|Year : 2022 | Volume
| Issue : 1 | Page : 49-55
Analysis of choroidal structure and vascularity indices with image binarization of swept source optical coherence tomography images: A prospective study of 460 eyes
Pukhraj Rishi1, Zeeshan Akhtar2, Rupesh Agrawal3, Ashutosh Agrawal4, Ekta Rishi5
1 Shri Bhagwan Mahavir Vitreoretinal Services, Chennai, Tamil Nadu, India
2 Elite School of Optometry, Medical Research Foundation, Sankara Nethralaya, Chennai, Tamil Nadu, India
3 Department of Visual Psychophysics, Srimathi Sundari Subramanian, Chennai, Tamil Nadu, India
4 National Healthcare Group Eye Institute, Tan Tock Seng Hospital, Singapore
5 School of Material Science and Engineering, Nanyang Technological, University Singapore, Singapore
|Date of Submission||03-Jul-2021|
|Date of Decision||31-Oct-2021|
|Date of Acceptance||20-Nov-2021|
|Date of Web Publication||02-Mar-2022|
Dr. Pukhraj Rishi
Shri Bhagwan Mahavir Vitreoretinal Services, Sankara Nethralya, 18 College Road, Chennai - 600 006, Tamil Nadu
Source of Support: None, Conflict of Interest: None
| Abstract|| |
PURPOSE: To evaluate choroidal vascularity index (CVI) among normal subjects using image binarization of swept source optical coherence tomography (SS-OCT).
MATERIALS AND METHODS: Four hundred and sixty eyes of 230 normal participants were included. Total circumscribed choroidal area, luminal area, stromal area (SA), and CVI were derived from SS-OCT scans using open-source software (ImageJ) with the modified Niblack method. Both CVI and subfoveal choroidal thickness (SFCT) were correlated with age, refractive error, intraocular pressure, and mean ocular perfusion pressure (MOPP) using mixed linear model analysis. Pearson's correlation coefficient was used to determine the relationship between age and each dependent factor. Analyses were performed using the SPSS software version 20.0 (IBM Corp., Armonk USA) and statistical significance was tested at 5%.
RESULTS: The mean age was 42.1 (±17.6) years. Mean SFCT was 307 ± 79 μm. Mean CVI was 66.80 (±3.8)%. There was statistically significant positive correlation between CVI and increasing age (r = 0.259, P < 0.0001) and statistically significant negative correlation between SFCT and age (r = −0.361, P < 0.0001). There was positive linear correlation between refractive error and CVI (r = 0.220, P < 0.0001) and negative correlation between SFCT and refractive error. There was no significant effect of MOPP on both CVI (P = 0.07) and SFCT (P = 0.7).
CONCLUSION: CVI and SFCT are significantly correlated with age and refractive error in normal Indian eyes.
Keywords: Choroid, choroidal thickness, choroidal vascularity index, eye, imaging, swept source optical coherence tomography
|How to cite this article:|
Rishi P, Akhtar Z, Agrawal R, Agrawal A, Rishi E. Analysis of choroidal structure and vascularity indices with image binarization of swept source optical coherence tomography images: A prospective study of 460 eyes. Oman J Ophthalmol 2022;15:49-55
|How to cite this URL:|
Rishi P, Akhtar Z, Agrawal R, Agrawal A, Rishi E. Analysis of choroidal structure and vascularity indices with image binarization of swept source optical coherence tomography images: A prospective study of 460 eyes. Oman J Ophthalmol [serial online] 2022 [cited 2022 May 19];15:49-55. Available from: https://www.ojoonline.org/text.asp?2022/15/1/49/338875
| Introduction|| |
The choroid forms the vascular middle coat of the eyeball. Its functions include providing blood supply to the outer retina and foveal avascular zone, retinal thermoregulation, intra-ocular pressure (IOP) modulation through vasomotor control, modulation of scleral vascularization and also a possible role in ocular emmetropization through changes in choroidal thickness (CT) and by controlling ocular elongation. In recent years, there has been an increasing interest to study subfoveal CT (SFCT) in retinal diseases such as central serous chorioretinopathy, age-related macular degeneration (AMD), polypoidal choroidal vasculopathy, diabetic retinopathy, retinitis pigmentosa, pathological myopia, and in various inflammatory and inherited pathologies. This has been possible largely because of the advent of enhanced depth imaging-optical coherence tomography (EDI-OCT), in which, the zero delay line is placed at the level of retinal pigment epithelium (RPE), thereby providing a better resolution of choroid. With this technique, the SFCT can be measured with good reproducibility. Swept source (SS)-OCT has better delineation of sclero-choroidal junction due to deeper penetration, leading to better characterization of choroidal details.
