Article Text

Download PDFPDF

Direct comparison of spectral-domain and swept-source OCT in the measurement of choroidal thickness in normal eyes
  1. Sergio Copete1,
  2. Ignacio Flores-Moreno1,
  3. Javier A Montero2,
  4. Jay S Duker3,
  5. José M Ruiz-Moreno1,4
  1. 1Department of Ophthalmology, Castilla La Mancha University, Albacete, Spain
  2. 2Ophthalmology Unit, Pio del Rio Hortega University Hospital, Valladolid, Spain
  3. 3New England Eye Center, Tufts Medical Center, Boston, Massachusetts, USA
  4. 4Vitreo-Retinal Unit, Alicante Institute of Ophthalmology, VISSUM, Alicante, Spain
  1. Correspondence to Dr Sergio Copete Piqueras, Department of Ophthalmology, Castilla -La Mancha University, 14th Almansa Avenue, Albacete 02006, Spain; sergiocp.ab{at}gmail.com

Abstract

Objective To compare spectral-domain optic coherence tomography (SD-OCT) and swept-source OCT (SS-OCT) in the study of choroidal thickness (CT) in healthy eyes.

Methods Prospective, cross-sectional, single-centre study. 82 healthy eyes of 46 patients were included. In a single session, Topcon 3D-2000 SD-OCT and 1050 nm SS-OCT prototype devices were used to perform OCT scans using a single line protocol. Two masked investigators independently, manually determined 13 CT measurements consisting of one subfoveal (SFCT), and six measurements on either side of the fovea (nasal and temporal) taken every 500 microns apart. The mean CT (MCT) was the mean average of these 13 measurements.

Results SD-OCT was able to reproducibly measure the CT in 74.4% of eyes vs 100% with SS-OCT (p<0.05; Fisher's Exact test). In those eyes measured by both systems, mean SFCT was 279.4±96.9 μm (range, 84–506) with SD-OCT vs 285.7±88.9 μm (range 130–527) with SS-OCT (p=0.11; Student's t test paired data). Mean MCT was 243.8±78.8 μm (range 103.6–433.2) with SD-OCT vs 242.2±81.8 μm (range 97.6–459) with SS-OCT (p=0.64; Student's t test paired data). The difference in SFCT and MCT was not statistically significant between both devices. Intraclass correlation coefficient was higher than 0.9 interobserver and interdevice measurements. SFCT Bland–Altman plots showed 95% interobserver measurement agreement within ±34 for SD-OCT, ±22 for SS-OCT and ±60 μm intersystems.

Conclusions SS-OCT permitted accurate identification of the choroido-scleral border in 100% of normal eyes, suggesting that SS-OCT was the superior modality for the measurement of CT.

  • Anatomy
  • Choroid
  • Imaging
  • Macula

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Since Huang et al introduced optic coherence tomography (OCT) technology in 1991, continuous hardware and software developments have resulted in higher resolution while decreasing the time of acquisition.1 OCT has become an essential ancillary test for studying eye diseases, especially those that involve the macula and optic nerve.

The most important known functions of the choroid are nourishment of the external retinal layers, retinal thermoregulation, light absorption and modulation of intraocular pressure.2 The study of the choroid has been challenging since it cannot be directly visualised, and non-invasive complementary tests, such as ultrasonography or MRI, do not have enough resolution. Indocyanine green angiography is clinically useful, but does not provide cross-sectional information.

Currently, most commercially available OCT systems employ broadband light sources operating at 800–870 nm wavelengths allowing high-definition images of the retina.3 Until recently, choroidal evaluation has not been possible using OCT due to signal roll-off with depth and signal scattering by pigmented tissues or media opacities. Spaide et al4 introduced a technique called enhanced depth imaging OCT (EDI-OCT) based on spectral-domain OCT (SD-OCT) that provided better choroidal visualisation and allowed choroidal thickness (CT) measurement. OCT is now able to demonstrate changes in CT associated with eye diseases such as central serous chorioretinopathy,5 age-related macular degeneration,6 or high myopia.7

Although evolution of SD-OCT devices has improved the anatomic evaluation of the choroid, light scattering can make it difficult to study, especially in individuals with abnormally thickened choroid. In recent years, OCT prototypes with longer wavelength light sources have been developed,8 ,9 showing higher penetration and improving choroidal visualisation. Swept-source OCT (SS-OCT) prototypes at 1 μm wavelength have been used to study the choroid in normal,10 ,11 and pathologic eyes,12–14 showing more detailed images than obtainable with commercially available systems. Both systems have been compared although a small number of healthy eyes have been analysed.15

The aim of this study is to simultaneously assess the ability of both, SD-OCT and SS-OCT, systems for the study of the choroid in a series of non-pathologic eyes.

