Single camera Dual camera J45 (D) Single camera Dual camera E Single camera Dual camera

Sw

R

CoV

ICC

ICC 95%

0.02 0.02

0.06 0.07

0.34 0.39

0.992 0.989

0.987, 0.996 0.981, 0.994

0.03 0.03

0.09 0.10

0.52 0.59

0.985 0.977

0.973, 0.992 0.961, 0.988

0.04 0.04

0.11 0.12

0.60 0.67

0.977 0.969

0.961, 0.988 0.948, 0.984

5.48 13.29

15.09 36.81

3.02 7.38

0.775 0.115

0.645, 0.870 0.073, 0.345

0.06 0.06

0.15 0.17

17.63 18.36

0.776 0.725

0.646, 0.872 0.577, 0.839

0.02 0.04

0.06 0.12

d d

0.854 0.605

0.761, 0.917 0.423, 0.718

0.03 0.03

0.09 0.08

d d

0.740 0.672

0.597, 0.848 0.506, 0.804

0.03 0.08

0.09 0.22

10.48 33.88

0.962 0.916

0.935, 0.979 0.858, 0.953

P Value* NS

NS

NS

NS

NS

NS

NS

.00

CoV Z coefficient of variation; E Z asphericity coefficient; ICC Z intraclass correlation coefficient; ICC 95% Z intraclass correlation coefficient with 95% confidence interval; J0 Z Jackson cross-cylinder, axes at 0 degrees and 90 degrees; J45Z Jackson cross-cylinder, axes at 45 degrees and 135 degrees; K Z keratometry; R Z repeatability coefficient; Sim Z simulated; Sw Z within-subject standard deviation *t test; comparison variance between the 2 devices for examiner 1

There was a significant difference between the total corneal refractive power of the single-camera device and the total corneal power of the dual-camera device (P Z .00) (Table 14, Figure 5).

G2 dual-camera device as well as the agreement of measurements. We found that both devices performed more precisely than the older models, which previous studies tested and found to be very reliable.2,9 On the other hand, it is essential for the clinician to know the uncertainty of the measurements of any new instrument because relevant decisions can depend on the measurements in areas such as keratoconus progression diagnosis and IOL power calculations.

DISCUSSION This study was designed to compare the precision of the Pentacam HR single-camera device and the Galilei

Table 3. Repeatability of total cornea measurements for measurements by the first examiner. Parameter/Device Ant/post Single camera Dual camera Total power (D) Single camera Dual camera

Sw

R

CoV

ICC

ICC 95%

0.00 0.00

0.01 0.01

0.35 0.35

0.938 0.945

0.894, 0.966 0.907, 0.970

0.05 0.12

0.14 0.34

0.11 0.30

0.999 0.991

0.998, 0.999 0.985, 0.995

P Value* NS

.01

Ant/post Z ratio between mean anterior radius of curvature and mean posterior radius of curvature; CoV Z coefficient of variation; ICC Z intraclass correlation coefficient; ICC 95% Z intraclass correlation coefficient with 95% confidence interval; R Z repeatability coefficient; Sw Z within-subject standard deviation; Total power Z total corneal refractive power for single-camera device and total corneal power for dual-camera device *t test; comparison variance between the 2 devices for examiner 1

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1939

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 4. Repeatability of total cornea wavefront error (6.0 mm area of analysis) measurements by the first examiner. Parameter/Device HOA RMS (mm) Single camera Dual camera SA Z(4,0) (mm) Single camera Dual camera Coma 90 degrees (mm) Single camera Dual camera Coma 0 degrees (mm) Single camera Dual camera Trefoil 30 degrees (mm) Single camera Dual camera Trefoil 0 degrees (mm) Single camera Dual camera

Sw

R

CoV

ICC

ICC 95%

0.03 0.12

0.07 0.33

23.96 20.26

0.564 0.687

0.364, 0.736 0.518, 0.819

0.03 0.04

0.08 0.10

13.55 16.26

0.814 0.833

0.696, 0.897 0.724, 0.908

0.09 0.09

0.24 0.26

d d

0.740 0.809

0.586, 0.851 0.688, 0.894

0.02 0.13

0.06 0.37

d d

0.961 0.868

0.934, 0.980 0.778, 0.928

0.08 0.13

0.23 0.37

d d

0.478 0.477

0.266, 0.674 0.265, 0.673

0.05 0.15

0.14 0.41

d d

0.509 0.510

0.300, 0.697 0.300, 0.670

P Value* .02

NS

NS

.04

NS

.00

CoV Z coefficient of variation; HOA RMS Z root mean square of higher-order aberrations; ICC Z intraclass correlation coefficient; ICC 95% Z intraclass correlation coefficient with 95% confidence interval; R Z repeatability coefficient; SA Z spherical aberration; Sw Z within-subject standard deviation *t test; comparison variance between the 2 devices for examiner 1

There were significant differences in the anterior, posterior, and total corneal curvature repeatability and reproducibility between the 2 devices. The single-camera device was approximately 2 times more precise than the dual-camera device for all curvature parameters. The simulated K Sw with the singlecamera device was 0.05 D for examiner 1 and 0.06 D for examiner 2. Such low figures have not been previously reported with any keratometry-measuring instrument (ie, with an automatic keratometer or a Placido topographer). Davies et al.10 found an intrasession SD of 0.11 D with an automatic keratometer.

