Translational Vision Science & Technology Cover Image for Volume 14, Issue 6
June 2025
Volume 14, Issue 6
Open Access
Retina  |   June 2025
Three-Dimensional Choroidal Contour Mapping in Healthy and Diseased Eyes
Author Affiliations & Notes
  • Supriya Arora
    Bahamas Vision Centre and Princess Margaret Hospital, Nassau, Bahamas
  • Sumit Randhir Singh
    Akhand Jyoti Eye Hospital, CoE Mastichak, Saran, Bihar, India
  • Sharat Chandra Vupparaboina
    Department of Ophthalmology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA
  • Brian Rosario
    UPMC Eye Center, University of Pittsburgh, Pittsburgh, PA, USA
  • Mohammed Nasar Ibrahim
    Department of Ophthalmology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA
  • Amrish Selvam
    UPMC Eye Center, University of Pittsburgh, Pittsburgh, PA, USA
  • Arman Zarnegar
    UPMC Eye Center, University of Pittsburgh, Pittsburgh, PA, USA
  • Sanjana Harihar
    UPMC Eye Center, University of Pittsburgh, Pittsburgh, PA, USA
  • Vinisha Sant
    UPMC Eye Center, University of Pittsburgh, Pittsburgh, PA, USA
  • Jose Alain Sahel
    UPMC Eye Center, University of Pittsburgh, Pittsburgh, PA, USA
  • Kiran Kumar Vupparaboina
    Department of Ophthalmology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA
  • Jay Chhablani
    UPMC Eye Center, University of Pittsburgh, Pittsburgh, PA, USA
  • Correspondence: Jay Chhablani, UPMC Eye Center, University of Pittsburgh, 203 Lothrop St., Pittsburgh, PA 15213, USA. e-mail: [email protected] 
Translational Vision Science & Technology June 2025, Vol.14, 16. doi:https://doi.org/10.1167/tvst.14.6.16
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Supriya Arora, Sumit Randhir Singh, Sharat Chandra Vupparaboina, Brian Rosario, Mohammed Nasar Ibrahim, Amrish Selvam, Arman Zarnegar, Sanjana Harihar, Vinisha Sant, Jose Alain Sahel, Kiran Kumar Vupparaboina, Jay Chhablani; Three-Dimensional Choroidal Contour Mapping in Healthy and Diseased Eyes. Trans. Vis. Sci. Tech. 2025;14(6):16. https://doi.org/10.1167/tvst.14.6.16.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: Quantitative evaluation of choroidal curvature including choroidal inner boundary (CIB) and choroidal outer boundary (COB) and report a comparison between healthy and diseased eyes.

Methods: This retrospective study was conducted on 97 eyes of 97 patients. Eyes were divided into three groups: central serous chorioretinopathy (CSCR), dry age-related macular degeneration (AMD), and healthy eyes. Delineation of CIB and COB was performed using a hybrid method based on our previously validated deep learning and three-dimensional (3D) smoothing methods for choroidal layer segmentation. Quantitative analysis of the surfaces was based on best-fit spherical radius (R). R for overall surface, as well as for each region (central/nasal/temporal/superior/inferior region), was estimated. Statistical analysis was done using SPSS software.

Results: There were 35 healthy eyes, 32 eyes with CSCR, and 30 eyes with dry AMD. At CIB and COB; RCSCR > Rhealthy > RAMD (P ≤ 0.001). The central region had the lowest R among all the regions within a group at CIB and COB (P < 0.001) in healthy, CSCR, and AMD eyes. There was moderate positive correlation of R of CIB and COB with subfoveal choroidal thickness in healthy eyes and a negligible/weak correlation in CSCR and AMD eyes.

Conclusions: Contour of choroid at CIB and COB was the flattest in CSCR and steepest in AMD. Central region was the steepest among all sectors in both healthy and diseased eyes.

Translational Relevance: Quantitative study of surface at CIB and COB in diseases helps in understanding the pathophysiological changes and provides a clinical biomarker in disease monitoring and treatment as well.

