Agreement in anterior segment measurements between swept-source optical coherence and dual Scheimpflug tomography devices in keratoconus eyes

Abstract

Purpose

To evaluate the agreement of ocular biometry measured using a swept-source optical coherence (Casia 2) and a dual Scheimpflug (Galilei G6) tomography in keratoconus.

Methods

This retrospective study included 102 eyes from 102 keratoconus patient examined using both devices. Parameters compared included flat (Kf) and steep (Ks) keratometry, astigmatism of anterior, posterior, and total keratometry, central (CCT) and thinnest (TCT) corneal thickness. To assess the agreement, intraclass coefficient (ICC) and Bland-Altman analysis with 95% limits of agreement (LoA) were used.

Results

Anterior and total Ks and Kf showed moderate or good agreement with 95% limits of agreement (LoA) range over 9.75 D. Posterior Ks and Kf were lower in the Galilei G6 (all p < 0.001). Astigmatism showed moderate, poor, and moderate agreement for anterior, posterior, and total keratometry, respectively. CCT and TCT showed excellent agreement; however, the 95% LoA range was over 60 μm.

Conclusion

The agreement between the two devices was not excellent for most parameters used to diagnose keratoconus and assess disease progression, and the differences were clinically significant. Therefore, the measurements from these two devices are not interchangeable for patients with keratoconus.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-76031-7.

Introduction

In recent years, various optical biometers have been developed for the precise measurement of the curvature and thickness of the cornea using different methods, including slit-scanning and Placido disc technology, Scheimpflug topography, swept-source optical biometer, and swept-source anterior segment optical coherence tomography (AS-OCT)¹. To confirm the concordance and reproducibility of the measured values using the above equipment, previous studies has been conducted on healthy cornea^2–6. However, there is lack of data describing the performance of these devices when measuring corneas that deviate from normal parameters, such as keratoconus^2,7–9. In ectatic corneal diseases, including keratoconus, the accuracy and precision of the ocular biometric measurements are important for diagnosing the disease, determining its progression, and deciding the timing of surgical intervention^10,11.

Casia 2 (Tomey Corp., Nagoya, Japan), a recently introduced AS-OCT, is a swept-source optical coherence tomography (SS-OCT)-based device specifically designed for cross-sectional imaging of the anterior segment structure of the eye with high resolution and scan speed, allowing the anterior and posterior corneal curvature to be obtained¹². The Galilei dual Scheimpflug Analyzer (Galilei G6; Ziemer, Port, Switzerland) combines dual-rotating Scheimpflug cameras and a Placido disc to assess the anterior segment of the eye. Although both instruments have been found to exhibit high individual performance, systemic deviation or bias may occur when using different instruments, even for the same parameters. Furthermore, whether these devices are interchangeable is controversial^2,4–6,13. Moreover, there is no current study comparing Galilei G6 and Casia 2 in the keratoconus eye.

Therefore, this study aimed to assess difference and agreement in anterior segment parameters, including anterior, posterior, and total corneal keratometry (K), anterior chamber depth (ACD), central corneal thickness (CCT), and thinnest corneal thickness (TCT) measurements obtained from Galilei G6 and Casia 2 in keratoconus eye.

Materials and methods

This retrospective study included patients with keratoconus who visited Seoul National University Hospital (SNUH), Seoul, Republic of Korea, in between November 2021 and June 2024. All procedures were conducted in accordance with the Declaration of Helsinki, and the study design was approved by the Seoul National University Hospital Institutional Review Board (IRB No. 2207-095-1341). Owing to the retrospective design of the study and the use of de-identified patient information, the review board waived the need for written informed consent.

Patients were diagnosed with keratoconus based on typical slit-lamp findings (central corneal thinning, conical protrusion, Vogt striae or Fleisher ring) as well as topographic patterns (asymmetric bow-tie pattern with or without skewed axes)^13–15. The keratoconus was graded stage 1–4 according to Amsler-Krumeich (AK) classification¹¹.