Our group found an Sw of 0.12 D for the IOLMaster 500 device (Carl Zeiss Meditec AG) and the Lenstar LS900 device (Haag-Streit AG) in normal young eyes (data not published). Savini et al.11 report an Sw of 0.10 D for a new single Scheimpflug–Placido device (Sirius, Costruzione Strumenti Oftalmici). The older Pentacam model was reported to have an SD of 0.09 D.12 With the Pentacam HR device, Read et al.13 found a worse Sw simulated K value (0.09 D) than ours and M
odis et al.14 found strangely high Sw values by 2 investigators for K flat (0.28 D) and 0.56 D) and K steep (0.37 and 0.74). In our study, the dual-camera

Table 5. Repeatability of measurements of ACD and corneal pachymetry at the center and at the thinnest point of the cornea from the first examiner. Parameter/Device Central pachymetry (mm) Single camera Dual camera Thinnest pachymetry (mm) Single camera Dual camera ACD (mm) Single camera Dual camera

Sw

R

CoV

ICC

ICC 95%

4.32 2.30

11.96 6.36

0.76 0.40

0.999 0.991

0.998, 0.999 0.985, 0.995

4.29 1.74

11.88 4.82

0.76 0.31

0.981 0.996

0.967, 0.989 0.993, 0.998

0.02 0.02

0.05 0.06

0.62 0.71

0.997 0.995

0.995, 0.998 0.992, 0.998

P Value* .00

.00

NS

ACD Z anterior chamber depth; CoV Z coefficient of variation; ICC Z intraclass correlation coefficient; ICC 95% Z intraclass correlation coefficient with 95% confidence interval; R Z repeatability coefficient; Sw Z within-subject standard deviation *t test; comparison variance between the 2 devices for examiner 1

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1940

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 6. Reproducibility of anterior cornea measurements between 2 observers (mean of 3 measurements compared between 2 observers with each device). P Value* Parameter/Device Sim K (D) Single camera Dual camera K flat (D) Single camera Dual camera K steep (D) Single camera Dual camera Axis flat (degrees) Single camera Dual camera Astigmatism (D) Single camera Dual camera J0 (D) Single camera Dual camera J45 (D) Single camera Dual camera E Single camera Dual camera

Sw

R

CoV

Interobserver Reproducibility

0.04 0.07

0.12 0.18

0.10 0.15

NS NS

0.04 0.14

0.12 0.40

0.10 0.34

NS NS

0.06 0.09

0.16 0.26

0.13 0.21

NS NS

9.82 9.56

27.19 26.47

5.38 5.15

NS .03

0.06 0.20

0.16 0.57

6.63 21.02

NS NS

0.03 0.13

0.09 0.37

d d

NS NS

0.04 0.08

0.12 0.22

d d

NS NS

0.01 0.04

0.03 0.10

4.12 18.09

NS NS

Interdevice Variance .04

.00

.01

NS

.00

.01

.06

NS

CoV Z coefficient of variation; E Z asphericity coefficient; ICC Z intraclass correlation coefficient; ICC 95% Z intraclass correlation coefficient with 95% confidence interval; J0 Z Jackson cross-cylinder, axes at 0 degrees and 90 degrees; J45Z Jackson cross-cylinder, axes at 45 degrees and 135 degrees; K Z keratometry; R Z repeatability coefficient; Sim Z simulated; Sw Z within-subject standard deviation *Paired t test

device rendered a simulated K Sw of 0.10 D for both examiners, similar to the 0.09 D value of Wang et al. 9 and the 0.12 D value of Savini et al.15 0.12 D (both older Galilei model). These results show that both devices overcome, or at least equal, the precision of autokeratometers, making them suitable for follow-up of keratometry stability or for IOL power calculations. Knowledge of precision limits is relevant in answering questions such as whether there was keratoconus progression between 2 visits or whether a refractive shift after laser in situ keratomileusis was caused by regression. In the case of multivariable equations, such as IOL power calculation, we can use these values to determine the error contribution of simulated K to the final random error in IOL power using Gaussian error-propagation analysis.16 Norrby17 estimated that the refraction prediction error contribution of the anterior corneal radius variance was 2.32%, assuming 0.02 mm (0.12 D) as the SD of anterior radius measurement. With the values we found, this error contribution of the anterior cornea would drop to 0.65%, assuming

0.01 mm (0.06 D) as the SD of the anterior cornea measurement and keeping the rest of values unchanged. The posterior curvature precision was very good when measured in diopters. The repeatability and reproducibility Sw values were 0.02 D for both devices (PO.05). Savini et al.11 found similar results with the Sirius device (0.02 D) and with the Galilei device (0.03 D).15 Wang et al.9 report a value of 0.03 D with the Galilei device. Pi~ nero et al.18 found a value of 0.02 mm (0.02 D) for the posterior best-fit sphere with the Pentacam device. Chen and Lam12 found a value of 0.03 D for the posterior M vector with the Pentacam device. It is interesting to translate these curvature figures in diopters into their corresponding values in millimeters (radius of curvature); the results show that the precision is similar for the anterior surface and posterior surface. The Sw for the anterior surface and the posterior surface would be 0.01 mm and 0.02 mm, respectively, with the single-camera device and 0.02 mm for both surfaces with the dualcamera device.