Introduction
In the past, the only way to study choroid in health and disease were ultrasound scanning and indocyanine green angiography. With a recent revolution in technology of optical coherence tomography (OCT), introduction of swept-source OCT (SS-OCT), and enhanced depth imaging OCT, it is now possible to do an in-depth analysis of choroidal structure in vivo.1 The traditional way of assessment of choroid in OCT is evaluation of cross-sectional B scans. Measurement of choroidal thickness (CT) and its applications have been extensively studied1 and recently measurement of choroidal vascularity index (CVI)25 is becoming an important biomarker too. Various other approaches have been attempted including en-face structural OCT,6 choroidal vascularity mapping7 and even three-dimensional (3D) visualization of choroidal vessels is underway. All the imaging studies evaluating choroid are based on choroidal evaluation within the choroidal slab of the OCT scans; however, there are no studies that evaluated the “surface” of the choroid. Considering the impact of biometric and pathological changes, there is a possible change in the surface/contour of the choroid, either in inner or outer surfaces. 
We recently did a study to develop an algorithm for quantitative characterization of curvature of choroidal surfaces including inner and outer surfaces on healthy eyes.8 In this study, we aim to apply that algorithm and report a comparison between healthy and diseased eyes. In diseased eyes, we chose eyes with central serous chorioretinopathy (CSCR) on one end and eyes with dry age-related macular degeneration (AMD) on the other end because of their contrasting choroidal features in comparison to healthy eyes. CSCR is known to have a thick choroid and higher CVI,9 whereas AMD is known to have a thin choroid and lower CVI.10 
Methods
This was a retrospective study carried out in accordance with tenets of the Declaration of Helsinki. Ethical clearance was obtained by the institutional review board of the University of Pittsburgh. Informed consent was obtained from all participants to include their retrospective data between November 2021 to April 2022 in the study. The site of enrollment was UPMC Eye Centre in University of Pittsburgh, and the ethnicity of patients was mainly White. We included eyes in three groups with a known diagnosis of (i) CSCR, (ii) dry AMD, and (iii) Healthy eyes. The exclusion criteria were as follow: (i) based on patient profile: neurological disorder, cervical spine/orthopedic issues, non-cooperation, non-consenting; (ii) based on ocular condition: significant refractive error (≥ +4 D or ≥ −6 D), high myopia (≥ −6 D), axial length < 23 mm or > 25 mm, corneal opacity/degeneration/dystrophy, pterygium, keratoconus, pupil abnormalities, glaucoma, cataract/ subluxated lens, asteroid hyalosis, choroidal tumors, squint, uveitis, optic disc pit, congenital anomalies, trauma; (iii) any other retinal disease: diabetic retinopathy, vascular occlusion, epiretinal membrane, macular hole, inherited retinal degenerations, inflammatory conditions, chorioretinal infections, retinal tumors, choroidal neovascularization, vitreous hemorrhage; (iv) previous surgery/procedure (other than an uneventful cataract surgery), laser, intravitreal injection: refractive surgery, corneal surgery, complicated cataract surgery, posterior capsular opacification, decentered intraocular lens implant, scleral or corneal perforation repair, glaucoma surgery, vitrectomy, laser treatment (cornea/iris/glaucoma/retina), intravitreal injections; (v) poor OCT scans: due to patient movement during scanning, due to media opacity or any other reason for a poor quality OCT scan. The quality of scan was ensured by the in-built scoring system in the SS-OCT machine. A score out of 10 is rewarded by the machine for every scan. Scans with score ≥6 (highlighted as green) were accepted for the analysis. Only treatment-naïve eyes were included in the CSCR group. Eyes with pachychoroid-related diseases (other than CSCR) were excluded. In the AMD group, eyes with wet AMD were excluded (Fig. 1). 
Figure 1.
 
Flowchart depicting categorization of eyes into three groups and inclusion and exclusion criteria.
Figure 1.
 
