The recent introduction of the spectral-domain optical coherence tomography (OCT) has allowed imaging of the retinal nerve fiber layer (RNFL) at a higher scan speed in a higher resolution than the time-domain OCT. The Cirrus HD-OCT (Carl Zeiss Meditec Inc, Dublin, Calif) is a commercially available spectral-domain OCT instrument. With a scan speed of 27,000 axial measurements per second, the Cirrus HD-OCT captures an area of 6 × 6 mm2 (composed of 200 × 200 measurement pixels) around the optic nerve head in approximately 1.5 seconds.1 An RNFL thickness map is constructed for 3-dimensional visualization of the RNFL distribution (Fig. 1A). The RNFL thickness at each superpixel (4 × 4 pixels) is then analyzed with reference to a built-in, age-matched normative data to generate an RNFL thickness deviation map (Fig. 1B). A superpixel would be indicated in yellow or red if the RNFL measurement falls within the lower 95% to 99% or the lower 99% normal distribution, respectively. Although analysis of the RNFL map may facilitate detection of RNFL defect that may not be recognized in optic disc stereophotograph, it is unclear if the quality of OCT images would affect the interpretation of the RNFL map. A quantifiable measure of the quality of OCT images is the signal strength (SS). It ranges between 0 (worst quality) and 10 (best quality). Media opacity (eg, cataract), corneal surface irregularity (eg, corneal drying related to infrequent blinking), and poor image focus could result in poor image quality and falsely lower the value of RNFL thickness.2–6 Although the manufacturer recommends an SS of at least 5 for RNFL imaging in the Cirrus HD-OCT,1 the scientific basis for this recommendation has not been validated. The purposes of this study were to establish the relationship of SS and RNFL thickness in the Cirrus HD-OCT and to investigate the influence of SS on interpretation of the RNFL map.
MATERIALS AND METHODS
Thirty-seven healthy Chinese subjects who met the inclusion criteria were recruited in this study (see Statistical Analysis for sample size calculation). We included only normal subjects for the reason of avoiding potential RNFL abnormities that may confound the analysis of the association between SS and RNFL thickness. All subjects were examined at the University Eye Center, The Chinese University of Hong Kong, from January 2008 to December 2008. Other than refractive error, all included eyes had no concurrent ocular pathological features, had best corrected visual acuity of at least 20/40, spherical refractive error within the range of −6.00 to +4.00 diopters (D), and a normal visual field (described below). Subjects with parapapillary atrophy extending more than 1.7 mm from the center of the disc (the radius of RNFL measurement circle), intraocular pressure more than 21 mm Hg, visual field defects (see Visual Field Testing), history of intraocular surgery, neurological disease, or diabetes were excluded. In a randomly selected eye in each subject, Cirrus HD-OCT (Carl Zeiss Meditec Inc) imaging was performed. The study was conducted in accordance with the ethical standards stated in the Declaration of Helsinki and was approved by the local clinical research ethics committee. Informed consent was obtained from each participant.
Visual Field Testing
Standard visual field testing was performed with the static automated white-on-white threshold perimetry (Swedish interactive threshold algorithm standard 24-2, Humphrey Field Analyzer II; Carl Zeiss Meditec Inc). A visual field was defined as reliable when fixation losses were less than 20%, false-positive and false-negative rates were less than 15%. A visual field defect was defined as having 3 or more significant (P < 0.05) nonedge contiguous points with at least 1 at the P < 0.01 level on the same side of the horizontal meridian in the pattern deviation plot and classified as outside normal limits in the Glaucoma hemifield test. All subjects in this study had a reliable and normal visual field without visual field defect.
Cirrus HD-OCT Imaging
Pupils were dilated during OCT imaging. The Optic Disk Cube 200 × 200 scan protocol was used to generate a cube of data in a 6 × 6-mm2 grid with 200 × 200 axial measurements. The “center” and “enhance” modes were used to optimize the Z-offset and scan polarization, respectively. An SS indicator from 0 (poor image quality) to 10 (best image quality) was displayed in the “review screen” after a scan was captured. Images with different SSs (5, 6, 7, 8, 9, and 10) were acquired sequentially in the same eye by adjusting the scan polarization with a slide bar in the operation screen manually. A 10-second break was given between the scans for blinking and resting. After each capture, motion artifact was checked with the line scanning ophthalmoscope image with the OCT en face overlaid. Rescanning was performed if motion artifact was detected. The RNFL tomogram was checked for RNFL segmentation. Individuals would not be included if a complete set of OCT images with the required SSs was not obtainable. Two subjects were excluded in the analysis because of repeated motion artifact detected in the image series.