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1941

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 7. Reproducibility of posterior cornea measurements between 2 observers (mean of 3 measurements compared between 2 observers with each device). P Value* Parameter/Device Sim K (D) Single camera Dual camera K flat (D) Single camera Dual camera K steep (D) Single camera Dual camera Axis flat (degrees) Single camera Dual camera Astigmatism (D) Single camera Dual camera J0 (D) Single camera Dual camera J45 (D) Single camera Dual camera E (D) Single camera Dual camera

Sw

R

CoV

Interobserver Reproducibility

0.02 0.02

0.05 0.06

0.29 0.36

.04 NS

0.02 0.04

0.06 0.10

0.36 0.61

NS NS

0.03 0.04

0.08 0.12

0.43 0.65

NS NS

5.00 5.87

13.84 16.25

2.76 3.26

NS NS

0.03 0.06

0.10 0.18

10.89 19.05

NS NS

0.02 0.03

0.06 0.07

d d

NS NS

0.02 0.04

0.06 0.10

d d

NS NS

0.02 0.09

0.05 0.25

5.43 35.43

.03 .07

Interdevice Variance NS

NS

NS

NS

NS

NS

NS

NS

CoV Z coefficient of variation; E Z asphericity coefficient; ICC Z intraclass correlation coefficient; ICC 95% Z intraclass correlation coefficient with 95% confidence interval; J0 Z Jackson cross-cylinder, axes at 0 degrees and 90 degrees; J45Z Jackson cross-cylinder, axes at 45 degrees and 135 degrees; K Z keratometry; R Z repeatability coefficient; Sim Z simulated; Sw Z within-subject standard deviation *Paired t test

Anterior cornea astigmatism precision in terms of the J0 ICC and J45 ICC was within the limit of reliability with the single-camera device and could not be considered reliable with the dual-camera device. The Sw for

axis reproducibility was 9.82 degrees with the singlecamera device and 9.56 degrees with the dualcamera device (PO.05). The precision of astigmatism measurement is lower when the astigmatism value is

Table 8. Reproducibility of total cornea measurements between 2 observers (mean of 3 measurements compared between 2 observers with each device). P Value* Parameter/Device Ant/post Single camera Dual camera Total power (D) Single camera Dual camera

Sw

R

CoV

Interobserver Reproducibility

0.00 0.00

0.01 0.01

0.27 0.31

.08 NS

0.04 0.08

0.10 0.21

0.09 0.19

NS NS

Interdevice Variance NS

.03

Ant/post Z ratio between mean anterior radius of curvature and mean posterior radius of curvature; CoV Z coefficient of variation; R Z repeatability coefficient; Sw Z within-subject standard deviation; Total power Z total corneal refractive power for single-camera device and total corneal power for dual-camera device *Paired t test

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1942

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 9. Reproducibility in total cornea wavefront error measurements between 2 observers (mean of 3 measurements compared between 2 observers with each device) with 6.0 mm area of analysis. P Value* Parameter/Device HOA (mm) Single camera Dual camera SA Z(4,0) (mm) Single camera Dual camera Coma 90 degrees (mm) Single camera Dual camera Coma 0 degrees (mm) Single camera Dual camera Trefoil 30 degrees (mm) Single camera Dual camera Trefoil 0 degrees (mm) Single camera Dual camera

Sw

R

CoV

Interobserver Reproducibility

0.02 0.11

0.05 0.29

17.51 17.93

NS NS

0.02 0.02

0.05 0.06

7.86 9.95

NS NS

0.05 0.10

0.14 0.27

d d

NS NS

0.01 0.09

0.04 0.25

d d

NS NS

0.05 0.08

0.14 0.22

d d

NS NS

0.03 0.12

0.10 0.32

d d

NS NS

Interdevice Variance .03

NS

NS

.00 NS

.00

CoV Z coefficient of variation; HOA Z higher-order aberrations; R Z repeatability coefficient; SA Z spherical aberration; Sw Z within-subject standard deviation *Paired t test

low, as shown by Kobashi et al.19 This is probably the explanation for the bad result in this sample, in which the mean astigmatism was 0.86 G 0.50 D. Surgeons must be aware of this variability when assessing the indications for surgery to correct astigmatism (toric IOL or incisional techniques) in eyes with low astigmatism. The precision in keratometric astigmatism measurements is variable in the literature. Karabatsas et al.20 report axis SD values of 4 degrees and 20

degrees for a manual keratometer and for the TMS-1 Placido topographer (Tomey), respectively. The cylinder magnitude SD was 0.14 D and 0.63 D, respectively. Read et al.13 found that Sw for J0 and J45 was 0.05 with the Pentacam HR device and 0.04 with the E300 Placido topographer (Medmont) calculated over a 6.0 mm diameter. Chen and Lam12 report an SD for J0 and J45 of 0.06 and 0.10, respectively, with the older Pentacam model. Goggin et al.21 compared 3 autokeratometers

Table 10. Reproducibility of corneal pachymetry and ACD measurements between 2 observers (mean of 3 measurements compared between 2 observers with each device). P Value* Parameter/Device Central pachymetry (mm) Single camera Dual camera Thinnest pachymetry (mm) Single camera Dual camera ACD (mm) Single camera Dual camera

Sw

R

CoV

Interobserver Reproducibility

3.39 1.36

9.39 3.77

0.59 0.24

.33 .13

3.41 1.34

9.45 3.72

0.60 0.24

.25 .72

0.02 0.02

0.04 0.04

0.54 0.52

.13 .20

Interdevice Variance .01

.01

.95

ACD Z anterior chamber depth; CoV Z coefficient of variation; R Z repeatability coefficient; Sw Z within-subject standard deviation *Paired t test

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1943

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 11. Agreement in anterior cornea measurements between single-camera device and dual-camera device (mean of 6 measurements with each device; 3 per observer). Device/ Parameter Single camera (mean G SD) Dual camera (mean G SD) Mean difference SD 95% CI P value*

Axis Flat (Degrees)

Astig (D)

