To evaluate and compare central corneal thickness measurements using rotating Scheimpflug camera, scanning-slit topography, and ultrasound pachymetry in virgin, healthy corneas.
Prospective, observational, cross-sectional study.
Central corneal thickness in 157 healthy eyes of 157 patients without ocular abnormalities other than refractive errors was measured, in a sequential order, once with rotating Scheimpflug camera and scanning-slit topography and 3 times with ultrasound pachymetry as the last part of examination. All measurements were performed by a single experienced examiner. The results from scanning-slit topography are given with and without correction for “acoustic correction factor” of 0.92.
The average measurements of central corneal thickness by rotating Scheimpflug imaging, scanning-slit pachymetry, and ultrasound were 537.15 ± 32.98 μm, 542.06 ± 39.04 μm, and 544.07 ± 34.75 μm, respectively. The mean differences between modalities were 6.92 μm between rotating Scheimpflug and ultrasound ( P < .0001), 2.01 μm between corrected scanning-slit and ultrasound ( P = .204), and 4.91 μm between corrected scanning-slit and rotating Scheimpflug imaging ( P = .001). According to Bland-Altman analysis, highest agreement was between ultrasonic and rotating Scheimpflug pachymetry.
In the assessment of normal corneas, rotating Scheimpflug topography measures central corneal thickness values with higher agreement to ultrasound pachymetry.
High accuracy in the measurement of central corneal thickness (CCT) has become increasingly relevant in ophthalmic practice. One of the applications with utmost importance is the measurement of CCT in the preoperative evaluation for refractive surgery. In addition, CCT measurement is especially important in the field of glaucoma, as accurate intraocular determination may require correction of applanation tonometry readings according to CCT; moreover, CCT could be an independent risk factor for glaucoma.
Until recently, ultrasonic pachymetry was the most commonly used and relied-upon clinical method to measure CCT, and it is widely accepted as the gold standard for corneal thickness measurement. Although handheld ultrasound-based systems offer the advantages of portability and relative ease of use, this technique has several important limitations. First of all, ultrasonic pachymetry requires probe-corneal contact, so measurement may yield slightly thinner readings as a result of tissue indentation. The probe could displace tear film and this may contribute to inaccurate measurement. In addition, it is proposed that using ultrasonic pachymetry, the posterior corneal reflection point might be located between the Descemet membrane and anterior chamber, resulting in inaccurate thickness measurements. Moreover, placement of the probe exactly on the center of the cornea is highly operator dependent and challenging in many instances. Needs for topical anesthesia, mild patient discomfort, risk of epithelial erosions, and the risk of infection transmission are additional concerns when a contact technique is applied.
In recent years, several noninvasive alternative techniques have been introduced that are noncontact, user friendly, and less dependent on operator skills. These methods, including optical techniques such as rotating Scheimpflug camera (Pentacam, OCULUS Optikgerate GmbH, Wetzlar, Germany) and scanning-slit topography (Orbscan, Bausch & Lomb, Rochester, New York, USA), are used in many clinical centers for preoperative measurement of corneal thickness in keratorefractive surgery candidates.
Pentacam provides the potential to model the anterior chamber and also has high accuracy in showing irregularities such as keratoconus. The system is based on a rotating Scheimpflug camera, which scans and measures the complete cornea and anterior chamber parameters in approximately 2 seconds.
Scanning-slit topography provides a topographic map of the corneal anterior and posterior surface and measures the thickness between the air–tear film interface and the posterior corneal surface. The device provides a corneal thickness map and determines the thinnest point.
However, several studies have shown significant differences in CCT measurements obtained using these 2 instruments and with ultrasonic pachymetry. In addition, several of these studies have limited sample size with methodological and statistical drawbacks. Therefore, the story of choosing the best method for the measurement of CCT and the possibility of using different methods in the follow-up of a patient is not yet clear.
In the present study, we evaluated the commonly used optical methods of CCT measurement and compared them with ultrasonic pachymetry as the gold standard and most commonly used clinical method to measure CCT.