Apart from retino-choroidal conditions, the CT can be affected by physiologic variables such as age, gender, axial length (AXL), refractive error, and diurnal variation. This has led to the search for a relatively more reliable biomarker of choroidal status. The choroidal vascularity index (CVI) has been suggested as one such novel biomarker using image processing of the EDI-OCT scans. With the help of customized image binarization algorithms, choroid was segmented into vascular or luminal areas (LA) and stromal areas (SA) to compute CVI. CVI was defined as the ratio of LA to the total circumscribed choroidal area (TCA). CVI has lesser variability and was influenced by fewer physiologic factors as opposed to SFCT in a cohort study of Singapore Malay Eyes. Calculation of the ratio of LA to SA in the circumscribed cross-sectional choroidal area has also been proposed as an index of choroidal vascularity, although the difference in the two techniques may purely be academic since TCA is equal to the sum of LA and SA. There are limited reports for CVI based on SS OCT scans and also there is no normative database for CVI from Indian eyes. In the present study, we aim to evaluate the CVI and factors affecting CVI and SFCT, amongst normal healthy subjects from India using SS-OCT scans.
| Materials and Methods|| |
This was a prospective, observational, cross-sectional study. Patients were enrolled between July 2015 and September 2016. Institutional Review Board (IRB) approval was obtained for the study, and the tenets of the Declaration of Helsinki were adhered to. IRB approval for contact procedures such as Applanation tonometry, AXL measurement, and central corneal thickness (CCT) was witheld for patients under 18 years of age. Written informed consent was obtained from all the participants. Four hundred and sixty eyes of 230 normal healthy subjects, with no history of ocular or systemic disorder, were enrolled in the study. Subjects with more than six diopters of refractive correction, poor image quality of OCT scan, abnormal OCT scan, and history of ocular surgery in the past 3 months, were excluded.
All participants underwent a detailed ocular examination including best-corrected visual acuity using Snellen's chart, slit-lamp biomicroscopy, and dilated fundus evaluation by indirect ophthalmoscopy. IOP was measured by applanation tonometry (Haag Streit AG, Switzerland), CCT was measured by the DGH ultrasonic Pachymeter Pachette 2® (DGH Technology Inc., 110 Summit Drive, Suite B, Exton, PA 19341, USA), and AXL measurement was performed using ultrasound biometry (OcuScan® RxP, Alcon Laboratories, 6201 South Freeway, Fort Worth, TX, USA). Blood pressure (BP) was recorded in the right arm after 5 min of rest, in sitting position, using a mercury sphygmomanometer (Diamond® Industries, Pune, India). Two readings were recorded and the mean value was used for analysis. Ocular perfusion pressure (OPP) was calculated using the following formula: mean OPP (MOPP) = 2/3 × mean arterial pressure (MAP) – IOP, where MAP = diastolic BP + 1/3 (systolic BP – diastolic BP). Systolic and diastolic OPP were calculated using the following equation: Systolic OPP = systolic BP - IOP. IOP used in the formula was adjusted for CCT using Ehler's formula.
Choroidal images were acquired using SS OCT (SS-OCT, Deep Range Imaging, OCT-1, Atlantis, Topcon, Tokyo, Japan); 12-mm horizontal, vertical and radial scans centered on fovea were obtained. The SFCT was manually measured (using calipers on SS OCT) as the distance in microns between the Bruch's membrane (lower boundary of RPE) and the choroid-scleral interface (CSI). CVI was derived using the technique described by Agrawal et al.'s modification of Sonada's technique; using an open source software ImageJ [Figure 1]., ImageJ is a Java-based image processing program developed at the National Institute of Health.
|Figure 1: Composite image shows the binarization of line scans for the left eye swept source image. Region of interest is identified (a) followed by conversion of the image to grey scale image (b). The image is next binarised using Niblack alogrithm (c). The binarized segmented image is superimposed on the original line scan (d) illustrating the segmentation of the scan to luminal and stromal areas|
Click here to view
The OCT image was semi-automatically segmented using the plug in, “Niblack binarisation algorithm” into SA and LA. Apparently, LA appear dark on binarization, whereas SA appears white. Measurements were made across the entire B-scan. The upper border of this region of interest (ROI) was made at the level of RPE, and the lower border at CSI [Figure 1]. The total area of the ROI was the TCA, which comprised of LA and SA. Ratio of LA to TCA gave the CVI. All measurements were taken manually by a single co-author (ZA) between 11:00 am to 3:00 pm to avoid the impact of confounding variable of diurnal variation on SFCT and CVI. Measurements of both eyes of each participant were obtained.