Patients and methods

This is a cross-sectional prospective study from a single centre. Only eyes without pathology and spherical equivalent (SE) between ±6.0 diopters (D) were included. Age, sex, best corrected visual acuity (BCVA) and SE were recorded. Eyes with a history of perforating eye injuries, traumatic choroidal rupture, retinal detachment, uveitis, corneal leucoma or intraocular laser treatment were excluded. The same ophthalmologist performed a complete ophthalmic examination of every patient. The study was conducted in accordance with the tenets of the Declaration of Helsinki.

Two different OCT devices (SD-OCT and SS-OCT) were used to measure the CT within a few minutes of each other. High-resolution SD imaging was performed using a 3D-2000 OCT (Topcon, Tokyo, Japan). This device uses a super luminescence diode at 840 nm wavelength as light source, and provides 5–6 μm of axial resolution, 20 μm of transverse resolution and maximum scan velocity of 27 000 A-scans per second. For high-penetration measurement, we used the SS-OCT Topcon prototype (Topcon), which uses a tunable laser as a light source to provide a 1050 nm centred wavelength. This prototype reaches a scanning speed of 100 000 A-scans per second, yielding 8 and 20 μm axial and transverse resolution in tissue, respectively.

The scanning protocol ‘line’ centred in the fovea was used in both systems. In 1 s, this protocol averages 50 B-scans and generates a 6 mm line on a SD-OCT device, while a 12 mm line with 96 B-scans averaged is obtained in the SS-OCT prototype. For the Topcon 3D-2000 device, the reference position was set at ‘choroid’ moving the zero-delay line behind the retinal pigment epithelium (RPE), improving the visualisation of the choroid, similar to the EDI mode in Heidelberg Spectralis and Zeiss Cirrus (Carl Zeiss Meditec, Dublin, California, USA) OCT systems. Pupil dilation was used in all eyes.

All scans were reviewed prior to the inclusion into the study. Those with image artefacts, as misalignment bias, or presence of choroid limits alterations, as scars, that did not allow their correct measurements were excluded.

CT measurements were performed manually by two independent observers (SCP and IF-M). CT was determined as the perpendicular distance from the external portion of the RPE to the inner edge of the sclera below the thinnest point of the fovea (subfoveal CT, SFCT) and 3 mm temporal and nasal to the fovea at 500 μm intervals, obtaining 13 determinations (figure 1). The mean CT (MCT) was calculated as their average values. Those eyes that showed good quality retina images, but in which one or both investigators were not able to measure the CT at the subfoveal location, two consecutives or three non-consecutive locations were included but classified as non-measurable.

Figure 1

6 mm line imaged with spectral-domain optic coherence tomography (SD-OCT) (A) and 12 mm swept-source OCT (SS-OCT) (B).

Statistical analysis was performed using the IBM Statistical Package for the Social Sciences V.20 (SPSS, Chicago, Illinois, USA) for Macintosh. Values were presented as mean value±SD. A p value<0.05 was considered statistically significant. Student's t test was used to compare age, refractive error and BCVA between groups and interdevice results. However, if one group was smaller than 30 individuals or its distribution was not normal, we used a Mann–Whitney U test. The OR showed how strongly two properties were associated, so we used it to know the association between eyes non-measured by SD-OCT depending on SD-OCT SFCT. Fisher's exact test was used to determine association between the systems used and whether CT was measured. Pearson's intraclass correlation test and Bland–Altman plots were used to compare CT between observers and between systems. The Bland–Altman method plots show the difference between two variables as a function of their average, showing correlation information between two variables in a graphic way, and providing the amount of disagreement.

Results

Eighty-two healthy eyes from 46 patients (42 women; mean age 47±24 years; range, 5–86) were evaluated. No statistically significant difference was observed by gender for age, SE or BCVA. CT could be measured in 61 eyes (74.4%) by SD-OCT and in the 82 eyes by SS-OCT (p<0.05; Fisher's exact test). In the eyes classified as ‘non-measurable’ by SD-OCT (n=21) 18 eyes were unable to be measured by both or either observers, while three eyes were measured by one of the observers.

The reason the SD-OCT system was not able to measure CT in all cases, was failure to identify the choroido-scleral border clearly. In the group of eyes that could be measured by both systems, the mean age was 45±22 years (range, 5–86), while the mean SE was −0.9±2.1 D (range, −6.0–2.8) and mean LogMAR BCVA was 0.04±0.12 (range, 0.00–1.00) (table 1). In the group of eyes that could not be measured by SD-OCT, the mean age was 48±15 years (range, 26–75), the mean SE was 0.7±1.0 D (range, −0.5–2.5) and mean LogMAR BCVA was 0.07±0.25 (range 0.00–1.00) (table 2).