Materials and Methods
This was a cross-sectional study comparing central corneal thickness measured by scanning-slit corneal topography, rotating Scheimpflug camera, and ultrasonic pachymetry. We investigated the measurements in 157 eyes of 157 healthy candidates for keratorefractive surgery.
Subjects with nonrefractive ocular abnormalities, including corneal pathology, history of any previous ocular surgery, ocular trauma, and risk factors for corneal edema, were excluded.
During morning office hours, a single experienced optometrist, under supervision of an ophthalmologist, performed all measurements.
Measurements were made using optical techniques including rotating Scheimpflug camera (Pentacam, OCULUS Optikgerate GmbH, Wetzlar, Germany) and scanning-slit topography (Orbscan IIz, Bausch & Lomb, Rochester, New York, USA), first followed by ultrasonic pachymetry (Tomey, Nagoya, Japan). The measurements were sequential, without significant time interval between different techniques. Ultrasonic pachymetry measurements were scheduled last to prevent influence on the other modalities because of possible corneal indentation. All measurements were taken on both eyes of subjects as parts of routine preoperative evaluation for keratorefractive surgery candidates in our center; however, only 1 eye per subject was randomly selected for final analysis in our study.
Rotating Scheimpflug imaging was performed with the patient seated using a chin rest and forehead strap. The patient was asked to look at a fixation target for about 1.5 to 2 seconds, and 1 measurement was taken on the cornea.
Immediately after the rotating Scheimpflug camera measurements, corneal thickness was measured by scanning-slit topography. Subjects were seated in typical position using the chin rest; the instrument was aligned and the cornea was scanned. For scanning-slit topography, in addition to analyzing raw pachymetry data, the acoustic equivalent correction factor was used to achieve equivalence with the ultrasonic evaluation, and was set at 0.92 as recommended by the manufacturer. After scanning-slit topography measurement, CCT was measured using a handheld ultrasonic pachymeter. The cornea was anesthetized with topical tetracaine 0.5% (Anestocaine; Sina Darou, Tehran, Iran), and 3 consecutive measurements were made by ultrasonic pachymeter. The patient was brought into a face-up position on the examination chair and asked to fixate on a target on the ceiling. The probe of the ultrasonic pachymeter was placed manually on the center of the cornea as precisely as possible. CCT was recorded as the mean of 3 measurements. The instrument was calibrated frequently, according to the manufacturer recommendations.
Normal distribution of CCT measurements with the 3 methods was evaluated using Kolmogorov-Smirnov test. Magnitude of correlation between different measurement techniques was evaluated using intraclass correlation coefficient (ICC) and significance of differences between methods was examined with paired t test. A Bonferroni adjustment to the alpha level was used for multiple comparisons. Bland-Altman method and plots were used to demonstrate the between-method agreements. All of the statistical analysis was carried out using SPSS version 16.0 (SPSS Inc, Chicago, Illinois, USA). The significance level was set at P < .05 level. The calculated power of study exceeds 90% for most comparisons.
One hundred fifty-seven eyes of 157 patients were included in the study. Subjects’ age in study population was 24.88 ± 4.37 years (range, 20–40 years). Ninety-nine participants (63%) were women. There were 87 right eyes (55.4%) in the final analysis. The average measurements of CCT by rotating Scheimpflug camera, scanning-slit topography, and ultrasound pachymetry were 537.15 ± 32.98 μm, 542.06 ± 39.04 μm, and 544.07 ± 34.75 μm, respectively. Without correction with acoustic factor, average CCT measurement by Orbscan was 589.19 ± 42.43 μm. According to 1-sample Kolmogorov-Smirnov test, CCT measurements with all 3 techniques exhibit a normal distribution ( P > .3 for all); the difference between each pair of measurement techniques also had a normal distribution ( Figure 1 ).