Descriptive statistics such as mean, standard deviation, and median were obtained for systemic and ocular factors defined on continuous scale, whereas frequencies and percentage were obtained for the categorical factor, i.e., gender. Linear mixed model was used to obtain univariate relationship of each systemic and ocular factor with the dependent, i.e., CVI, as well as SFCT, treating patient as a subject and observations on right and left eyes as repeated on each patient. Unstructured repeated covariance type was used throughout the analysis. The factors showing significant univariate relationship were included in the multivariate model to obtain adjusted relationship of each factor with the dependent. Pearson's correlation coefficient was used to determine the relationship between age and each dependent factor. Furthermore, the patients were binned according to age and the visualization of CVI and SFCT trends was obtained in terms of boxplots across bins. Similar analysis was obtained for refractive error and the dependent factors. All the analyses were performed using the SPSS software version 20.0 (IBM Corp., Armonk USA), and the statistical significance was tested at 5%.
| Results|| |
Four-hundred and sixty eyes of 230 patients were enrolled and analyzed in the study. The demographic and clinical details at the presentation are mentioned in [Table 1]. Of 230 patients, 105 (46%) were female participants, and mean age of the study cohort was 42.13 ± 17.59 years. IOP, CCT, and AXL measurements were carried out only for patients >18 years of age (n = 200, mean IO P = 13.2 mmHg and MOPP = 47.60 mmHg in both eyes) for reasons mentioned in the methods section. The distribution of choroidal parameters according to various age groups in normal healthy participants is provided in [Table 1]. Mean SFCT for right eye and left eye was 307.6 ± 79 um (105–529.3 um) and 307.2 ± 79.7 um (103.8–527.3 um), respectively [Table 1]. The mean CVI in the right eye was 67.00 (±3.83)% while in the left eye was 66.60 (±3.78)%. Frequency distribution for CVI and SFCT for the study sample is depicted in [Figure 2].
|Table 1: Description of demographic, ocular, and systemic parameters of patients (n=230)|
Click here to view
|Figure 2: Frequency distribution for choroidal vascularity index (a) and sub-foveal choroidal thickness (b) for the study sample|
Click here to view
The distribution of CVI and SFCT across various age groups is illustrated on the scatter plots [Figure 3]. It can be seen from the scatter plot [Figure 3] that the highest CVI is in the 50 to 80 years age group. There was statistically significant linear positive correlation between CVI and increasing age (r = 0.259, P < 0.0001) and statistically significant linear negative correlation between SFCT and age (r = -0.361, P < 0.0001). The distribution of CVI and SFCT across refractive errors is shown on the scatter plots [Figure 4]. As seen on the plots, there was positive linear correlation between refractive error and CVI (r = 0.220, P < 0.0001), and negative correlation was obtained between SFCT and refractive error.
|Figure 3: Scatter plots showing relationship between age and (a) choroidal vascularity index and (b) sub-foveal choroidal thickness|
Click here to view
|Figure 4: Scatter plots showing relationship between refractive error and (a) choroidal vascularity index and (b) choroidal sub-foveal thickness|
Click here to view
The relationship of different systemic and ocular parameters including age, refractive errors, and AXL was further evaluated with SFCT, and CVI using linear mixed model analysis and results are shown in [Table 2] and [Table 3], respectively. Both age (P = 0.004) and refractive error (P < 0.0001) were significantly correlated with CVI on multivariate regression analysis. However, IOP (P = 0.237), AXL (P = 0.090), and MOPP (P = 0.078) did not have any effect on CVI values. Similarly, age and refractive error were significantly correlated with SFCT (P < 0.0001) on multivariate regression analysis. OPP was significantly correlated with SFCT on univariate regression analysis but no statistical significance was obtained on multivariate regression analysis.