Flowchart depicting categorization of eyes into three groups and inclusion and exclusion criteria.
Evaluation of participants included history taking, complete ophthalmic examination and dilated imaging on wide-field SS-OCT 12 × 12 mm on the Plex Elite 9000 device (Carl Zeiss Meditec, Dublin, CA, USA) centered on the fovea. OCT scans of the subjects were taken between 9 AM to 12 noon. SS-OCT scans were exported as complete eight-bit volumes. Each OCT volume comprised 1024 B-scans, and the resolution of each scan was 1024 × 1536. 
Image Analysis
Delineation of Choroidal Inner Boundary (CIB) and Choroidal Outer Boundary (COB)
For segmentation of COB and CIB, we adopted a modified choroid segmentation algorithm combining our previously validated deep learning and 3D smoothing methods for choroid layer segmentation.11 We combined two methods to suit our dataset (i.e., SS-OCT volume scans) and to reduce the computation cost (execution time) per volume maintaining the accuracy of the segmentation. In particular, we replaced the initial choroid boundary estimation step used in an article by Ibrahim et al.12 with the ResUNet deep learning method reported by Vupparaboina et al.13 Note that there may be misalignment with some B-scans of a volume during image acquisition, and we see a deviation in the position of structures in those scans when compared with adjacent B-scans. To mitigate these abrupt shifts in boundaries obtained in B-scans, we performed robust locally estimated scatterplot smoothing in the direction orthogonal to B-scans. A smoothing parameter of 0.1 is chosen to keep the local neighborhood window small so that it does not alter the overall profile of the surface.12,1416 In the study, we only included the volumes whose choroid boundary segmentation are verified and approved by the expert grader. In addition, we also performed Dice coefficient (DC)−based accuracy analysis where two scans from each volume were randomly picked from 60 volumes (20 each from the Healthy, CSCR, and AMD categories) and obtained manual segmentation to compare against the automated segmentations. For these 120 B-scans, we calculated the DC between the manual and the automated segmentation. We observed a mean DC value of 95.08%, 93.62%, and 95.19% for the healthy, CSCR, and AMD categories. On representative healthy, CSCR, and AMD OCT B-scans, Figure 2 shows the subjective comparison of segmentation of the proposed approach vis-a-vis corresponding manual segmentation. 
Figure 2.
 
Comparison of choroid segmentations by manual and proposed automated methods. Two scans of a healthy eye, two scans of an eye with CSCR, and two scans of an eye with AMD have been shown to compare manual versus automated segmentation. For each eye sequential OCT B scans were used to generate OCT volume data.
Figure 2.
 
Comparison of choroid segmentations by manual and proposed automated methods. Two scans of a healthy eye, two scans of an eye with CSCR, and two scans of an eye with AMD have been shown to compare manual versus automated segmentation. For each eye sequential OCT B scans were used to generate OCT volume data.
Features for Quantitatively Characterizing Curvature of Choroidal Inner and Outer Boundary (CIB and COB)
To evaluate the surface curvature changes on the CIB and COB obtained from healthy, CSCR and AMD subjects, we performed quantitative analysis of the surfaces based on analysis of best-fit spherical radius (R)17,18 as done in our previously published article.8 The quantitative analysis was performed for the overall CIB (en-face CIB) and COB surfaces (en-face COB). CIB and COB were partitioned into five regions each, namely: central region, nasal region, temporal region, superior region, and inferior region. 
To facilitate the expert grading, analysis, and division into regions, we developed an in-house MATLAB-based graphical user interface (GUI) to view the choroidal surfaces, thickness maps and curvatures maps and to generate respective spreadsheets consisting of region-wise statistics (Fig. 3). In GUI, the user selects the foveal center on the RPE en-face image. Based on the user input, a predefined grid mask consisting of five regions is overlaid on the RPE en-face image with grid's center at the foveal center. 
Figure 3.
 
The division into five regions (central, nasal, temporal, superior, inferior) to facilitate the study of region-wise statistics. In the GUI, the user selects the foveal center on the RPE en face image. The user also selects two more points: any point nasal to the fovea (based on the optic disc) and any point superior to the fovea (based on optic disc and fovea). Based on the user input, a predefined grid mask of size 1500 × 1500 pixels (much larger than the en-face image size of 1024 × 750 pixels) consisting of five regions is overlaid on the RPE en-face image with the grid's center matching with the foveal center. More specifically, the grid mask has a circle of radius 1 mm (around ∼85 pixels) at its center and the remaining region is four equal regions separated by lines positioned at 45°, 135°, 225°, and 315°. The circle of radius 1 mm around the foveal center constitutes the central region. Further, based on the user input, nasal and superior regions are identified. The region opposite to the nasal corresponds to the temporal region, and, similarly, the region opposite to the superior region is the inferior region.
Figure 3.
 
The division into five regions (central, nasal, temporal, superior, inferior) to facilitate the study of region-wise statistics. In the GUI, the user selects the foveal center on the RPE en face image. The user also selects two more points: any point nasal to the fovea (based on the optic disc) and any point superior to the fovea (based on optic disc and fovea). Based on the user input, a predefined grid mask of size 1500 × 1500 pixels (much larger than the en-face image size of 1024 × 750 pixels) consisting of five regions is overlaid on the RPE en-face image with the grid's center matching with the foveal center. More specifically, the grid mask has a circle of radius 1 mm (around ∼85 pixels) at its center and the remaining region is four equal regions separated by lines positioned at 45°, 135°, 225°, and 315°. The circle of radius 1 mm around the foveal center constitutes the central region. Further, based on the user input, nasal and superior regions are identified. The region opposite to the nasal corresponds to the temporal region, and, similarly, the region opposite to the superior region is the inferior region.
Figure 4.
 