To investigate agreement and correlation between different measurement methods, intraclass correlation analysis was used. There was a correlation of 0.88 between ultrasonic and Pentacam pachymetries, which was highly statistically significant ( P < .0001). Similar analysis demonstrated a correlation of 0.86 between ultrasonic and Orbscan pachymetries after correction with acoustic factor as recommended by the manufacturer, with considerable statistical significance ( P < .0001). Without acoustic factor, their correlation coefficient was 0.84 ( P < .0001). In addition, there was a correlation of 0.88 between corrected Orbscan and Pentacam pachymetry; the correlation was statistically significant ( P < .0001). Without correction with acoustic factor, there was still a high correlation of 0.86 ( P < .0001) between Orbscan and Pentacam.
In Bland-Altman analysis of agreement, although the mean difference between Orbscan and ultrasonic pachymetry was the least (2 μm), the highest agreement in measurement with narrowest 95% limit of agreement was detected between Pentacam and ultrasonic pachymetry, followed by high agreement between Orbscan and Pentacam pachymetry; on the other hand, the least agreement was between Orbscan and ultrasonic pachymetry, corresponding to the widest range in 95% limit of agreement in Bland-Altman plotting. Considering uncorrected Orbscan data without acoustic factor correction, there was still least agreement between Orbscan and other pachymetry methods ( Table , and Figure 2 ).
|Pentacam – Ultrasonic Pachymetry||Corrected Orbscan – Ultrasonic Pachymetry||Corrected Orbscan – Pentacam||Uncorrected Orbscan – Ultrasonic Pachymetry||Uncorrected Orbscan – Pentacam|
|SD of difference||16.16||19.73||17.90||21.49||19.94|
|Mean + 1.96 SD (95% CI)||24.75 (22.52 to 26.98)||36.66 (33.93 to 39.39)||39.99 (37.52 to 42.46)||87.25 (84.28 to 90.22)||91.12 (88.37 to 93.87)|
|Mean − 1.96 SD (95% CI)||−38.59 (−40.82 to −36.36)||−40.68 (−43.41 to −37.95)||−30.17 (−32.64 to −27.70)||3.01 (0.04 to 5.98)||12.95 (10.20 to 15.70)|
|95% LoA width||63.34||77.34||70.16||84.24||78.12|
|95% prediction interval b||−37.86 to 37.86||−49.04 to 49.04||−46.12 to 46.12||−48.28 to 48.28||−42.53 to 42.53|
|95% PI width||75.72||98.08||92.24||96.56||85.06|
|LoA – PI difference||12.38||20.74||22.08||12.32||6.94|
In paired t testing, although we found high levels of agreement between different measurement techniques, there was a statistically significant difference between Pentacam and ultrasound ( P < .0001) and between Pentacam and corrected Orbscan ( P = .001). The difference between Pentacam and uncorrected Orbscan measurements was also statistically significant ( P < .0001). The least difference was between corrected Orbscan and ultrasonic pachymetry ( P = .204); however, without correction for acoustic factor, uncorrected Orbscan and ultrasonic pachymetry yielded statistically significantly different measurement ( P < .0001). The significance levels for all comparisons were well below the adjusted P value of .016 after Bonferroni alpha adjustment for multiple comparisons. As mentioned previously, least difference does not indicate highest agreement, as the former indicates that the mean difference is least, while the true variability in either direction could be high.
Although methodologically questionable, regarding the popularity of the method in similar literatures, we also performed regression analysis on CCT measurement by different techniques. There was high correlation between CCT values obtained by different techniques; the regression line for corrected Orbscan–ultrasound pachymetry correlation was the closest to equality ( Figure 3 ). FLOAT NOT FOUND
To be statistically more precise, we also calculated the 95% prediction interval and compared this limit with the 95% limit of agreements (Table, Figure 4 ). The closest relationship between 95% prediction interval and 95% limit of agreements was observed for Pentacam–uncorrected Orbscan comparison, followed by ultrasonic pachymetry–uncorrected Orbscan comparison. In this regard, Pentacam and ultrasonic pachymetry also had a close 95% prediction interval and 95% limits of agreement. FLOAT NOT FOUND