|Table 2: Relationship of different systemic and ocular parameters with choroidal vascularity index using univariate and multivariate analysis|
Click here to view
|Table 3: Relationship of different systemic and ocular parameters with sub-foveal choroidal thickness using univariate and multivariate analysis|
Click here to view
| Discussion|| |
Various techniques have been described to study the morphology and structure of the choroid in the literature. These include histology of the choroid, indocyanine green angiography (ICGA), laser doppler flowmetry, B-scan, C-scan, Doppler OCT, OCT angiography, EDI-OCT, SS-OCT, and magnetic resonance imaging (MRI). Histology of the choroid is an invasive technique and is also limited by shrinkage of tissues during fixation, which precludes quantitative measurements of the choroid. ICGA and Doppler flowmetry provides information about choroidal blood flow but are unable to provide an anatomical cross-section of the choroid and are also invasive imaging modalities. B-scan has an axial resolution of 150–200 um and is unable to provide fine details of the choroidal structure. A C-scan can provide volumetric data of the choroid by processing images obtained on 3D SD-OCT, but involves a lengthy and time-consuming process., A Doppler OCT measured the frequency shift to visualize choroidal vessels but is limited to measuring blood flow oriented transversely to the image direction while OCT angiography is not dependent on flow rate and orientation for visualization of vessels. An MRI is expensive and lacks the spatial resolution to provide choroidal details. EDI-OCT and SS-OCT provide good anatomical detail of the choroidal angio-architecture, although in eyes with a thicker choroid, the visualization of the CSI may get impaired more in EDI-OCT.
In this study, we measured CVI in 460 healthy eyes of 230 participants using SS-OCT images. There have been few studies, which have measured choroidal parameters in healthy participants, but there are no studies acquiring the images at the same diurnal time point to alleviate the possible effect of diurnal variation on this dynamic parameter. There are also limited studies of CVI on SS-OCT and no normative database of CVI from Indian eyes. In our current study, the mean age of the cohort was 42.1 ± 17.6 (range: 12–80) years. Branchini et al. studied 42 eyes of 42 healthy participants with a mean age of 51.6 ± 21.02 years (range: 23–89 years). Sonoda et al. studied 180 eyes of 180 healthy volunteers with 106 of them being female participants. The mean age of the patients in their study was 55.9 ± 18.8 (range: 22–90) years. Agrawal et al. studied 345 eyes from 345 participants, with 190 (55%) subjects being female. The mean age of patients in this study was 61.53 ± 8.77 (47.2–86.7) years. The relatively lower mean age in our study can be accounted for by the wider age range and inclusion of participants below 18 years of age. This study presents the data for both eyes using SS OCT and using linear mixed model regression analysis, analyzing the association between the potential confounding variables and CVI and SFCT measurements. [Table 4] compares various studies on choroidal vascularity parameters with the current study.
|Table 4: Comparison of ocular and choroidal parameters with other studies|
Click here to view
Sonoda et al. studied the choroidal structure in normal eyes and in eyes with wet AMD after photodynamic therapy using SD-OCT. The mean spherical equivalent was −3.4 D as compared to −0.2 D in our study. Furthermore, choroidal parameters including TCA, LA, and SA were slightly lower as compared to our study. It is known that myopic patients have a thinner choroid as compared to emmetropic patients due to lesser stromal components in the choroid. Although no reduction has been reported in the vascular components of the sub-foveal choroid in myopic eyes, this might explain the slightly reduced TCA (and subsequently LA and SA) in the study by Sonoda et al. but the almost comparable CVI as compared to the present study.
Another difference in both studies conducted by Sonoda et al. as compared to the present study was in the technique of binarization of the images. The present study did not pre-select vessels more than 100 um in size. The binarization technique in our study correlates more closely with the study done by Agrawal et al. Agrawal et al. studied 345 healthy eyes from 345 participants of Singapore Malay descent. The CVI of eyes of Malay descent in their study was comparable to the Indian population in our study, even though the mean SFCT was much lower at 241.34 μm ± 97.11 μm (range, 40.24–519.48 μm) in the former. This indicates to CVI being a closer representation of choroidal vascularity than SFCT.
The ratio of luminal to SA in our study was 2 ± 0.4 in both eyes in our study. Branchini et al. used a custom-software to determine the mean light-dark ratio of the subfoveal choroid to be 0.27 ± 0.08. They interpreted that the subfoveal choroid of a healthy eye has a higher proportion of choroidal vessel lumen (dark pixels) than choroidal stroma (bright pixels). The luminal to stromal ratio was similar in our study was similar to other studies, reflecting it as an alternative method to assess choroidal vascularity.,
In the present study, CVI was found to vary significantly across various age groups (P < 0.01). Age was found to be significantly correlated with CVI in the univariate regression analysis but did not correlate significantly in multivariate analysis in both the right and the left eyes [Table 2] and [Table 4]. This is similar to the results published by Agrawal et al. Multivariate regression analysis in our study showed CVI to be significantly related with LA directly and significant inverse correlation with SA. Sonoda et al. reported that both age and AXL were significantly and negatively correlated with TCA, LA, and SA by multivariate analysis, while Agrawal et al. reported SFCT to be significantly and positively correlated with CVI.