Patients with a history of ocular trauma or surgery, corneal or conjunctival lesions, corneal transplantation, and those who failed the Galilei G6 or Casia 2 test were excluded. If both eyes met the criteria, the more severe staged eye was selected.

Study devices

Ocular biometric evaluations were conducted using Galilei G6 and Casia 2. All devices were placed in the same room. Every time we examined the patient, the room was constantly illuminated under 10 lx, as measured using a light meter (LX-1102, Lutron, Taiwan). Measurements were performed by a single examination specialist on the same day using a standard methodology. During the measurement, all patients were asked to keep their chin and forehead in position and look at the fixation light. Complete eye blinking was required before each automatic capture. A short break was allowed after each measurement. The measurement was accepted for analysis if the image quality status “OK” showed on the device screen. Otherwise, the measurement were repeated until a high-quality image was obtained. The device was realigned to its default position before subsequent measurements.

The measured parameters, including anterior K (ACP), anterior corneal astigmatism (ACA), posterior K (PCP), posterior corneal astigmatism (PCA), total K (TCP), total corneal astigmatism (TCA) (vector summation of anterior and posterior corneal astigmatism), ACD, CCT and TCT were recorded. A vector analysis for astigmatism was also evaluated in J0 and J45 . To compare the right and left eyes, we neutralized the enantiomorphism of the corneal astigmatism. The left-eye data were transformed by mirroring the vectors along the y-axis to avoid cancellation due to the sign when averaging the results¹⁷.

Statistical analysis

The main outcome measure was the agreement of the biometric parameters between the two devices. Intraclass correlation coefficients (ICC, two-way random, single measure) were calculated to assess the agreement between the measurements obtained from the devices. ICC is a widely used reliability index in reliability and agreement analyses. The ICC values range from 0 to 1, with 1 indicating perfect agreement. ICC was categorized as follows: < 0.50, poor reliability; 0.50 to < 0.75, moderate reliability; 0.75 to < 0.90, good reliability; and ≥ 0.90, excellent reliability¹⁸.

Bland-Altman plots were used to compare measurements between device pairs by plotting the differences between measurements against their means and the constructs of the limits of agreement (LoA). The 95% LoA (mean difference ± 1.96 x standard deviation) defines the range within which most differences between measurements from the two devices will lie.

Kolmogorov–Smirnov test was used to analyze normal data distributions. Biometric parameters were compared using the paired t-test for normally distributed datasets and the Wilcoxon rank test for non-normally distributed datasets. Statistical significance was set at p < 0.05. Statistical analyses were performed using SPSS (version 23.0) for Windows (IBM, Chicago, Illinois, USA) and Prism 10.1 (GraphPad Software, San Diego, CA, USA).

Results

A total of 102 eyes (37 right eye, 65 left eye) from 102 patients (68 males, 34 females) were included in this study. The mean age was 33.19 ± 11.33 years (range, 16–69 years). According to the AK keratoconus staging system, three patients (2.9%) were categorized in stage 1, 42 patients (41.2%) in stage 2, 21 patients (20.6%) in stage 3, and 36 patients (35.3%) in stage 4. The basic characteristics of the patients are presented in Table 1.

Table 1.

Demographics and clinical characteristics of patients.

Data
Number of patients, n	102
Male, n (%)	68 (66.7%)
Female, n (%)	34 (33.3%)
Number of eyes, n	102
Right, n (%)	37 (36.3%)
Left, n (%)	65 (63.7%)
Age (years old)	33.19 ± 11.33
Amsler-Krumeich keratoconus stage
Stage 1	3 (2.9%)
Stage 2	42 (41.2%)
Stage 3	21 (20.6%)
Stage 4	36 (35.3%)

Values are presented as the mean ± standard deviation.