Sequential OCT B scans were used to generate OCT volume. Choroidal contour on CIB and COB and choroidal thickness map have been demonstrated in representative cases of AMD eye, healthy eye, and eye with CSCR.
Figure 4.
 
Sequential OCT B scans were used to generate OCT volume. Choroidal contour on CIB and COB and choroidal thickness map have been demonstrated in representative cases of AMD eye, healthy eye, and eye with CSCR.
First, R is estimated for the overall surface to understand the overall curvature of the surface. Then R is computed for each point on the surface using the user-contributed sphereFit code from MATLAB File Exchange developed based on the least-squares regression  
\begin{eqnarray*}{\left( {x - a} \right)^2} + {\left( {y - b} \right)^2} + {\left( {z - c} \right)^2} = {R^2},\end{eqnarray*}
where (a, b, c) indicate the center and R indicates the radius of the sphere.18 In particular, it extends the Kasa modified least squares method for circle fitting.19,20 
Subsequently, we estimated R for each region (i.e., for central, nasal, temporal, superior, and inferior regions). Intuitively, a flatter surface will have a larger radius and vice versa. 
Statistical Analysis
Normality of the data was assessed using the Shapiro-Wilk test at a 5% level of significance (α = 0.05). The Levene test for equality of variances was performed to ensure that the assumption of equal variances holds. One-way analysis of variance, the nonparametric alternative Kruskal-Wallis H test, and the Friedman test were used to compare quantitative parameters among the three groups (healthy, CSCR, and AMD eyes) based on the pattern of data distribution. Correlation analyses were done using Pearson or Spearman correlation based on the normality distribution. Correlation was considered negligible when r = 0.00–0.10; it was considered a weak correlation when r = 0.10–0.39; a moderate correlation when r = 0.40–0.69; a strong correlation when r = 0.70–0.89, and a very strong correlation when r = 0.90–1.00.21 Statistical analysis was done using Statistical Package for the Social Sciences (SPSS software; version 23; IBM, Armonk, NY, USA). 
Results
The study cohort included 97 eyes of 97 patients. There were 35 healthy eyes, 32 eyes with CSCR, and 30 eyes with dry AMD. Baseline demographic features like mean age, sex, visual acuity, and eye laterality have been tabulated in Table 1. For comparison of the proposed segmentation approach versus the manual segmentation method, DC-based accuracy analysis was performed as mentioned in the Methods section. Table 2 compares overall R at CIB, as well as COB in each of the three groups: 
  • 1. There are three variables in the first row are: Variable 1 (V1): mean R at CIB in healthy eyes (R healthy CIB); variable 2 (V2): mean R at CIB in CSCR eyes (R CSCR CIB); variable 3 (V3): mean R at CIB in dry AMD eyes (R AMD CIB). Hypothesis was that there is no difference in distribution of R across the three groups. The Shapiro-Wilk test for assessing normal distribution of data was not satisfied. The Levene test for equality of variance was satisfied. The Kruskal-Wallis test was used, and with a P value < 0.001, the hypothesis was rejected. Dunn's post hoc pairwise comparison showed AMD – CSCR (P < 0.001); AMD – Healthy (P = 0.047); CSCR – Healthy (P = 0.066).
  • 2. V1: R healthy at choroidal outer boundary (COB); V2: R CSCR COB; V3: R AMD COB. Hypothesis: there is no difference in distribution of R across three groups. Normality (Shapiro-Wilk test): not satisfied; Variance (Levene test): satisfied. Kruskal-Wallis test, P = 0.001. Hypothesis rejected. Dunn's post hoc pairwise comparison showed AMD – CSCR (P = 0.001); AMD – Healthy (P = 0.050); CSCR – Healthy (P = 0.562).
Table 1.
 
Demographic Features Including Healthy Eyes, CSCR Eyes, and Dry AMD Eyes
Table 1.
 
Demographic Features Including Healthy Eyes, CSCR Eyes, and Dry AMD Eyes
Table 2.
 
Overall Features (En-Face Region) at CIB and COB in the Three Groups: Healthy, CSCR, and Dry AMD Eyes
Table 2.
 
