Ophthalmic Surgery, Lasers and Imaging Retina

Imaging Review 

Choroidal Vascularity Index Using Swept-Source and Spectral-Domain Optical Coherence Tomography: A Comparative Study

Rupesh Agrawal, MD; Sophia Seen, MD; Shiva Vaishnavi, MD; Kiran Kumar Vupparaboina, MD; Abhilash Goud, MD; Mohammed Abdul Rasheed, MD; Jay Chhablani, MD

Abstract

BACKGROUND AND OBJECTIVE:

Choroidal vascularity index (CVI) is a noninvasive tool to assess choroidal structure. The objective of the current study was to compare the CVI measurements on swept-source optical coherence tomography (SS-OCT) and spectral-domain OCT (SD-OCT) scans using the same image binarization protocol.

PATIENTS AND METHODS:

This prospective study was conducted from July 2016 through January 2017 and involved 54 healthy volunteers at a tertiary referral eye care institute in Southern India. Choroidal scans were obtained using both SS- and SD-OCT machines. An automated binarization algorithm was used to compute CVI.

RESULTS:

The mean CVI with SS-OCT scans was 53.88% ± 12.54% (range: 20.46% to 73.93%), whereas the mean CVI with SD-OCT scans was 51.11% ± 7.97% (range: 29.90% to 67.72%)(P < .001). The unadjusted (95% confidence interval [CI], 0.554–0.851) and adjusted (95% CI, 0.607–0.871) intraclass correlation (ICC) estimates were quite similar and indicate moderate-to-good reliability of measurements by two machines. The interval estimate for a conversion factor between SD-OCT and SS-OCT can be calculated as follows: SD = [0.383*SS+19.467, 0.586*SS+30.661].

CONCLUSION:

CVI is a noninvasive, robust, and reliable measurement of choroidal vascularity and CVI measurements obtained using both SS-OCT and SD-OCT concur with each other.

[Ophthalmic Surg Lasers Imaging Retina. 2019;50:e26–e32.]

Abstract

BACKGROUND AND OBJECTIVE:

Choroidal vascularity index (CVI) is a noninvasive tool to assess choroidal structure. The objective of the current study was to compare the CVI measurements on swept-source optical coherence tomography (SS-OCT) and spectral-domain OCT (SD-OCT) scans using the same image binarization protocol.

PATIENTS AND METHODS:

This prospective study was conducted from July 2016 through January 2017 and involved 54 healthy volunteers at a tertiary referral eye care institute in Southern India. Choroidal scans were obtained using both SS- and SD-OCT machines. An automated binarization algorithm was used to compute CVI.

RESULTS:

The mean CVI with SS-OCT scans was 53.88% ± 12.54% (range: 20.46% to 73.93%), whereas the mean CVI with SD-OCT scans was 51.11% ± 7.97% (range: 29.90% to 67.72%)(P < .001). The unadjusted (95% confidence interval [CI], 0.554–0.851) and adjusted (95% CI, 0.607–0.871) intraclass correlation (ICC) estimates were quite similar and indicate moderate-to-good reliability of measurements by two machines. The interval estimate for a conversion factor between SD-OCT and SS-OCT can be calculated as follows: SD = [0.383*SS+19.467, 0.586*SS+30.661].

CONCLUSION:

CVI is a noninvasive, robust, and reliable measurement of choroidal vascularity and CVI measurements obtained using both SS-OCT and SD-OCT concur with each other.

[Ophthalmic Surg Lasers Imaging Retina. 2019;50:e26–e32.]