Table 2 is a descriptive summary of the ocular biometric measurements taken by the two devices and their ICC values. Figure 1 demonstrates the Bland-Altman plots for each parameter. In the measurement of anterior steep and flat K, agreement was good (ICC = 0.884 and 0.803, respectively). However, the anterior steep and flat K showed a broad 95% LoA of -5.41 to 4.37 D and − 5.84 to 5.12 D, respectively as shown in Fig. 1a-b. Posterior steep and flat K was significantly different between the two devices (all p < 0.001) and showed good agreement (ICC = 0.851 and 0.794, respectively) (Table 2). For TCP and TCA, no significant differences were identified between the two devices (p > 0.05). However agreement of total steep and flat K was good (ICC = 0.860) and moderate (ICC = 0.686), respectively. They also showed a broad 95% LoA of -5.50 to 4.25 D and − 6.71 to 5.26 D, respectively (Fig. 1g-h). In the measurement of CCT, no significant difference was observed between the two devices (p = 0.484), however, there was a significant difference in TCT. Although the CCT and TCT measurements showed excellent agreement between the two devices (ICC = 0.944 and 0.937, respectively), they showed a broad 95% LoA ranges of -32.2 to 28.7 μm and − 30.5 to 39.5 μm, respectively, which are clinically unacceptable. ACD was also significantly different between the two devices (p < 0.001) and showed good agreement (ICC = 0.810).

Table 2.

Comparison and intraclass correlation coefficient of various biometric parameters between Casia 2 and Galilei G6.

	Casia 2 (range)	Galilei G6 (range)	Mean difference	p-value	ICC	Reliability	Confident limit
							Lower	Upper
Anterior
Ks, D	51.77 ± 5.60 (41.9–71.0)	51.26 ± 4.68 (41.9–63.0)	-0.53 ± 2.49	0.605	0.884	Good	0.831	0.920
Kf, D	47.46 ± 4.79 (38.6–64.9)	47.10 ± 4.17 (36.3–57.4)	-0.36 ± 2.79	0.067	0.803	Good	0.722	0.863
ACA, D	4.32 ± 2.77 (0.2–15.1)	4.17 ± 2.66 (0.4–14.0)	-0.17 ± 2.40	0.417	0.620	Moderate	0.485	0.727
J0	1.19 ± 1.78 (-3.20–6.67)	0.62 ± 1.93 (-4.67–4.79)	-0.57 ± 1.22	< 0.001	0.752	Good	0.590	0.845
J45	0.29 ± 1.40 (-3.54–3.88)	0.32 ± 1.38 (-3.35–5.19)	0.04 ± 1.15	0.533	0.662	Moderate	0.537	0.759
Posterior
Ks, D	-8.05 ± 1.15 (-11.2 - -1.2)	-8.36 ± 4.23 (-12.0 - -6.1)	-0.30 ± 0.60	< 0.001	0.851	Good	0.716	0.914
Kf, D	-7.28 ± 1.03 (-10.5 - -5.4)	-7.47 ± 1.20 (-10.7 - -3.7)	-0.19 ± 0.70	< 0.001	0.794	Good	0.703	0.859
PCA, D	-0.78 ± 1.06 (-2.7–8.7)	-0.89 ± 0.64 (-3.8–0.11)	-0.11 ± 0.70	0.281	0.255	Poor	0.068	0.426
J0	-0.22 ± 0.30 (-1.21–0.41)	-0.18 ± 0.41 (-1.84–0.67)	0.04 ± 0.34	0.063	0.533	Poor	0.379	0.658
J45	0.00 ± 0.26 (-0.70–0.70)	-0.19 ± 0.26 (-1.15–0.52)	-0.19 ± 0.29	< 0.001	0.319	Poor	0.060	0.522
Total
Ks, D	49.80 ± 5.32 (39.9–68.4)	49.18 ± 4.24 (39.9–61.5)	-0.63 ± 2.48	0.156	0.860	Good	0.796	0.905
Kf, D	45.75 ± 4.38 (37.1–62.4)	44.93 ± 3.63 (34.1–55.0)	-0.73 ± 3.05	0.763	0.686	Moderate	0.562	0.779
TCA, D	4.05 ± 2.73 (0.1–14.4)	4.25 ± 2.92 (0.4–14.4)	0.10 ± 2.34	0.850	0.593	Moderate	0.452	0.706
J0	1.05 ± 1.77 (-3.35–6.39)	0.48 ± 2.04 (-5.35–4.84)	-0.57 ± 1.36	< 0.001	0.716	Moderate	0.564	0.814
J45	0.36 ± 1.28 (-3.38–3.78)	0.18 ± 1.38 (-3.58–4.81)	-0.18 ± 1.04	0.013	0.689	Moderate	0.571	0.778
CCT, μm	466.37 ± 44.36 (300–559)	464.63 ± 48.35 (316–577)	-1.75 ± 15.52	0.484	0.944	Excellent	0.918	0.962
TCT, μm	430.71 ± 49.59 (278–527)	435.23 ± 53.51 (271–525)	4.52 ± 17.86	< 0.001	0.937	Excellent	0.906	0.958
ACD, mm	3.33 ± 0.30 (2.26–4.00)	3.13 ± 0.24 (2.18–3.52)	-0.10 ± 0.12	< 0.001	0.810	Good	0.450	0.915