Strengths of our study included a large sample size and use of publicly available software for validation by other researchers. Furthermore, diurnal variation was taken into account by performing the scans at a specific time. There were a few limitations of the study. A manual segmentation method was used to make the choroidal measurements. Furthermore, a single grader was used to process the images so intra-observer bias could not be eliminated. In conclusion, CVI of normal, healthy Indian participants seem comparable to that of other studies despite differences in SFCT. Furthermore, our study further validates the method of image binarization and segmentation to quantitatively measure choroidal parameters. It would be prudent to further study the choroidal vascularity in various chorioretinal diseases as compared to normal eyes and also as a measure of treatment outcomes.
Presentation at a meeting
This study was presented at ARVO Imaging meeting, Hawaii, May 2018.
This study analyses choroidal structural indices in 460 normal eyes using image binarization of swept-source OCT scans and validates this method for quantitatively measuring choroidal parameters. It also shows choroidal vascularity index increases with age.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Nickla DL, Wallman J. The multifunctional choroid. Prog Retin Eye Res 2010;29:144-68.
Chhablani J, Barteselli G. Clinical applications of choroidal imaging technologies. Indian J Ophthalmol 2015;63:384-90.
] [Full text]
Koay CL, Quo MJ, Subrayan V. Reproducibility of choroidal thickness measurements in subjects on 3 spectral domain optical coherence tomography machines. Int Ophthalmol 2017;37:655-71.
Ouyang Y, Heussen FM, Mokwa N, Walsh AC, Durbin MK, Keane PA, et al.
Spatial distribution of posterior pole choroidal thickness by spectral domain optical coherence tomography. Invest Ophthalmol Vis Sci 2011;52:7019-26.
Li XQ, Larsen M, Munch IC. Subfoveal choroidal thickness in relation to sex and axial length in 93 Danish university students. Invest Ophthalmol Vis Sci 2011;52:8438-41.
Nickla DL, Jordan K, Yang J, Totonelly K. Brief hyperopic defocus or form deprivation have varying effects on eye growth and ocular rhythms depending on the time-of-day of exposure. Exp Eye Res 2017;161:132-42.
Tan CS, Ouyang Y, Ruiz H, Sadda SR. Diurnal variation of choroidal thickness in normal, healthy subjects measured by spectral domain optical coherence tomography. Invest Ophthalmol Vis Sci 2012;53:261-6.
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.
Sonoda S, Sakamoto T, Yamashita T, Uchino E, Kawano H, Yoshihara N, et al.
Luminal and stromal areas of choroid determined by binarization method of optical coherence tomographic images. Am J Ophthalmol 2015;159:1123-31.e1.
Hayreh SS. Blood flow in the optic nerve head and factors that may influence it. Prog Retin Eye Res 2001;20:595-624.
Ehlers N, Bramsen T, Sperling S. Applanation tonometry and central corneal thickness. Acta Ophthalmol (Copenh) 1975;53:34-43.
Schneider CA, Rasband WS, Eliceiri KW. NIH image to ImageJ: 25 years of image analysis. Nat Methods 2012;9:671-5.
Huynh E, Chandrasekera E, Bukowska D, McLenachan S, Mackey DA, Chen FK. Past, present, and future concepts of the choroidal scleral interface morphology on optical coherence tomography. Asia Pac J Ophthalmol (Phila) 2017;6:94-103.
Mrejen S, Spaide RF. Optical coherence tomography: Imaging of the choroid and beyond. Surv Ophthalmol 2013;58:387-429.
Zhang L, Lee K, Niemeijer M, Mullins RF, Sonka M, Abràmoff MD. Automated segmentation of the choroid from clinical SD-OCT. Invest Ophthalmol Vis Sci 2012;53:7510-9.
Branchini LA, Adhi M, Regatieri CV, Nandakumar N, Liu JJ, Laver N, et al.
Analysis of choroidal morphologic features and vasculature in healthy eyes using spectral-domain optical coherence tomography. Ophthalmology 2013;120:1901-8.
Tan CS, Ngo WK, Cheong KX. Comparison of choroidal thicknesses using swept source and spectral domain optical coherence tomography in diseased and normal eyes. Br J Ophthalmol 2015;99:354-8.
Sonoda S, Sakamoto T, Yamashita T, Shirasawa M, Uchino E, Terasaki H, et al.
Choroidal structure in normal eyes and after photodynamic therapy determined by binarization of optical coherence tomographic images. Invest Ophthalmol Vis Sci 2014;55:3893-9.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3], [Table 4]