Introduction

The choroid is the highly vascular structure of the eye that lies between the retina and the sclera and provides nourishment to the external layers of the retina.1 Due to the challenges faced in direct visualization of the choroid with high velocity in choroidal circulation, studying the choroid and choroidal circulation has been fraught with significant impendences, and there is currently no robust technique to assess choroidal vasculature or ischemia. Ultrasonography or magnetic resonance imaging of the eye and orbits has not provided sufficient resolution for useful information to be acquired from the choroidal structures. In addition, although indocyanine green angiography provides en face images of the choroidal vasculature with reasonable resolution, it fails at providing cross-sectional information.2

The development of enhanced depth imaging (EDI) optical coherence tomography (OCT) scans has enabled us to acquire high-resolution images with greater depth of penetration, which has allowed us to better visualize the choroid. During the past decade, fervent research has been conducted to quantify the changes in choroidal thickness (CT), a proxy measure of choroidal vascularity, in normal subjects and in patients with ocular and systemic diseases.3–6 Choroidal vascularity index (CVI) has been derived from image binarization of EDI OCT B-scans and utilizes a relatively novel algorithm to compute the vascularity of the choroid. CVI is an noninvasive tool for measuring choroidal vascularity using image segmentation algorithms.7 CVI provides information about choroidal vascularity as, unlike CT, it isolates the vessels and focuses on changes in the vasculature and stroma of the choroid. On the other hand, CT can be influenced by physiological changes; hence, changes in the CT are not specific to the changes in choroidal vasculature or diseases affecting the choroidal structure.

Swept-source OCT (SS-OCT) and spectral-domain OCT (SD-OCT) use the wavelengths around 1,000 nm to 1,100 nm and 800 nm to 900 nm, respectively. Light scattering was a commonly faced issue when using SD-OCT for individuals with abnormally thickened choroid. This issue is reduced with the use of longer wavelength SS-OCT.2 In addition, the longer wavelength of SS-OCT is safer for the eye; hence, it allows for the use of greater laser power for better detection of weak signals scattered from the choroids. At a given laser power, SS-OCT has a reduced sensitivity roll-off along the imaging depth range in comparison with SD-OCT. Moreover, the faster scanning rates of SS-OCT allows for denser scanning patterns with comparable acquisition times as for SD-OCT.8

Numerous studies have been published from our group as well as from other groups validating CVI in various chorioretinal diseases, including age-related macular degeneration,9 central serous retinopathy,10 Vogt-Koyanagi-Harada disease,11 and diabetic retinopathy,12 using SD-OCT. As discussed before, SS-OCT provides better signal of the choroid; therefore, we postulate that it may provide more-accurate and possibly different values for CVI compared to SD-OCT. In this index study, we seek to establish discrepancies in CVI measurements on SS-OCT and SD-OCT scans using the same protocol and to quantify and qualify the differences between the respective scans.

Patients and Methods

Fifty-four healthy volunteers with refractive error of ± 3 diopters (D) were recruited prospectively for this comparative study. The study period was from July 2016 to January 2017 at a tertiary referral eye care institute in Southern India. Prior approval was obtained from the institutional review board, and informed consent was obtained from each subject prior to the acquisition of the scans. The study was conducted in accordance with the tenets of the Declaration of Helsinki. We recorded the demographic details of all patients, and all patients underwent a comprehensive ophthalmic examination. Exclusion criteria included history of active or previous retinal or chorioretinal disease, media opacity precluding fundus imaging, previous ocular surgery (other than cataract surgery), and any other significant ocular comorbidity that had previously been noted.

Imaging Devices and Protocol

Imaging of the choroid was obtained via SS-OCT scan using DRI-OCT Triton (Topcon, Tokyo, Japan) and Cirrus HD-OCT (Carl Zeiss Meditec, Dublin, CA). Images were first acquired with Triton DRI-OCT and within a 5-minute interval with Cirrus SD-OCT. The Topcon SS-OCT uses a tunable laser as a light source to provide a 1,050-nm centered wavelength. The device reaches a scanning speed of 100,000 A-scans per second, yielding 8 μm and 20 μm axial and transverse resolution in tissue, respectively. The Cirrus light source is centered on 800-nm wavelength and achieves 5 μm axial resolution in tissue.

Image Binarization

We used an automated algorithm to analyze choroidal vascularity, as described before by our group.13,14 Briefly, choroidal stroma and vessel area analysis involved (i) automated binarization of an OCT B-scan and (ii) automated segmentation of the binarized choroid layer. The task of automated binarization involved (a) preprocessing, (b) exponential and non-linear enhancement, and (c) thresholding.