ICC, intraclass correlation coefficient; K, keratometry; s, steep; f, flat; ACA, anterior corneal astigmatism; PCA, posterior corneal astigmatism; TCA, total corneal astigmatism; J0, Jackson cross-cylinder axes at 0°and 90°; J45, Jackson cross-cylinder axes at 45°and 135°; CCT, central corneal thickness; TCT, thinnest corneal thickness; ACD, anterior chamber depth; D, diopters.

Values are presented as the mean ± standard deviation.

The ICC values range from 0 to 1, with 1 indicating perfect agreement. ICC was categorized as follows: < 0.50, poor reliability; 0.50 to < 0.75, moderate reliability; 0.75 to < 0.90, good reliability; and ≥ 0.90, excellent reliability.

^*p-value Wilcoxon rank test.

Fig. 1.

The Bland-Altman plots of the (a) anterior Ks, (b) anterior Kf, (c) ACA, (d) posterior Ks, (e) posterior Kf, (f) PCA, (g) total Ks, (h), total Kf, (i) TCA, (j) CCT, (k) TCT, and (l) ACD. The solid line represents the mean difference, while the dotted lines on each side show upper and lower 95% LoA. K = keratometry; s = steep; f = flat; ACA = anterior corneal astigmatism; PCA = posterior corneal astigmatism; TCA = total corneal astigmatism; CCT = central corneal thickness; TCT = thinnest corneal thickness; ACD = anterior chamber depth; D = diopters; LoA = limits of agreement.

Generally, the mean differences in keratometric values were negative except for TCA, indicating that Casia 2 measured larger values than Galilei G6. Furthermore, in the anterior, posterior, and total corneal K (steep, flat, astigmatism), as the K value increased, the difference in the measured values ​​between the two instruments increased, and the variance of the deviation increased.

Next, we divided patients into two groups (AK stages 1–2 and 3–4) and analyzed their ocular biometric measurements accordingly (Table 3, supplementary Fig. 1). CCT, TCT, and ACD showed good to excellent reliability in both groups, however, keratometric values which showed moderate to good reliability in AK stages 1–2 group, were confirmed to have poor to moderate reliability in AK stage 3–4 group. Except for anterior Kf, total Kf, and TCT, there was no difference in the mean difference between the two measurements in the two groups. Excluding anterior Ks, the 95% LoA was wider in the AK stage 3–4 group.

Table 3.

Comparison and intraclass correlation coefficient of various biometric parameters between Casia 2 and Galilei G6 according to AK stage.