43.22 G 1.43 42.78 G 1.44 43.65 G 1.46

183.75 G 18.20

0.86 G 0.50

0.37 G 0.27

0.04 G 0.18 0.29 G 0.10

43.19 G 1.39 42.70 G 1.38 43.67 G 1.42

185.43 G 20.28

0.97 G 0.45

0.38 G 0.27

0.06 G 0.21 0.20 G 0.13

0.11 0.22 0.03, 0.18 0.01

0.01 0.10 0.04, 0.03 NS

Sim K (D)

0.03 0.12 0.07, 0.01 NS

K Flat (D)

K Steep (D)

0.08 0.17 0.14, 0.02 0.01

0.03 0.16 0.03, 0.08 NS

1.68 12.86 2.74, 6.10 NS

J0 (D)

J45 (D)

0.02 0.13 0.06, 0.02 NS

E

0.09 0.10 0.13, 0.06 .00

Astig Z astigmatism; CI Z confidence interval; E Z asphericity coefficient; K Z keratometry; J0 Z Jackson cross-cylinder, axes at 0 degrees and 90 degrees; J45Z Jackson cross-cylinder, axes at 45 degrees and 135 degrees; NS Z not significant *Two-tailed t test

and the Pentacam HR device and found that variability was larger with the Scheimpflug device. The mean astigmatism variability was 0.24 G 0.2 D with the ARK510A device (Nidek Co. Ltd.), 0.22 G 0.13 D with the ARK599 device (Humphrey Instruments, Inc.), 0.28 G 0.25 D with the IOLMaster device (Carl Zeiss Meditec AG), and 0.46 G 0.46 D with the Pentacam HR device. Kobashi et al.19 found that the repeatability ICC was very good with the ARK-700A autokeratometer (Nidek Co., Ltd.) (J0 ICC, 0.974; J45 ICC, 0.908) and with the Atlas 991 Placido topographer (Carl Zeiss Meditec AG) (J0 ICC, 0.921; J45 ICC; 0.904). The mean astigmatism in their sample was higher (1.20 G 0.63 D). The precision of posterior cornea astigmatism measurements was similar between the 2 devices, but with a slight advantage for the single-camera instrument. The reproducibility Sw was 0.02 for J0 and 0.02 for J45 with the single-camera device and 0.03 and 0.04,

respectively, with the dual-camera device; however, the differences were not statistically significant. Chen and Lam12 report a worse repeatability SD with the older Pentacam device (J0, 0.05; J45, 0.04). Szalai et al.22 evaluated the Pentacam HR device but had worse astigmatism repeatability (Sw, 0.066). Wang et al.9 report similar figures for posterior astigmatism repeatability with Galilei device (J0, 0.03; J45, 0.03). Both instruments used in our study calculate the central corneal optical power by ray tracing through the anterior and posterior surfaces. The precision of such parameter depends on both corneal surfaces, which proportionally contribute to the parameter’s precision. The repeatability Sw was 0.05 D and 0.06 D for examiner 1 and examiner 2, respectively, with the single-camera device and 0.12 D for both examiners with the dual-camera device (P Z .02). The reproducibility Sw was 0.04 D with the single-camera device and 0.08 D with the dual-camera device (P Z .03).

Table 12. Agreement in posterior cornea measurements between single-camera device and dual-camera device (mean of 6 measurements with each device; 3 per observer). Device/ Parameter Single camera (mean G SD) Dual camera (mean G SD) Mean difference SD 95 % CI P value*

Axis Flat (Degrees)

Astig (D)

6.38 G 0.25

180.71 G 11.21

0.32 G 0.10

0.14 G 0.05

0.00 G 0.06 0.33 G 0.18

6.07 G 0.24

6.41 G 0.24

179.97 G 7.98

0.34 G 0.10

0.16 G 0.05

0.00 G 0.04 0.26 G 0.27

0.01 0.05 0.03, 0.01 NS

0.03 0.05 0.05, 0.01 .00

0.02 0.07 0.00, 0.04 .05

0.01 0.03 0.02, 0.00 .00

Sim K (D)

K Flat (D)

K Steep (D)

6.21 G 0.24

6.06 G 0.24

6.24 G 0.24 0.02 0.04 0.04, 0.01 .01

0.75 9.85 4.13, 2.64 NS

J0 (D)

J45 (D)

0.00 0.05 0.02, 0.02 NS

E

0.08 0.14 0.12, 0.03 .00

Astig Z astigmatism; CI Z confidence interval; E Z asphericity coefficient; K Z keratometry; J0 Z Jackson cross-cylinder, axes at 0 degrees and 90 degrees; J45Z Jackson cross-cylinder, axes at 45 degrees and 135 degrees; NS Z not significant *Two-tailed t test

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1944

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Table 13. Agreement in total corneal wavefront error between single-camera device and dual-camera device (mean of 6 measurements with each device; 3 per observer). Device/Parameter Single camera (mean G SD) Dual camera (mean G SD) Mean difference SD 95 % CI P value*

HOA RMS (mm)

SA (mm)

0.11 G 0.03 0.59 G 0.20 0.48 0.19 0.42, 0.55 .00

0.21 G 0.06 0.22 G 0.08 0.01 0.05 0.01, 0.03 NS

Coma 90 (mm) 0.04 G 0.15 0.04 G 0.22 0.00 0.20 0.08, 0.07 NS

Coma 0 (mm) 0.00 G 0.11 0.00 G 0.34 0.03 0.31 0.14, 0.08 NS

Trefoil 30 (mm) 0.01 G 0.06 0.13 G 0.15 0.14 0.18 0.21, 0.08 .00

Trefoil 0 (mm) 0.00 G 0.05 0.00 G 0.15 0.00 0.18 0.07, 0.06 NS

HOA RMS Z root mean square of higher-order aberrations; SA Z spherical aberration *Two-tailed t test