Finally, the binarized choroidal layer was used to quantify the stromal area, vessel area and CVI (vessel: [stromal + vessel] area ratio) (Figure 1). The automated algorithm was applied on total choroidal scans to get the CVI across the entire volume of the choroid.

Optical coherence tomography (OCT) of the left eye of a 27-year-old male using swept-source OCT (A) and spectral-domain OCT (B), respectively. (C, D) Choroid segmentation using automated algorithm. (E, F) Binarized images, which were used to calculate the choroidal vascularity index (0.512 and 0.479, respectively)

Figure 1.

Optical coherence tomography (OCT) of the left eye of a 27-year-old male using swept-source OCT (A) and spectral-domain OCT (B), respectively. (C, D) Choroid segmentation using automated algorithm. (E, F) Binarized images, which were used to calculate the choroidal vascularity index (0.512 and 0.479, respectively)

Statistical Analysis

Statistical analysis was performed using SPSS Version 22.0 (SPSS, Chicago, IL). Interclass correlation (ICC) and Bland-Altman plots (BAPs) were used to establish the concordance or discordance for CVI measured using two machines. Pearson's correlation coefficient was used to identify the association of confounding variables such as age, best-corrected visual acuity (BCVA), and spherical equivalent (SE) with CVI readings of respective machines.

Results

There were 54 eyes of 54 patients analyzed for the current study. The mean age was 37 years ± 15.88 years (range: 21 years to 66 years), the mean SE was −0.73 D ± 1.15 D, and the mean BCVA was 0.029 logMAR units ± 0.0924 logMAR units. All patients were phakic. The mean CVI score for SS-OCT was 53.88% ± 12.54% (range: 20.46% to 73.93%), whereas the mean CVI score for SD-OCT was 51.11% ± 7.97% (29.90% to 67.72%) (P < .001) (Figure 2).

Scatter plot showing the relationship between the choroidal vascularity index measurements by two machines

Figure 2.

Scatter plot showing the relationship between the choroidal vascularity index measurements by two machines

Intraclass Correlation (Unadjusted)

To determine the correlation and the extent of agreement between the CVI values by two machines, ICC analysis was performed. The “two-way mixed” model was selected for ICC analysis. The ICC for the unadjusted measurements was 0.742 with a 95% confidence interval (CI) of 0.554–0.851. The correlation was statistically significant with P value less than .001.

BAPs (Figure 3) were constructed to demonstrated the discrepancies for CVI between the two machines.

Bland-Altman plots demonstrating good correlation between choroidal vascularity index measurements obtained using swept-source optical coherence tomography (OCT) and spectral-domain OCT machines.

Figure 3.

Bland-Altman plots demonstrating good correlation between choroidal vascularity index measurements obtained using swept-source optical coherence tomography (OCT) and spectral-domain OCT machines.

Intraclass Correlation (Adjusted)

The Pearson's correlation coefficient for age, BCVA, and SE was obtained with SS-OCT as well as SD-OCT, with the results shown in Table 1.

Correlation Coefficient for Two Devices

Table 1:

Correlation Coefficient for Two Devices

It is evident from the above tables that age is negatively correlated with SS-OCT CVI measurements with coefficient of −0.331 (P = .015), whereas BCVA and SE showed insignificant correlation with these measurements. SD-OCT CVI measurements showed insignificant correlation with all the three covariates. Moreover, the difference of mean SS-OCT and SD-OCT between gender was statistically insignificant (P = .385 and P = .309, respectively).

In view of the above findings, the adjustment of CVI measurements by SS and SD were performed using one-way MANCOVA. The measurements SS-OCT CVI and SD-OCT CVI were treated as dependents, whereas age and gender (0: Female; 1: Male) were considered as covariates. The correlation between the two dependent variables was 0.67, which was statistically significant (P < .001. The parameter estimates obtained for the two dependent variables are given in Table 2.