	Casia 2 (range)	Galilei G6 (range)	Mean difference	p-value	ICC	Reliability	Confident limit
							Lower	Upper
AK stage 1 or 2 (n = 45)
Anterior
Ks, D	49.12 ± 5.04 (41.9–70.9)	48.90 ± 4.07 (41.9–60.4)	-0.52 ± 2.53	0.968	0.847	Good	0.738	0.913
Kf, D	45.22 ± 3.36 (38.6–58.1)	44.77 ± 3.37 (36.3–53.1)	-0.45 ± 2.48	0.776	0.728	Moderate	0.554	0.841
ACA, D	4.21 ± 3.18 (0.2–15.1)	4.14 ± 2.65 (0.8–11.0)	-0.07 ± 2.00	0.904	0.766	Good	0.612	0.864
J0	1.31 ± 1.79 (-1.84–6.67)	1.00 ± 1.84 (-4.45–4.79)	-0.33 ± 1.13	0.390	0.831	Good	0.713	0.904
J45	0.02 ± 1.36 (-3.54–2.54)	0.22 ± 1.30 (-3.30–3.39)	0.20 ± 1.06	0.231	0.686	Moderate	0.493	0.814
Posterior
Ks, D	-7.52 ± 1.04 (-11.2 - -6.2)	-7.79 ± 1.09 (-11.9 - -6.1)	-0.26 ± 0.55	0.002 †	0.866	Good	0.770	0.924
Kf, D	-6.80 ± 0.80 (-9.7 - -5.7)	-6.94 ± 0.84 (-8.8 - -5.4)	-0.15 ± 0.61	0.009	0.726	Moderate	0.551	0.839
PCA, D	-0.73 ± 0.54 (-2.7 - -0.1)	-0.84 ± 0.57 (-3.6–0.1)	-0.11 ± 0.57	0.178†	0.502	Moderate	0.247	0.692
J0	-0.25 ± 0.28 (-1.21–0.22)	-0.21 ± 0.36 (-0.64–0.63)	0.05 ± 0.27	0.231†	0.661	Moderate	0.458	0.798
J45	0.03 ± 0.25 (-0.42–0.70)	-0.16 ± 0.25 (-0.73–0.52)	-0.20 ± 0.25	0.385†	0.518	Moderate	0.268	0.703
Total
Ks, D	47.62 ± 4.82 (39.9–68.3)	47.11 ± 3.74 (39.9–56.8)	-0.52 ± 2.38	0.679	0.848	Good	0.739	0.913
Kf, D	43.72 ± 3.03 (37.1–55.5)	43.05 ± 3.13 (34.1–50.6)	-0.67 ± 2.36	0.053	0.704	Moderate	0.520	0.826
TCA, D	3.90 ± 3.12 (0.4–14.4)	4.05 ± 2.71 (0.4–11.2)	0.15 ± 1.91	0.350	0.786	Good	0.641	0.876
J0	1.17 ± 1.84 (-2.10–6.39)	0.87 ± 1.89 (-4.27–4.78)	-0.30 ± 1.23	0.927	0.784	Good	0.638	0.875
J45	0.13 ± 1.34 (-3.38–2.43)	0.08 ± 1.30 (-3.58–3.12)	-0.05 ± 0.90	0.472	0.746	Moderate	0.581	0.852
CCT, μm	485.62 ± 33.46 (422–559)	483.84 ± 37.13 (421–557)	-1.78 ± 13.67	0.399	0.925	Excellent	0.868	0.958
TCT, μm	455.13 ± 34.74 (401–527)	457.20 ± 38.39 (381–525)	2.07 ± 17.81	0.333	0.882	Good	0.795	0.933
ACD, mm	3.26 ± 0.28 (2.26–3.78)	3.13 ± 0.23 (2.18–3.46)	-0.07 ± 0.10	< 0.001 †	0.924	Excellent	0.859	0.960