Wang et al.9 report a similar Sw (0.09 D) with the older Galilei model. Savini et al.15 found an Sw of 0.12 with a Scheimpflug–Placido instrument (Sirius) and of 0.13 with the older Galilei device. The precision of the ratio of the anterior corneal radius of curvature to the posterior radius of curvature (AP ratio) has not been published before to our knowledge. We found that the 2 devices had the same Sw for repeatability and reproducibility (0.004). This means that the proportion between the 2 surfaces is kept constant, even if there is some variability in the curvature radii measurements. This is an interesting finding for those who calculate arbitrary indices of refraction to determine the total corneal power by measuring the anterior surface only, as done when keratometers and Placido topographers are used.23 The precision in measuring HOAs, expressed by means of Zernike polynomial expansion, gives these systems the ability to measure tiny irregularities in the corneal surface. In the present study, both systems showed some variability in RMS HOA parameters. The single-camera device Sw was 0.03 mm for 1 examiner and 0.01 mm for the examiner 2. The dual-camera device values were 0.12 mm and 0.16 mm, respectively Table 14. Agreement in cornea anterior/posterior ratio and total power between the single-camera device and the dual-camera device (6 measurements with each device; 3 per observer). Device/Parameter Single camera (mean G SD) Dual camera (mean G SD) Mean difference SD 95 % CI P value*

Ant/Post

Total Power (D)

1.21 G 0.02 1.22 G 0.02 0.00 0.01 0.00, 0.01 0.00

42.95 G 1.48 41.37 G 1.33 1.58 0.22 1.66, 1.51 0.00

Ant/post Z ratio between mean anterior radius of curvature and mean posterior radius of curvature; Total power Z total corneal refractive power for single-camera device and total corneal power for dual-camera device *Paired t test

(P Z .02). At this point, we must address the significant difference in the means of this parameter; the mean was 0.11 G 0.03 mm with the single-camera device and 0.59 G 0.20 mm with the dual-camera device. Therefore the CoV of the examiners was not very different between the 2 devices (single camera, 23.96% for examiner 1 and 8.06% for examiner 1; dual camera, 20.26% and 27.00%, respectively). The reason for this variability may be outliers we found in the measurements in 3 eyes with the single-camera device and in 4 eyes with the dual-camera device. These outliers affected the measurement of the superior periphery due to incomplete eye opening; this especially affected the 90-degree coma coefficient and both trefoil coefficients. All cases had an OK quality reading by the software, and the Scheimpflug scores were above 90% in most cases. Reproducibility, which was calculated by averaging 3 measurements per observer, was much better and probably would provide a better comparison (Table 9). The HOA RMS CoV was very similar for the 2 devices (17.51%, single-camera device; 17.93%, double-camera device). The same was true for spherical aberration Z(4,0) (7.86%, single-camera device; 9.95% dual-camera device). The coma and trefoil coefficients had better reproducibility with singlecamera device than with the dual-camera device. Spherical aberration was the most reliable Zernike coefficient and had the best ICC value of all Zernike coefficients. The reproducibility Sw was 0.02 mm for both devices, with no statistically significant difference between instruments. Other studies have also found relative variability in HOA measurements with Scheimpflug topographers. In the 2 most relevant Pentacam studies, the analysis was performed using external software, not the Pentacam software. This means that it might not be possible to compare the results in the studies. In an evaluation of 101 young healthy eyes, Read et al.13 exported the anterior surface elevation data obtained with Pentacam HR device to purpose-designed software, calculating the polynomial terms up to the 8th radial

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

1945

Table 15. Agreement in central and thinnest point pachymetry and ACD measurements between the single-camera device and the dualcamera device (6 measurements with each device; 3 per observer). Device/Parameter Single camera (mean G SD) Dual camera (mean G SD) Mean difference SD 95 % CI P value*

Central Pachymetry (mm)

Thinnest Pachymetry (mm)

ACD (mm)

570.79 G 29.78 568.03 G 27.39 2.76 4.52 4.32, 1.21 0.00

567.04 G 29.85 565.15 G 27.26 1.89 4.67 3.50, 0.29 0.02

3.04 G 0.33 3.14 G 0.32 0.10 0.05 0.08, 0.12 0.00

ACD Z anterior chamber depth *Paired t test

order. Their repeatability Sw values were similar to those in our study. The most relevant values in Read et al.’s study are as follows: HOA RMS, 0.04; Z(4,0), 0.03 mm; and 90-degree coma, 0.04 mm. They found that the repeatability of higher frequency terms was worse, especially for the tetrafoil aberration. Shankar et al.24 exported anterior elevation data obtained with the old Pentacam system to Vol Pro software (Sarver & Associates) and calculated HOAs. They reported higher mean and repeatability values than authors of any other study. The mean HOA RMS was 0.88 G 0.24 mm. The intraobserver SD was 0.17 mm for HOA RMS, 0.15 mm for coma, and 0.16 mm for trefoil. They found that elevation data were missing, and the Pentacam software interpolated them without warning. In the study of Shankar et al., it was again found that repeatability of higher frequency terms was worse. The authors suggest that interpolation

between samples in the periphery, where space between them is maximum, was the reason for the worse repeatability. Wang et al.9 tested the aberrometry repeatability of the older Galilei model and found a similar mean HOA RMS (0.56 G 0.14 mm). Their Sw values were slightly better than our mean value for all terms as follows: 0.02 mm for Z(4,0), 0.08 mm for 90-degree coma, 0.10 mm for 0-degree coma, 0.12 mm for 30-degree trefoil, and 0.11 mm for 0-degree trefoil. As in other studies, Wang et al. report a decrease in repeatability for the higher frequency terms. Savini et al.15 report an Sw value of 0.05 mm for the Z(4,0) coefficient with the older Galilei model and 0.02 mm with the Sirius device,11 which has a single Scheimpflug camera with a Placido disk. In future studies, it would be interesting to determine whether methodology improvements, such as more careful measurements, the discarding of maps