Parameter Estimates for the Regression Models

Table 3:

Parameter Estimates for the Regression Models

Age has a significant impact on SS-OCT CVI (Table 3). As age increased, SS-OCT CVI reduced (P = .017). Using the above parameter estimates, the adjusted values for SS-OCT CVI and SD-OCT CVI measurements were obtained. The descriptive statistics for the adjusted measurements, as given in Table 3.

Descriptive Statistics for the Adjusted Measurements of CVI by Two Machines

Table 3:

Descriptive Statistics for the Adjusted Measurements of CVI by Two Machines

The ICC was obtained for the adjusted measurements using the two-way mixed model. Referring to average measures, the ICC obtained was 0.776 (95% CI, 0.607–0.871) and showed statistical significance with a P value of less than .001. A graphical visualization of the adjusted CVI measurements by two machines is given below (Figure 4).

Scatter plot showing the correlation of adjusted choroidal vascularity index measurements by two machines

Figure 4.

Scatter plot showing the correlation of adjusted choroidal vascularity index measurements by two machines

In summary, the unadjusted and adjusted ICC estimates are quite similar and indicate moderate-to-good reliability of measurements for two machines. The effect of covariates on SS-OCT and SD-OCT measurements was marginal; as a result, there was a small increase in the ICC coefficient from 0.742 to 0.776.

The interval estimate for a conversion factor between SD-OCT and SS-OCT can be calculated as follows: standard deviation = [0.383*SS+19.467, 0.586*SS+30.661] (Figure 4).

Discussion

Although there have been several studies comparing CT in SD-OCT and SS-OCT, this is the first study that directly and simultaneously compares the CVI measurements from SD-OCT and SS-OCT. Based on the guidelines from the work of Cicchetti,15 an ICC coefficient of 0.776 can be interpreted as excellent, suggesting that there is good agreement between the CVI values and the measurements obtained using images from either OCT machines are comparable.

Previously published studies on CT in SD-OCT and SS-OCT reported contradictory results. Matsuo et al.16 and Ikuno et al.17 have demonstrated thicker subfoveal choroidal CT using SS-OCT than CT on SD-OCT. Matsuo et al. postulated that the CT measurements were thicker using SS-OCT because the choroid-scleral border seen on SD-OCT scans may not be the true border and hence was inaccurate. Copete et al., on the other hand, found that there was no difference in CT measured by SS-OCT and SD-OCT and that the ICC for SD-OCT and SS-OCT was 0.95.2

We used an automated segmentation algorithm in the current study to quantify the vascular areas. This eliminates possible subjective biases in manual measurements. Another strength of our study is that scans of each eye were performed within 5 minutes of each other. This minimized the possibility of CVI variations as a result of factors that have been found to be associated with CT changes, such as changes in circulating catecholamines, diurnal variations, and fluctuations of intraocular pressure,13,18,19 as these would have remained largely constant in the span of 5 minutes.

In general, the quality of the images obtained using SS-OCT is superior to that of SD-OCT.2,14 Moreover, previous data have suggested that images acquired from SD-OCT can be used to calculate CT most of the time (75%20 to 90%21), whereas images acquired from SS-OCT can be used to calculate CT all of the time (100%2). The specific steps carried out in our algorithm allowed for greater contrast of the different areas of the choroid, including in areas of poorer resolution, thus allowing for clearer demarcations for the calculation of CVI from scans acquired using SS-OCT. Although CVI results are comparable in SS-OCT and SD-OCT, SS-OCT is still preferable to SD-OCT for reasons not relating to CVI calculation. Based on results of other studies, SD-OCT performs poorly at determining the choroido-scleral interface in thicker choroids due to greater light scattering, increased signal loss, poorer resolution from more artifacts, and shadows cast by connective tissue between vessels resulting in poor image quality.2 The confidence interval and range of CVI for both the machines were quite significant, and along with the smaller sample size, the results obtained from this current study may not be totally relevant and significant.