Table 1.

Epidemiology and OCT values of the 61 patients measured by both devices

Table 2

Difference observed between patients measured by both systems and those only measured by SS-OCT

In the group of eyes that both devices were able to measure, mean SFCT by SD-OCT was 285.7±88.8 μm (range, 130.0–527.0) and 279.3±96.9 μm (range, 84.0–506.0) by SS-OCT (p=0.11; Student's t test). The MCT by SD-OCT was 243.8±78.8 μm (range, 103.6–433.2) vs 242.2±81.8 μm (range, 97.6–459) with SS-OCT (p=0.64; Student's t test).

Both devices showed similar choroidal profiles, with maximum thickness in the subfoveal area and progressive thinning towards temporal and especially, nasal aspects (see figure 2). Regression analysis according to age showed that SFCT decreased by 11.8 and 11.2 μm every decade with SD and SS-OCT, respectively, whereas MCT decreased by 14 and 12.7 μm.

Figure 2

Both devices showed a similar choroidal profile with the maximum thickness in the subfoveal location. Difference observed at temporal edges is due to the difference in the values measured.

The 21 patients whose SFCT could only be measured by SS-OCT showed a significantly higher hyperopic refractive error (p<0.05; Mann–Whitney U test) (table 2) and a thicker choroid (410.9±147.8 μm; range from 126.0 to 681.0) and MCT (347.9±123.1; range from 117.1 to 581.1; p<0.05, Mann–Whitney U test) compared to the eyes that we were able to measure with SD-OCT. We have calculated the OR of being unable to measure SFCT by SD-OCT being 1.4 (0.4–5.8), 8.07 (2.2–30.3), 10.3 (3.2–33.1) for SFCT thickness higher than 200, 300 and 400 μm, respectively.

Interobserver and interdevice correlations are shown in table 3. Pearson's test and intraclass correlation coefficient (ICC) for SFCT were 0.98 for SD-OCT and 0.99 for SS-OCT. Between both systems, Pearson's test and ICC for SFCT were 0.95. All these values decreased more peripheral to the fovea. SFCT Bland–Altman plots were obtained for SD-OCT and SS-OCT interobserver values, and between systems (see figure 3), showing 95% of values difference between ±34, ±22 and ±60 μm, respectively. Although the mean interobserver difference at SFCT with both devices was smaller than 7 μm, two eyes showed interobserver difference higher than 50 μm in both systems, reaching 100 μm in other choroidal locations. The mean interdevice difference for CT measurements ranged from −7 to 6 μm, depending on the location, with SD between 25 and 40 μm. Maximal interdevice difference was 122 μm at SFCT and reached 160 μm at other locations, with no clear deviation for any system. Student's t test did not show statistical difference for interdevice CT at SFCT or MCT, though statistically significant difference was observed at two locations (table 3).

Table 3

Correlation interobserver and interdevice

Figure 3

Bland–Altman plots showing correlation between both devices for SFCT, subfoveal choroidal thickness (SFCT) (A), and mean choroidal thickness (MCT) (B).

Discussion

This study represents a direct, simultaneous comparison between SD-OCT and SS-OCT devices that employ different wavelengths for the analysis of the CT in healthy eyes. Current CT determination by OCT is based on SD systems using 800–870 nm light sources. In the retina, OCT has become an essential ancillary test, but data obtained with different commercial systems cannot be easily compared due to the diverse acquisition protocols used.16 In recent years, SD-OCT devices have been used to evaluate the choroid and its thickness, showing significant interobserver and interdevice reproducibility.17–21 However, sensitivity can decrease in depth, due to light scattering or absorption by pigmented tissues, and important information can be missed, rendering the CT measurement impossible. Manjunath et al22 measured CT in 74% of the eyes from healthy patients with Cirrus HD OCT, and Wei et al23 in 93% of 3468 subjects with Heidelberg Spectralis OCT, being the largest study ever reported. Other studies showed varied proportion of choroid delimitation, with no clear differences between different commercial machines.19 ,20 ,22 Topcon SD-OCT, like the one used in the present study, has been previously used for the determination of CT with a reported rate between 65% and 90%.20 ,24 In our study, reliable measurement of CT was possible in 74% using the SD-OCT device, and in 100% using SS-OCT.