Figure 1. Difference in mean simulated K values between devices. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are shown for comparison (Sim K G Z simulated keratometry with the dual-camera device; Sim K P Z simulated keratometry with the single-camera device).

Figure 2. Difference in mean posterior simulated K values between devices. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are averaged for comparison (Sim K POST G Z simulated posterior keratometry with the dual-camera device; Sim K POST P Z simulated posterior keratometry with the single-camera device).

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1946

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

Figure 3. Difference in mean J0 power vector values between devices. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are averaged for comparison (J0 ANT G Z J0 anterior with the dual-camera device; J0 ANT P Z J0 anterior with the single-camera device).

Figure 4. Difference in mean J45 power vector values between devices. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are averaged for comparison (J45 ANT G Z J0 anterior with the dual-camera device; J45 ANT P Z J0 anterior with the single-camera device).

with peripheral irregularity, or and improved x, y, z alignment (maybe automatically), would lead to better repeatability of higher frequency Zernike terms. The study could determine whether the repeatability of higher frequency Zernike terms is an inherent limitation of Scheimpflug devices due to the number of scans used at present to model the peripheral cornea. Better repeatability figures for these higher frequency terms has been reported for Placido anterior surface aberrometry.13

Pachymetry precision was very good with both devices in our study. Central and thinnest-point measurements were very reliable (repeatability and reproducibility ICC O0.98 in all cases). The reproducibility Sw in the central cornea was 3.39 mm with the single-camera device and 1.36 mm with the dualcamera device (P Z .01). These results are similar to those published for the older Galilei model by Wang et al.,9 who found an Sw of 1.68 mm for central pachymetry, and by Savini et al.,15 who found values of

Figure 5. Difference in mean total corneal power of the dual-camera device and mean total corneal refractive power of the single-camera device. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are averaged for comparison (TOT POW G Z total corneal power with the dual-camera device; TOT POW P Z total corneal refractive power with the single-camera device).

Figure 6. Difference in mean HOA RMS values between devices. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are averaged for comparison (HOA G Z higher-order aberrationswith the dual-camera device; HOA P Z higher-order aberrations with the single-camera device).

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

1947

Figure 7. Difference in mean spherical aberration Z(4,0) values between devices. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are averaged for comparison (Z40 G Z spherical aberration with the dual-camera device; Z40 P Z spherical aberration with the single-camera device).

Figure 8. Difference in mean central pachymetry values between devices. The solid line is the mean difference (bias). The red dotted line is the equality. The maroon dashed lines represent the 95% LoA. Six measurements (3 per observer) are averaged for comparison (PAC APEX G Z pachymetry with the dual-camera device; PAC APEX P Z pachymetry with the single-camera device).

2.15 mm and 1.72 mm for the central pachymetry and thinnest-point pachymetry, respectively. With the older Pentacam model, Nam et al.25 found an Sw of 3.61 mm for the central cornea, while Fu et al.26 found an Sw of 4.20 mm. Compared with other technologies, the precision of Scheimpflug tomographers seems similar to or slightly better than that of ultrasound and scanning-slit tomography. Nam et al.25 found an Sw of 1.70 mm with Fourier-domain optical coherence tomography for the central cornea thickness; this is a little better than Pentacam and similar to Galilei findings in our study as well as in other studies. The reason for the better performance of the Galilei device in thickness measurements could be that the device has 2 opposite Scheimpflug cameras that compensate for the effect of x–y decentration in the pachymetry measurement. The ACD precision was very high as well; the value for the repeatability and reproducibility Sw was 0.02 mm. All ICC values were higher than 0.993. Similar or slightly worse results were found in previous Studies. Using the older Galilei model, Wang et al.9 found an Sw of 0.04 mm and Savini et al.,15 of 0.02 mm. Using the older Pentacam model, Fu et al.26 found an Sw of 0.03 mm. In our study, the agreement between the singlecamera device and the dual-camera device was good for most parameters; therefore, the measurements can be considered interchangeable for purposes such as IOL power calculation with no need for IOL constant adjustment. The most significant difference was in the total corneal power; the difference was 1.58 D (95% CI, 1.51-1.66) between the single-camera device total corneal refractive power and the dual-camera

device total corneal power. The reason is that in the new software version (5.2.1) of the Galilei G2 device changed the optical reference plane for the definition of total corneal power to the anterior corneal surface, decreasing the value in previous software versions by approximately 3%.27,B The Pentacam HR device calculates the total corneal refractive power in reference to the posterior corneal surface. Thus, although the names and definitions in the software of the 2 devices are similar, they are in fact different parameters and care must be taken in their use. An example is the American Society of Cataract and Refractive Surgery post-refractive surgery IOL calculator,C which uses Galilei total corneal power as input and asks the user to distinguish between software version 5.2.1 and previous versions. Another important disagreement was found in the mean HOA RMS value, which was 0.11 G 0.03 mm with the single-camera device and 0.59 G 0.20 mm with the dual-camera device; the mean difference was 0.48 mm (95% CI, 0.42-0.55). This difference can only be explained by the different calculation methods. As seen in the literature, the mean corneal HOA RMS values in normal eyes with a 6.0 mm optical zone differ based on factors such as the measuring technology, calculation method, and sample studied. However, in general, values reported previously are more similar to those in our study with the dualcamera device. For example, Pesudovs28 found a mean HOA RMS value of 0.38 G 0.07 mm using Orbscan device and Vol Pro software. Read et al.13 found a value of 0.409 G 0.100 mm for the anterior cornea using the Pentacam HR device and 0.367 G 0.105 with a videokeratoscope. Purpose-designed software was used in both cases. Wang et al.9 found a mean value