In conclusion, CVI can be a potentially noninvasive, robust, and reliable measurement of choroidal vascularity. CVI measurements obtained from both SS-OCT and SD-OCT concur with each other, although with significant spread in the range.

References

  1. Nickla DL, Wallman J. The multifunctional choroid. Prog Retin Eye Res. 2010;29(2):144–168. doi:10.1016/j.preteyeres.2009.12.002 [CrossRef]
  2. Copete S, Flores-Moreno I, Montero JA, Duker JS, Ruiz-Moreno JM. Direct comparison of spectral-domain and swept-source OCT in the measurement of choroidal thickness in normal eyes. Br J Ophthalmol. 2014;98(3):334–338. doi:10.1136/bjophthalmol-2013-303904 [CrossRef]
  3. Ikuno Y, Tano Y. Retinal and choroidal biometry in highly myopic eyes with spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2009;50(8):3876–3880. doi:10.1167/iovs.08-3325 [CrossRef]
  4. Ahmad M, Kaszubski PA, Cobbs L, Reynolds H, Smith RT. Choroidal thickness in patients with coronary artery disease. PloS One. 2017;12(6):e0175691. doi:10.1371/journal.pone.0175691 [CrossRef]
  5. Zhu D, Wang Y, Zheng YF, et al. Choroidal thickness in school children: The Gobi Desert Children Eye Study. PloS One. 2017;12(6):e0179579. doi:10.1371/journal.pone.0179579 [CrossRef]
  6. Sodi A, Lenzetti C, Murro V, et al. EDI-OCT evaluation of choroidal thickness in retinitis pigmentosa. Eur J Ophthalmol. 2018;28(1):52–57. doi:10.5301/ejo.5000961 [CrossRef]
  7. Agrawal R, Gupta P, Tan KA, Cheung CM, Wong TY, Cheng CY. Choroidal vascularity index as a measure of vascular status of the choroid: Measurements in healthy eyes from a population-based study. Sci Rep. 2016;6:21090. doi:10.1038/srep21090 [CrossRef]
  8. Zhang Q, Chen CL, Chu Z, et al. Automated quantitation of choroidal neovascularization: A comparison study between spectral-domain and swept-source OCT angiograms. Invest Ophthalmol Vis Sci. 2017;58(3):1506–1513. doi:10.1167/iovs.16-20977 [CrossRef]
  9. Koh LH, Agrawal R, Khandelwal N, Sai Charan L, Chhablani J. Choroidal vascular changes in age-related macular degeneration. Acta Ophthalmol. 2017;95(7):e597–e601. doi:10.1111/aos.13399 [CrossRef]
  10. Agrawal R, Chhablani J, Tan KA, Shah S, Sarvaiya C, Banker A. Choroidal vascularity index in central serous chorioretinopathy. Retina. 2016;36(9):1646–1651. doi:10.1097/IAE.0000000000001040 [CrossRef]
  11. Agrawal R, Li LK, Nakhate V, Khandelwal N, Mahendradas P. Choroidal vascularity index in Vogt-Koyanagi-Harada disease: An EDI-OCT derived tool for monitoring disease progression. Transl Vis Sci Technol. 2016;5(4):7. doi:10.1167/tvst.5.4.7 [CrossRef]
  12. Tan KA, Laude A, Yip V, Loo E, Wong EP, Agrawal R. Choroidal vascularity index - a novel optical coherence tomography parameter for disease monitoring in diabetes mellitus?Acta Ophthalmol. 2016;94(7):e612–e16. doi:10.1111/aos.13044 [CrossRef]
  13. Tan KA, Gupta P, Agarwal A, et al. State of science: Choroidal thickness and systemic health. Surv Ophthalmol. 2016;61(5):566–581. doi:10.1016/j.survophthal.2016.02.007 [CrossRef]
  14. Ting DS, Cheung GC, Lim LS, Yeo IY. Comparison of swept source optical coherence tomography and spectral domain optical coherence tomography in polypoidal choroidal vasculopathy. Clin Exp Ophthalmol. 2015;43(9):815–819. doi:10.1111/ceo.12580 [CrossRef]
  15. Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychological Assessment. 1994;6(4):284–290. doi:10.1037/1040-3590.6.4.284 [CrossRef]
  16. Matsuo Y, Sakamoto T, Yamashita T, Tomita M, Shirasawa M, Terasaki H. Comparisons of choroidal thickness of normal eyes obtained by two different spectral-domain OCT instruments and one swept-source OCT instrument. Invest Ophthalmol Vis Sci. 2013;54(12):7630–7636. doi:10.1167/iovs.13-13135 [CrossRef]
  17. Ikuno Y, Maruko I, Yasuno Y, et al. Reproducibility of retinal and choroidal thickness measurements in enhanced depth imaging and high-penetration optical coherence tomography. Invest Ophthalmol Vis Sci. 2011;52(8):5536–5540. doi:10.1167/iovs.10-6811 [CrossRef]
  18. Chakraborty R, Read SA, Collins MJ. Diurnal variations in axial length, choroidal thickness, intraocular pressure, and ocular biometrics. Invest Ophthalmol Vis Sci. 2011;52(8):5121–5129. doi:10.1167/iovs.11-7364 [CrossRef]
  19. Brown JS, Flitcroft DI, Ying GS, et al. In vivo human choroidal thickness measurements: Evidence for diurnal fluctuations. Invest Ophthalmol Vis Sci. 2009;50(1):5–12. doi:10.1167/iovs.08-1779 [CrossRef]
  20. Manjunath V, Taha M, Fujimoto JG, Duker JS. Choroidal thickness in normal eyes measured using Cirrus HD optical coherence tomography. Am J Ophthalmol. 2010;150(3):325–329.e1. doi:10.1016/j.ajo.2010.04.018 [CrossRef]
  21. Wei WB, Xu L, Jonas JB, et al. Subfoveal choroidal thickness: The Beijing Eye Study. Ophthalmology. 2013;120(1):175–180. doi:10.1016/j.ophtha.2012.07.048 [CrossRef]