Overall Features (En-Face Region) at CIB and COB in the Three Groups: Healthy, CSCR, and Dry AMD Eyes
Comparison Between Healthy, CSCR, and AMD Eyes
Figure 4 shows representative cases of healthy, CSCR, and AMD eyes. 
CIB Mean R
The Kruskal-Wallis H test showed significant difference of radius of CIB across the three groups (median = 34.14 mm, 38.85 mm, and 27.82 mm in healthy, CSCR, and AMD eyes, respectively; P < 0.001) as shown in Table 2. Dunn's post hoc test for pairwise comparison showed a significant difference between healthy and CSCR eyes (P = 0.066), CSCR and AMD eyes (P < 0.001), and healthy and AMD eyes (P = 0.047) (Table 2Fig. 5A). 
Figure 5.
 
Box and whisker plots for parameters: radius (R) for (A) CIB and (B) COB.
Figure 5.
 
Box and whisker plots for parameters: radius (R) for (A) CIB and (B) COB.
COB Mean R
Comparison among the three groups showed significant difference (median = 31.13 mm, 31.74 mm, and 25.05 mm in healthy, CSCR, and AMD eyes, respectively; P = 0.001). Pairwise comparison using Dunn's post hoc revealed a significant difference between CSCR and AMD and between AMD and healthy eyes (P = 0.001 and 0.050, respectively) (Table 2, Fig. 5B). 
Analysis Based on Generalized Linear Mixed Model
On comparing the three groups, two groups were shown to be significantly different from each other (Wald χ² (2) = 16.64, P = 0.002). On pairwise comparison: CSCR measurements were 18.8% higher than healthy (coefficient = −0.2088, P = 0.002). The difference between AMD and healthy eyes was not significant (coefficient = 0.0399, P = 0.567); but significant difference existed between CSCR and AMD, with CSCR measurements being 28.2% higher than AMD (coefficient = 0.2487, P = 0.005). 
On comparison of R COB versus R CIB across all the three groups, R COB measurements were 10.5% lower than R CIB measurements across all groups (CSCR, healthy and AMD). On assessment of R COB versus R CIB with respect to groups; it was observed that there was a significantly greater difference in boundaries (COB and CIB) in CSCR patients compared to healthy subjects (9% higher) (coefficient = −0.0948, P = 0.013). However, the difference in boundaries in AMD patients was similar to healthy subjects (coefficient = 0.0342, P = 0.145). On analysis of covariates, it was observed that refractive error had no significant effect on the R measurements (coefficient = 0.0988, P = 0.115), and age had no significant effect coefficient = −0.0015, P = 0.241) 
CVI and CT
Mean central CVI was 0.3893 ± 0.0376 in healthy eyes; 0.4096 ± 0.0235 in CSCR eyes; and 0.3519 ± 0.0461 in dry AMD eyes. Comparison of CVI among the three groups showed significant difference (P = 0.003, Kruskal-Wallis test). Pairwise comparison using Dunn's post hoc revealed a significant difference between CSCR and AMD (P = 0.003). Mean SFCT in micrometers was 266.03 ± 79.34 in the healthy eyes, 412.27 ± 90.21 in CSCR eyes, and 221.80 ± 97.19 in dry AMD eyes. The difference among the three groups was significant (P < 0.001; Kruskal-Wallis test), and pairwise comparison using Dunn's post hoc revealed a significant difference between healthy and CSCR (P < 0.001), between CSCR and AMD (P < 0.001), and between AMD and healthy (P = 0.04). 
Regional Analysis
An analysis of different regions (nasal, inferior, temporal, superior, and central) in each group was performed, and then Tables 3A is intragroup and 3B is intergroup comparisons were made. 
Table 3A.
 
Analysis of Radius of Curvature (R) Including Comparison of Different Regions (Nasal, Inferior, Temporal, Superior, and Central) Within a Group (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes): Intragroup Analysis at CIB and COB
Table 3A.
 
Analysis of Radius of Curvature (R) Including Comparison of Different Regions (Nasal, Inferior, Temporal, Superior, and Central) Within a Group (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes): Intragroup Analysis at CIB and COB
Table 3B.
 
Regional Mean Radius of Curvature (R) Compared Between the Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes: Intergroup Analysis at CIB and COB
Table 3B.
 