Measurement of RNFL Thickness
A built-in algorithm automatically detected the optic disc center and positioned a calculation circle of diameter 3.46 mm around the optic disc on the RNFL thickness map. Clock hour, quadrant, and average RNFL thicknesses were derived from the calculation circle and displayed in the analysis printout. Measurements were aligned based on right-eye orientation. Twelve o’clock corresponds to the superior region, 3 o’clock corresponds to the nasal region, 6 o’clock corresponds to the inferior region, and 9 o’clock corresponds to the temporal region, for both right and left eyes. The RNFL thickness deviation map (50 × 50 superpixels) was generated with reference to a built-in, age-matched normative data.7 A superpixel would be indicated in yellow if the RNFL measurement falls within 95% to 99% of the normal distribution or in red if it is less than the 99% normal distribution. The RNFL thickness deviation map was exported, and the number of abnormal superpixels (in yellow or red) was counted manually with the aid of an image analysis software (Sigma Scan Pro version 5.0; Systat Software Inc, Point Richmond, Calif).
Statistical analysis was performed using SPSS version 17.0 (SPSS, Inc, Chicago, Ill). In a previous study, the average RNFL thickness measured by the Cirrus HD-OCT in normal eyes was 96.81 (SD, 8.88) μm, and the intervisit repeatability was 5.12 μm.8 Assuming detecting a difference of 6 μm with an SD of 9 μm between images with different levels of SS was considered clinically significant, the minimal sample size required for an α of 0.05 and a β of 0.20 was 36. The correlations between SS and clock hour, quadrant, average RNFL thicknesses, and the number of abnormal superpixels were computed and expressed as Spearman correlation coefficient. The RNFL thicknesses and the number of abnormal superpixel at different SSs were compared with repeated-measures analysis of variance. Bonferroni correction was applied for multiple comparisons. P < 0.05 was considered statistically significant.
A total of 37 eyes from 37 normal Chinese subjects (28 women and 12 men) were included in the analysis. The mean age, axial length, spherical equivalent, and visual field mean deviation were 45.20 (SD, 11.17) years (range, 23–64 years), 23.49 (SD, 1.23) mm (range, 21.4–26.9 mm), −0.43 (SD, 1.71) D (range, −5.88 to 2.25 D), and −0.58 (SD, 1.00) dB (range, −2.34 to 2.06 dB), respectively.
Table 1 shows the clock hour, quadrant, and average RNFL thicknesses measured at each level of SS. Significant differences between RNFL thicknesses measured at SS of 10 and at other SSs were evident at the superonasal (1–3 o’clock) and inferotemporal (7–9 o’clock) sectors (repeated-measures analysis of variance, all with P ≤ 0.001). The RNFL thicknesses generally increased with the SS, although significant differences between images with SS of 8, 9, and 10 were observed only at 7 o’clock and in the average RNFL thickness. Figure 2 shows the frequency distribution of abnormal clock-hour RNFL measurements in red “outside normal limits” or yellow “borderline” obtained at an SS of 5. The nasal sector (1–5 o’clock), in particular at 1 o’clock, was the most frequent location that showed abnormal RNFL measurements. More than 10% of subjects had abnormal RNFL measurement at 1 o’clock.
The number of abnormal superpixels (in yellow or red) in the RNFL thickness deviation map was greatest in images with an SS of 5 [24.7 (SD, 17.6)] and generally reduced with each level increase in SS. However, there were no significant differences in the number of abnormal superpixels between images with an SS of 10 [8.45 (SD 10.9)] and those with an SS of 8 [8.60 (SD, 9.48)] or 9 [7.77 (SD 10.2)] (Table 1 and Fig. 3). Figure 4 illustrates the effect of different SSs on the RNFL thickness deviation map in a normal eye.
Correct interpretation of OCT RNFL measurement relies on good-quality OCT images. In agreement with previous studies, we found that SS is an important determinant of parapapillary RNFL thickness.4–6,9 In addition, we showed that the level of SS had a significant impact in the interpretation of the RNFL thickness deviation map. Images with SS less than 8 had a significant increase in the number of abnormal superpixels.
In a previous study, we showed that RNFL thickness was positively correlated with SS in the Stratus OCT. Significant differences in RNFL thickness were evident between images obtained with an SS of 10 and those obtained at lower levels.5 In the Cirrus HD-OCT, SS represents the average of signal intensity of the 200 × 200 pixels in the RNFL thickness map. Unlike the Stratus OCT, no significant differences in RNFL measurements were observed between images with SS of 10 and those with SS of 8 or 9 in most clock hours (all the differences were <4 μm) (Table 1). Likewise, there was no significant difference in the number of abnormal superpixels between images with SS of 10 and those with SS of 8 or 9 (Fig. 3). The differences in RNFL measurements between images with different levels of SS were also smaller in the Cirrus HD-OCT than in Stratus OCT. For example, the difference in average RNFL thickness was 4.24 μm between images with SS of 10 and 5 in the Cirrus HD-OCT, whereas it was 12.58 μm in the Stratus OCT. The apparent disparity in the relationship between SS and RNFL thickness could be related to the intrinsic differences in the measurement scale and the scanning protocol between the OCT instruments. It has been shown that the RNFL thickness measured by the Cirrus HD-OCT is generally smaller than that by the Stratus OCT and that the difference increases when the RNFL thickness increases.8,10,11 It is plausible that the different measurement scales may result in different degrees of association between SS and RNFL thickness. With regard to the scanning protocol, the Stratus OCT measures the parapapillary RNFL using a circle scan. Signal strength represents an average measure of image quality over 256 A-scans. In contrast, the Cirrus HD-OCT collects RNFL measurements in an area of 6 × 6 mm2 in which SS represents the image quality of 40,000 (200 × 200) A-scans as a whole. There may be differences in image quality at different regions of the scan, and the SS reported in the Cirrus HD-OCT may not directly signify the signal quality of the RNFL at the 3.46-mm circle.