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

1948

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

0.56 G 0.14 mm with the older Galilei model. Gatinel et al.29 found a mean value of 0.85 G 0.51 mm using the OPD-Scan device (Nidek Co. Ltd.) in 57 myopic eyes. In 228 eyes, Wang et al.30 found a mean value of 0.479 G 0.124 mm for the anterior cornea with the Atlas Placido topographer (Carl Zeiss Meditec AG). One limitation of this study is that some parameters share names between the 2 devices even though the calculation method is not exactly the same. In addition, although we addressed the most significant differences (HOA RMS, total corneal power), there were subtle differences in significant measurements, such as K. In conclusion, both tomographers had very good repeatability and reproducibility in measuring anterior and posterior corneal curvature, astigmatism, and pachymetry. The single-camera device was more precise in curvature and astigmatism measurements, and the dual-camera device was more precise in pachymetry measurements. Both devices gave precise spherical aberration measurements but lacked some precision in coma and trefoil measurements. The ACD measurement was very reliable with both devices. Agreement was good for most parameters, except total corneal power and HOA RMS.

3.

4.

5.

6.

7.

8. 9.

WHAT WAS KNOWN  The repeatability and reproducibility of Pentacam and Galilei devices in different samples of patients have been published. However, no direct comparison of all parameters has been presented, and most studies were performed with the older models.

10.

11.

WHAT THIS PAPER ADDS  Curvature and astigmatism precision was very good with both devices, with a significant advantage of the Pentacam HR device. The Galilei G2 was more precise in pachymetry measurements. There was significant disagreement in total corneal power and HOA RMS values between devices.  The information obtained is relevant for making decisions in the day-to-day practice as well in calculating error contributions in multivariable equations, such as those used in IOL power calculations, for example.

12.

13.

14.

15.

REFERENCES 1. Chen D, Lam AKC. Intrasession and intersession repeatability of the Pentacam system on posterior corneal assessment in the normal human eye. J Cataract Refract Surg 2007; 33:448–454 2. Kawamorita T, Uozato H, Kamiya K, Bax L, Tsutsui K, Aizawa D, Shimizu K. Repeatability, reproducibility, and agreement characteristics of rotating Scheimpflug photography and scanning-

16. 17. 18.

slit corneal topography for corneal power measurement. J Cataract Refract Surg 2009; 35:127–133 International Organization for Standardization. Accuracy (Trueness and Precision) of Measurement Methods and Results. Part 1. General Principles and Definitions. Geneva, Switzerland, ISO, 1994; (ISO 5725–1) International Organization for Standardization. Accuracy (Trueness and Precision) of Measurement Methods and Results. Part 2. Basic Methods for the Determination of Repeatability and Reproducibility of a Standard Measurement Method. Geneva, Switzerland, ISO, 1994; (ISO 5725–2) Thibos LN, Horner D. Power vector analysis of the optical outcome of refractive surgery. J Cataract Refract Surg 2001; 27:80–85 Bland JM, Altman DG. Measurement error [Statistics notes]. BMJ 1996; 313:744. Available at: http://www.ncbi.nlm.nih.gov/ pmc/articles/PMC2352111/pdf/bmj00560-0056b.pdf. Accessed July 16, 2012 Girden ER. ANOVA: Repeated Measures (Sage University Paper Series on Quantitative Applications in the Social Sciences, 07–084). Newbury Park, CA, Sage, 1992; Available at: http:// books.google.com/books?idZJomGKpjnfPcC&pgZPR2&lpgZ PR2&dqZSageCUniversityCpapers.Cgirden&sourceZbl& otsZmvWtCdQmbD&sigZR07F5DQTIIL14VhoS4-_fdDfXUM& hlZen&saZX&eiZ4QgEUK7YDKbZ0QGAh_2gBw&vedZ 0CEEQ6AEwAA#vZonepage&qZSage%20University%20 papers.%20girden&fZfalse. Accessed July 16, 2012 Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res 1999; 8:135–160 Wang L, Shirayama M, Koch DD. Repeatability of corneal power and wavefront aberration measurements with a dualScheimpflug Placido corneal topographer. J Cataract Refract Surg 2010; 36:425–430 Davies LN, Mallen EAH, Wolffsohn JS, Gilmartin B. Clinical evaluation of the Shin-Nippon NVision-K 5001/Grand Seiko WR-5100K autorefractor. Optom Vis Sci 2003; 80:320–324. Available at: http://journals.lww.com/optvissci/Fulltext/2003/ 04000/Clinical_Evaluation_of_the_Shin_Nippon_NVision_K.11. aspx. Accessed July 16, 2012 Savini G, Barboni P, Carbonelli M, Hoffer KJ. Repeatability of automatic measurements by a new Scheimpflug camera combined with Placido topography. J Cataract Refract Surg 2011; 37:1809–1816 Chen D, Lam AKC. Reliability and repeatability of the Pentacam on corneal curvatures. Clin Exp Optom 2009; 92:110–118. Available at: http://onlinelibrary.wiley.com/doi/10.1111/j.1444-0938. 2008.00336.x/pdf. Accessed July 16, 2012 Read SA, Collins MJ, Iskander R, Davis BA. Corneal topography with Scheimpflug imaging and videokeratography: comparative study of normal eyes. J Cataract Refract Surg 2009; 35: 1072–1081
dis L Jr, Szalai E, Kolozsva
ri B, Ne
meth G, Vajas A, Berta A. Mo Keratometry evaluations with the Pentacam high resolution in comparison with the automated keratometry and conventional corneal topography. Cornea 2012; 31:36–41 Savini G, Carbonelli M, Barboni P, Hoffer KJ. Repeatability of automatic measurements performed by a dual Scheimpflug analyzer in unoperated and post-refractive surgery eyes. J Cataract Refract Surg 2011; 37:302–309 Olsen T. Sources of error in intraocular lens power calculation. J Cataract Refract Surg 1992; 18:125–129 Norrby S. Sources of error in intraocular lens power calculation. J Cataract Refract Surg 2008; 34:368–376 ~ero DP, Saenz Gonza
lez C, Alio
JL. Intraobserver and interPin observer repeatability of curvature and aberrometric