Correlation Coefficient for Two Devices

AgeBCVASE
SS-OCT CVIPearson Correlation−0.331*−0.055−0.012
P Palue (Two-Tailed).015.692.93
N545454
SD-OCT CVIPearson Correlation−0.080.041−0.079
P Value (Two-Tailed).566.767.57
N545454

Parameter Estimates for the Regression Models

Dependent VariableBSEt-TestP Value95% CI
Lower BoundUpper Bound
SS-OCT CVIIntercept62.3804.47313.946.00053.40071.360
Age−0.2560.104−2.464.017−0.465−0.047
Gender2.4353.3310.731.468−4.2529.122
SD-OCT CVIIntercept53.6742.98717.968.00047.67759.671
Age−0.0440.069−0.640.525−0.1840.095
Gender−2.2442.224−1.009.318−6.7092.222

Descriptive Statistics for the Adjusted Measurements of CVI by Two Machines

NMinimumMaximumMeanStandard Deviation
SS-OCT CVI adjusted5422.4878.5153.88711.773
SD-OCT CVI adjusted5429.1767.4351.1077.862
Authors

From the National Healthcare Group Eye Institute, Tan Tock Seng Hospital, Singapore (RA); Yong Loo Lin School of Medicine, National University of Singapore, Singapore (RA, SS); and L.V. Prasad Eye Institute, Hyderabad, India (SV, KKV, AG, MAR, JC).

The authors report no relevant financial disclosures.

Address correspondence to Jay Chhablani, MD, Smt. Kanuri Santhamma Retina Vitreous Centre, L.V.Prasad Eye Institute, Kallam Anji Reddy Campus, Banjara Hills, Hyderabad, 500 034 India; email: jay.chhablani@gmail.com.

Received: December 04, 2017
Accepted: May 09, 2018

10.3928/23258160-20190129-15

Sign up to receive

Journal E-contents