Signal strength is an indicator of the signal-to-noise ratio in the OCT images. The boundaries of the RNFL are detected by the OCT software with reference to a certain level of threshold reflectivity (signal intensity). If the SS is low, the signal intensity of the RNFL would be reduced. This may result in a thinner RNFL measurement. As illustrated in Figure 4, the RNFL thickness deviation map can appear abnormal in a normal eye when the SS is low. Of note, “abnormal” RNFL thicknesses in normal eyes were most often found at the nasal clock hours (Fig. 2). As the RNFL defects in glaucomatous eyes are most often found at the superotemporal or inferotemporal sectors, the presence of abnormal superpixels at the nasal sector may not confound the detection of glaucomatous damage, although careful interpretation of RNFL measurement is always required. Further investigation is necessary to determine the underlying causes for the predilection of abnormal RNFL measurement at the nasal sector. Although media opacity and corneal surface irregularity are common causes of poor SS, it is difficult to standardize and vary these parameters to establish the precise relationship between SS and RNFL thickness in an individual eye. For this reason, RNFL measurements with different SSs were obtained in the same eyes by adjusting the level of polarization. Of note, there is no direct evidence suggesting that the association between RNFL measurement and SS changes induced by polarization adjustment and media opacity is the same. Nevertheless, as illustrated in the case example, there is a similar inverse relationship between SS and the number of abnormal pixels in the RNFL thickness deviation map. Our finding not only illustrates the importance of optimizing the polarization for OCT imaging, but also provides a useful reference for clinician to interpret the influence of SS on RNFL measurement.
In summary, the level of SS has a significant impact on the measurement of RNFL thickness in the Cirrus HD-OCT. It is important to take the level of SS into consideration for interpretation of RNFL measurement in the RNFL thickness deviation map.
1. Carl Zeiss Meditec Inc. Cirrus HD-OCT User Manual
. Dublin, CA: Carl Zeiss Meditec Inc; 2008. 4.18-9. Rev. A.
2. Stein DM, Wollstein G, Ishikawa H, et al.. Effect of corneal drying on optical coherence tomography. Ophthalmology. 2006; 113: 985–991.
3. Savini G, Zanini M, Barboni P. Influence of pupil size and cataract on retinal nerve fiber layer thickness measurements by Stratus OCT. J Glaucoma. 2006; 15: 336–340.
4. Wu Z, Vazeen M, Varma R, et al.. Factors associated with variability in retinal nerve fiber layer thickness measurements obtained by optical coherence tomography. Ophthalmology. 2007; 114: 1505–1512.
5. Cheung CY, Leung CK, Lin D, et al.. Relationship between retinal nerve fiber layer measurement and signal strength
in optical coherence tomography. Ophthalmology. 2008; 115: 1347–1351.
6. Wu Z, Huang J, Dustin L, et al.. Signal strength
is an important determinant of accuracy of nerve fiber layer thickness measurement by optical coherence tomography. J Glaucoma. 2009; 18: 213–216.
7. Gurses-Ozden R, Durbin M, Callan T, et al.. Distribution of retinal nerve fiber layer thickness using Cirrus HD-OCT spectral domain technology. Invest Ophthalmol Vis Sci. 2008; 49. E-abstract 4632.
8. Leung CK, Cheung CY, Weinreb RN, et al.. Retinal nerve fiber layer imaging
with spectral-domain optical coherence tomography
: a variability and diagnostic performance study. Ophthalmology. 2009; 116: 1257–1263.
9. Balasubramanian M, Bowd C, Vizzeri G, et al.. Effect of image quality on tissue thickness measurements obtained with spectral domain-optical coherence tomography. Opt Express. 2009; 17: 4019–4036.
10. Sung KR, Kim DY, Park SB, et al.. Comparison of retinal nerve fiber layer thickness measured by Cirrus HD and Stratus optical coherence tomography. Ophthalmology. 2009; 116: 1264–1270.
11. Knight OJ, Chang RT, Feuer WJ, et al.. Comparison of retinal nerve fiber layer measurements using time domain and spectral domain optical coherent tomography. Ophthalmology. 2009; 116: 1271–1277.