AK stage 3 or 4 (n = 57)
Anterior
Ks, D	53.65 ± 5.41 (44.7–71.0)	53.12 ± 4.29 (45.9–63.0)	-0.54 ± 2.48	0.579	0.871	Good	0.790	0.922
Kf, D	49.22 ± 5.01 (42.1–64.9)	48.94 ± 3.81 (41.6–57.4)	-0.29 ± 3.04	0.040	0.767	Good	0.634	0.856
ACA, D	4.43 ± 2.53 (0.2–11.4)	4.19 ± 2.69 (0.4–14.0)	-0.24 ± 2.69	0.510†	0.467	Poor	0.237	0.648
J0	1.09 ± 1.78 (-3.20–5.41)	0.32 ± 1.97 (-4.67–4.52)	-0.77 ± 1.31	< 0.001	0.758	Good	0.622	0.850
J45	0.49 ± 1.40 (-3.09–3.88)	0.40 ± 1.45 (-3.35–5.19)	-0.09 ± 1.21	0.815	0.642	Moderate	0.460	0.772
Posterior
Ks, D	-8.46 ± 1.06 (-11.1 - -6.9)	-8.81 ± 1.22 (-12.0 - -6.9)	-0.34 ± 0.64	< 0.001 †	0.846	Moderate	0.753	0.907
Kf, D	-7.66 ± 1.04 (-10.5 - -5.4)	-7.88 ± 1.28 (-10.7 - -3.7)	-0.23 ± 0.77	0.030 †	0.784	Moderate	0.659	0.867
PCA, D	-0.81 ± 0.45 (-1.8–0.5)	-0.92 ± 0.70 (-3.8 - -0.1)	-0.11 ± 0.79	0.789	0.081	Poor	-0.182	0.332
J0	-0.20 ± 0.32 (-0.82–0.41)	-0.16 ± 0.44 (-1.84–0.67)	0.03 ± 0.40	0.510†	0.459	Poor	0.228	0.642
J45	-0.02 ± 0.27 (-0.70–0.55)	-0.21 ± 0.27 (-1.15–0.34)	-0.19 ± 0.31	< 0.001 †	0.319	Poor	0.065	0.533
Total
Ks, D	51.53 ± 5.10 (43.0–68.4)	50.81 ± 3.91 (43.1–61.5)	-0.71 ± 2.57	0.128	0.838	Good	0.740	0.901
Kf, D	47.36 ± 4.63 (40.6–62.4)	46.59 ± 3.43 (39.1–55.0)	-0.77 ± 3.52	0.445	0.627	Moderate	0.440	0.762
TCA, D	4.17 ± 2.40 (0.1–11.4)	4.23 ± 2.93 (0.4–14.4)	0.06 ± 2.64	0.715	0.512	Moderate	0.293	0.681
J0	0.95 ± 1.72 (-3.35–5.40)	0.18 ± 2.12 (-5.35–4.84)	-0.78 ± 1.43	< 0.001	0.724	Moderate	0.573	0.828
J45	0.53 ± 1.29 (-2.98–3.78)	0.25 ± 1.46 (-3.49–4.81)	-0.28 ± 1.14	0.067†	0.657	Moderate	0.480	0.783
CCT, μm	451.18 ± 46.23 (300–532)	449.46 ± 51.00 (316–533)	-1.72 ± 16.96	0.796	0.939	Excellent	0.899	0.964
TCT, μm	411.42 ± 51.28 (278–501)	417.88 ± 57.55 (271–508)	6.46 ± 17.81	< 0.001	0.947	Excellent	0.911	0.968
ACD, mm	3.39 ± 0.31 (2.66–4.00)	3.14 ± 0.25 (2.55–3.52)	-0.12 ± 0.14	< 0.001	0.828	Good	0.688	0.908