In an attempt to improve choroidal visualisation, prototypes that use tunable lasers instead of super luminescence diode, for operating at 1 μm wavelength region, such as the one used in our study, have been developed. This wavelength centred at 1050 nm enables better imaging depth range with less sensitivity roll-off, allowing better visualisation of the choroid. SFCT in healthy eyes has been shown to be variable in different studies, Agawa et al10 and Ikuno et al11 obtained a thicker choroid measurement (mean SFCT was 355 and 354, respectively) with SS-OCT, than those reported by Margolis25 and Wei,23 with SD-OCT (287 and 254 μm, respectively). In our series, CT measured by SS-OCT was thicker in the 82 eyes initially included than those 61 measured by both systems (313 vs 279 μm for SFCT). On the other hand, the mean SFCT and MCT of those 61 eyes measured manually with both devices did not show statistically significant differences.

Measured by SS-OCT, those 21 patients who were not measured by SD-OCT showed a thicker choroid than the other 61 patients measured by both (table 2). CT appears to be the most important factor in determining if the choroido-scleral interface can be identified by SD-OCT. The OR for CT showed that a thicker choroid, as determined by SS-OCT, was related to a higher chance of non-measurability by SD-OCT. The distance from the external limit of the choroid to zero delay line in a relatively thicker choroid, or the increased light scattering induced by a thicker choroid, could explain these results.

Topcon SD-OCT uses a 6 mm line with 50 averaged images that only allowed measurements along the 6 mm when the fovea was perfectly centred in the image, losing temporal or nasal information when the fovea was displaced. On the other hand, the SS-OCT prototype used in this study provides a 12 mm line with 96 images averaged for studying retina and choroid, allowing us to measure CT up to 3 mm from the fovea in nasal and temporal direction in almost 100% of our eyes. Despite this difference, CT profile observed with both systems was similar, with the thickest choroid at subfoveal location, decreasing more peripheral to the fovea.

Excellent correlation was observed in the 13 locations measured and their average, showing interobserver and interdevice values higher than 0.9 in almost all locations. Correlation was higher in the central areas due to a small displacement of the measurement locations close to the limits. ICC interobserver was 0.99 for MCT and subfoveal measurements for SS-OCT and 0.98 for SD-OCT, decreasing slightly close to the edges, but conserving high ICC values. Interdevice ICC showed an excellent correlation for all measurements with values between 0.95 at SFCT and around 0.9 in the edges. These results agree with previous comparisons of different SD-OCT devices.17–21 Bland–Altman SFCT plots showed worse correlation interdevice than interobserber, and better for SS-OCT than SD-OCT (table 3). This difference could be caused by differences in source or visualisation, or displacement of the line at the time of acquisition.

Additionally, important maximal differences were observed in both groups with 50–70 μm interobservers and 122 μm interdevice at SFCT that reached 160 μm in other locations. Setting the choroidal limit is not always easy, and large vessels can be confused with choroid-sclera interface. So, it is necessary to exercise caution when different devices are used or when the measurements have been performed by different investigators. Although automatic software for CT measurement is available with SS-OCT, manual measurement was used in both devices in order to reduce the procedural bias, and because automatic segmentation could induce artefacts.

Our study has some limitations. No other commercially SD-OCT devices were available in the centre for comparison. ‘Line protocol’ was used to increase the quality of the images obtained, missing information when the image was not perfectly centred. Automatic determination could decrease bias induced by manual measurement, but it could add segmentation artefacts. Although SD-OCT averaged 50 scans (instead of 96, as with SS-OCT) both systems took 1 s, so movement artefacts are similar.

In summary, although excellent correlation was observed between SS-OCT and SD-OCT for CT measurement in healthy eyes, extrapolation of the data between different devices should be exercised with caution. The best quality of choroidal images was obtained with SS-OCT, which allows higher rates of measurement, mainly in those eyes with a thicker choroid.

References

Footnotes

  • Contributors All coauthors have contributed to conception and design, or analysis and interpretation of data; drafting the article and/or revising it critically for important intellectual content. All coauthors finally approved the version to be published. SC and JMR-M are responsible for the overall content as guarantors.

  • Funding This study has been supported in part by a grant of the Spanish Ministry of Health, Instituto de Salud Carlos III, Red Temática de Investigación Cooperativa en Salud ‘Patología ocular del envejecimiento, calidad visual y calidad de vida’ (RD07/0062/0019) and by a Research to Prevent Blindness. Unrestricted Grant to the New England Eye Center/Department of Ophthalmology, Tufts University School of Medicine and the Massachusetts Lions Clubs.

  • Competing interests JSD receives research support from Carl Zeiss Meditech, Inc, and Optovue, Inc.

  • Patient consent Obtained.

  • Ethics approval Institutional Review Board of Vissum Corporación Alicante.

  • Provenance and peer review Not commissioned; externally peer reviewed.