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

PRECISION AND AGREEMENT: DUAL VERSUS SINGLE SCHEIMPFLUG CAMERA

19.

20.

21.

22.

23.

24.

25.

26.

measurements of the posterior corneal surface in normal eyes using Scheimpflug photography. J Cataract Refract Surg 2009; 35:113–120 Kobashi H, Kamiya K, Igarashi A, Ishii R, Sato N, Wang G, Shimizu K. Comparison of corneal power, corneal astigmatism, and axis location in normal eyes obtained from an autokeratometer and a corneal topographer. J Cataract Refract Surg 2012; 38:648–654 Karabatsas CH, Cook SD, Papaefthymiou J, Turner P, Sparrow JM. Clinical evaluation of keratometry and computerised videokeratography: intraobserver and interobserver variability on normal and astigmatic corneas. Br J Ophthalmol 1998; 82:637–642. Available at: http://www.pubmedcentral. nih.gov/picrender.fcgi?artidZ1722645&blobtypeZpdf. Accessed July 16, 2012 Goggin M, Patel I, Billing K, Esterman A. Variation in surgically induced astigmatism estimation due to test-to-test variations in keratometry. J Cataract Refract Surg 2010; 36:1792–1793
dis L Jr. Reliability and repeatSzalai E, Berta A, Hassan Z, Mo ability of swept-source Fourier-domain optical coherence tomography and Scheimpflug imaging in keratoconus. J Cataract Refract Surg 2012; 38:485–494 Ho J-D, Tsai C-Y, Tsai RJ-F, Kuo L-L, Tsai I-L, Liou S- W. Validity of the keratometric index: evaluation by the Pentacam rotating Scheimpflug camera. J Cataract Refract Surg 2008; 34:137–145 Shankar H, Taranath D, Santhirathelagan CT, Pesudovs K. Repeatability of corneal first-surface wavefront aberrations measured with Pentacam corneal topography. J Cataract Refract Surg 2008; 34:727–734 Nam SM, Im CY, Lee HK, Kim EK, Kim T-I, Seo KY. Accuracy of RTVue optical coherence tomography, Pentacam, and ultrasonic pachymetry for the measurement of central corneal thickness. Ophthalmology 2010; 117:2096–2103 Fu J, Wang X, Li S, Wu G, Wang N. Comparative study of anterior segment measurement with Pentacam and anterior segment optical coherence tomography. Can J Ophthalmol 2010;

27.

28.

29.

30.

1949

45:627–631. Available at: http://www.eyesite.ca/CJO/4506/ i10-068.pdf. Accessed July 16, 2012 Wang L, Mahmoud AM, Anderson BL, Koch DD, Roberts CJ. Total corneal power estimation: Ray tracing method versus Gaussian optics formula. Invest Ophthalmol Vis Sci 2011; 52:1716– 1722. Available at: http://www.iovs.org/content/52/3/1716.full. pdf. Accessed July 16, 2012 Pesudovs K. Wavefront aberration outcomes of LASIK for high myopia and high hyperopia. J Refract Surg 2005; 21:S508– S512 Gatinel D, Adam P-A, Chaabouni S, Munck J, Thevenot M, Hoang-Xuan T, Pieger S, Fujieda M, Bains HS. Comparison of corneal and total ocular aberrations before and after myopic LASIK. J Refract Surg 2010; 26:333–340 Wang L, Dai E, Koch DD, Nathoo A. Optical aberrations of the human anterior cornea. J Cataract Refract Surg 2003; 29:1514–1521

OTHER CITED MATERIAL € te GmbH A. Pentacam HR Instruction Manual. Oculus Optikgera 2010 B. Galilei Software Version 5.2. Addendum to the Operator Manual (v5.0.). Ziemer Ophthalmic Systems, 2010 C. Hill W, Wang L, Koch DD. IOL power calculation in eyes that have undergone LASIK/PRK/RK. Version 4.1. Available at: http://iolcalc.org/. Accessed July 16, 2012

J CATARACT REFRACT SURG - VOL 38, NOVEMBER 2012

First author: Jaime Aramberri, MD Begitek Clínica Oftalmol
ogica, San Sebasti
an, Spain