ICC, intraclass correlation coefficient; K, keratometry; s, steep; f, flat; ACA, anterior corneal astigmatism; PCA, posterior corneal astigmatism; TCA, total corneal astigmatism; J0, Jackson cross-cylinder axes at 0°and 90°; J45, Jackson cross-cylinder axes at 45°and 135°; CCT, central corneal thickness; TCT, thinnest corneal thickness; ACD, anterior chamber depth; D, diopters.

Values are presented as the mean ± standard deviation.

The ICC values range from 0 to 1, with 1 indicating perfect agreement. ICC was categorized as follows: < 0.50, poor reliability; 0.50 to < 0.75, moderate reliability; 0.75 to < 0.90, good reliability; and ≥ 0.90, excellent reliability.

^*p-value Wilcoxon rank test.

Discussion

To the best of our knowledge, this is the first comparison of the performance of Galilei G6 and Casia 2 in the ocular biometry of keratoconus. The comprehensive analysis results showed that the agreement of most parameters related to corneal curvatures between the two devices was not excellent, with a broad 95% LoA. CCT and TCT showed excellent agreement, however, wide ranges of 95% LoA were also observed.

Periodic repetition of ocular biometric measurements is recommended in keratoconus as treatment options may differ as the disease progresses. However, there may be occasions when patients are unable to use the same device during the follow-up period (e.g., when visiting another hospital, when there are different types of machines in the hospital, or when hospital upgrade or replace their devices). Therefore, comparing the reliability and agreement between the devices is important in these contexts.

Several previous studies have investigated the agreement and repeatability of keratometry using the same measurement methods in normal corneas. A study comparing ACP, PCP, and TCP using Casia 2 and rotating Scheimpflug tomography (Pentacam high resolution [HR], Oculus, Wetzlar, Germany) in 49 healthy young subjects with an average age of 24.8 ± 4.36 years showed generally good agreement in those values¹⁹. In a study comparing Casia 2 and Galilei G6 in 100 normal subjects suitable for cataract surgery, no significant difference was found in TCP, but a significant difference was found in PCP²⁰. Including the previously mentioned study²⁰, even though studies that demonstrated statistically significant differences between the two measurement platforms in keratometry have found that the differences were less than 0.5 D, which were not significant for clinical application purposes^5,6,20,21. However, when it comes to keratoconus, the results varied in each study. In one study comparing the biometry of keratoconus using Casia2, Pentacam AXL (with axial length measurement [AXL], Oculus), and SS-OCT (IOLMaster 700, Carl Zeiss, AG, Jena, Germany), the agreement of TCP measured by three devices was excellent, although IOLMaster 700 and Casia 2 showed a difference of more than 1.0 D in TCP². In the study², the majority of patients (73.53%) were in the early stage of keratoconus belonging to stage 1, thus the results may differ from those of our study, which included patients in higher stages. In our study, unlike in previous studies, the agreement between ACP and TCP was moderate to good. In addition, since the with 95% limits of agreement (LoA) range over 9.75 D, the two devices were not considered clinically interchangeable. In another study involving patients with early keratoconus, the agreement between AS-OCT combined with Placido-disc topography (MS-39, Construzione Strumenti Oftalmici, Florence, Italy) and Placido-Scheimpflug imaging system (Sirius; Construzione Strumenti Oftalmici) showed agreement was moderate for TCT and ACD and poor for most other measured parameters⁹. It is well known that as the stages of keratoconus increase, the irregularity of the corneal curvature becomes severe and the possibility of stromal scars resulting from breaks in the Bowman layer increases; therefore, less repeatability, reproducibility, and inaccurate keratometry have been reported^11,22. Among studies conducted on higher grade of keratoconus, Flockerzi et al., concluded that although both devices appeared to be reliable, differences between the devices regarding the measured tomographic parameters indicated that the Pentacam HR and Casia 2 measurements were not interchangeable in keratoconus. They also concluded that the reliability of measurements decreases with disease severity⁷. Another study reported statistical differences in arithmetic values at all stages of keratoconus, but the differences were within the clinically acceptable level (< 0.5 D)⁸. Considering that the values ​​that match in healthy cornea do not in those of ectatic cornea, image quality or ectasia may affect measurements more than differences in hardware, software, or image acquisition between instruments.

As observed in previous studies⁶, the Casia 2 demonstrated higher keratometry values than the Galilei G6. This may be due to the different technologies and image analysis principles of the instruments. Casia 2 takes 32 measuring points from the central 3.2 mm area of the cornea and connects two points symmetrically centered on the corneal apex to form 16 straight lines. Among these 16 lines, the line with strongest D is steep K, and the radius of curvature in the direction with an angle of 90◦ with steep K is flat K. Using this principle, Casia 2 can calculate the anterior and posterior corneal curvatures and the real corneal curvature by paraxial calculation using the Gullstrand model eye refractive index. Casia 2 utilizes a wavelength of 1,310 nm, and its axial resolution is < 10 μm. In Contrast, the Galilei G6 combines 20 Placido rings with a dual-rotating Scheimpflug camera. Simulated keratometry is calculated from the 0.5 to 2.0 mm annular (semichord) zone and represented as diopters using a refractive index of 1.3375. The posterior mean K is calculated using a refractive index of 1.376 for the cornea and 1.336 for the aqueous humor over a 4 mm diameter area. In addition to differences in measurement methods, changes in corneal scarring that may occur as keratoconus progresses can cause strong reflections at the air/cornea interface, making it difficult to accurately identify the edge. Moreover, posterior surface evaluation is hindered by the errors on these front surfaces. Therefore, it is believed that differences in arithmetic values can occur between the two instruments when using different measurement methods.

When planning crosslinking²³ or implantation of intracorneal ring segments²⁴ as a treatment option of keratoconus, corneal pachymetry is important. Thus, several studies investigated the agreement and reliability of corneal thickness measured using Scheimpflug tomography and AS-OCT. In a comparative study between MS-39 and Sirius, TCT showed a statistically significant difference, but the difference was 6.58 ± 14.74 μm, which has little impact from a clinical point of view⁹. A study comparing CCT with Casia 2 and Pentacam AXL also revealed a difference of < 5 μm². In a study using AS-OCT (Anterion, Heidelberg Engineering GmbH, Heidelberg, Germany) and Pentacam HR, the difference was < 5 μm for CCT and < 10 μm for TCT for both mild and moderate keratoconus⁸. However, in another study of keratoconus using Casia SS-1000 (an earlier version of Casia 2, AS-OCT) and Pentacam HR, the differences were 27.54 ± 23.02 μm and 18.46 ± 25.14 μm, respectively, which is much larger than those reported in any other studies. This result was attributed to the fact that Casia SS-1000 measured these values manually, not automatically¹⁵. Previous studies have reported that TCT reduction of > 20 μm⁶, 10 μm²⁵, or 5% ²⁶ can be considered disease progression factors. TCT measurement differences in between two devices > 20 μm and 10 μm were 10.29% and 51.47%, respectively in our study. Considering wide 95% LoA and reduction in TCT of > 10 μm as criteria for keratoconus, although the CCT and TCT showed high agreement and excellent reliability between the two devices in this study, caution is required when using devices interchangeably.

This study had certain limitations. First, our sample comprised of patients with relatively severe keratoconus, which could have potentially impacted the results. Second, we did not perform repeated measurements for each device to assess the repeatability of the results. However, we included only high-quality measurements in this study. Third, all the participants were of Asian descent. Consequently, the generalizability of our findings to other ethnicities is limited.

Conclusion

This study analyzed the agreement between Casia 2 and Galilei G6 in keratoconus. For most parameters used to diagnose keratoconus and assess disease progression, the agreement between the two devices was not excellent and the differences were clinically significant. Therefore, it is not recommended to use the measurements of these two devices interchangeably for patients with keratoconus.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

YL participated in analysis and interpretation of data, and writing the manuscript. JYO, HJC, and MKK participated in conceptualization, drafting, and revising the manuscript. CHY participated in conceptualization, study design, analysis and interpretation of data, and writing and revising the manuscript. All authors have approved the final submitted version and have agreed to be personally accountable for each contribution.

Funding

None.

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

This retrospective study adhered to the ethical standards of the Declaration of Helsinki and was approved by the Seoul National University Hospital Institutional Review Board (IRB No. 2207-095-1341).

Consent for publication

The informed consent from patients was waived by the Seoul National University Hospital Institutional Review Board because this study was based on the retrospective chart review and involves no more than minimal risk.