Regional Mean Radius of Curvature (R) Compared Between the Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes: Intergroup Analysis at CIB and COB
Intragroup Comparisons (Table 3A)
Analysis of CIB and COB radius of curvature in different regions (nasal, inferior, temporal, superior and central) showed statistically significant difference in all the three groups (all P values < 0.001). 
CIB (Radius)
The central region showed a minimum radius of curvature (mean ± SD = 18. 63 ± 10.07 mm) whereas the inferior region had maximum radius (38.34 ± 13.32 mm) in healthy eyes. For CSCR group, inferior region had the maximum radius (45.76 ± 23.21 mm) followed by superior (44.67 ± 14.22 mm), temporal (34.81 ± 13.7 mm), nasal (26.96 ± 16.17 mm), and central (19.84 ± 12.71 mm) (P < 0.001). Similar to healthy and CSCR eyes, central region had the least radius of curvature (14.44 ± 8.66 mm) with the inferior region (33.99 ± 5.59 mm) showing the maximum radius (P < 0.001) in AMD eyes. 
COB (Radius)
Similar to CIB, radius of curvature of central region (9.05 ± 4.74 mm) was minimum, with inferior (29.68 ± 8.15) and temporal (29.1 ± 8.62) regions showing the maximum radius in healthy eyes. In CSCR eyes, there was wide variation in the radius of curvature, with the central region showing the least radius of curvature (8.36 ± 3.72 mm), whereas the temporal (30.28 ± 11.74 mm) and superior regions (30.27 ± 7.19 mm) had a radius of curvature approximately 3.6 times compared to the central region. The AMD group showed similar results, with central (10.14 ± 5.84 mm) and inferior (27.73 ± 4.26 mm) regions showing minimum and maximum radius, respectively. 
The CIB radius is greater than the COB radius in every region in all the three groups. (Healthy: nasal P = 0.1, inferior P < 0.001, temporal P = 0.03, superior P < 0.001, central P < 0.001) (CSCR: nasal P = 0.003; inferior P < 0.001; temporal P = 0.001; superior P < 0.001; central P < 0.001) (AMD: nasal P = 0.1; inferior P < 0.001; temporal P = 0.03; superior P < 0.001; central P < 0.001). Overall (en face region: union of all five regions) radius of curvature at CIB is also greater than COB in all the three groups (P < 0.001in each of the three groups). (Test: Friedman test, P < 0.001, post-hoc: paired t test if assumptions were met, and Wilcoxon signed rank test if assumptions were not met): 
  • 1. Row 1: regional R in five regions (nasal, inferior, temporal, superior, central) at CIB in healthy eyes compared with each other. Hypothesis: There is no difference in distribution of R across the five regions. Normality (Shapiro Wilk test): not satisfied; Variance (Levene's test): satisfied. Friedman test, P < 0.001. Result: Hypothesis rejected, the difference in distribution of region-wise radius of curvature is statistically significant.
  • 2. Row 2: regional R in five regions at CIB of CSCR eyes compared with each other. Hypothesis: There is no difference in distribution of R across the five regions. Normality (Shapiro Wilk test): not satisfied; Variance (Levene's test): satisfied. Friedman test, P < 0.001. Result: Hypothesis rejected, the difference in distribution of region-wise radius of curvature is statistically significant.
  • 3. Row 3: regional R in five regions at CIB in dry AMD eyes compared with each other. Hypothesis: There is no difference in distribution of R across the five regions. Normality (Shapiro Wilk test): not satisfied; Variance (Levene's test): satisfied. Friedman test, P < 0.001. Result: Hypothesis rejected, the difference in distribution of region-wise radius of curvature is statistically significant.
  • 4. Row 4: regional R in five regions at COB in healthy eyes compared with each other. Hypothesis: There is no difference in distribution of R across the five regions. Normality (Shapiro Wilk test): not satisfied; Variance (Levene's test): satisfied. Friedman test, P < 0.001. Result: Hypothesis rejected, the difference in distribution of region-wise radius of curvature is statistically significant.
  • 5. Row 5: regional R in five regions at COB of CSCR eyes compared with each other. Hypothesis: There is no difference in distribution of radius across the five regions. Normality (Shapiro Wilk test): not satisfied; Variance (Levene's test): not satisfied. Friedman test, P < 0.001. Result: Hypothesis rejected, the difference in distribution of region-wise radius of curvature is statistically significant.
  • 6. Row 6: regional R in five regions at COB in dry AMD eyes compared with each other. Hypothesis: There is no difference in distribution of R across the five regions. Normality (Shapiro Wilk test): not satisfied; Variance (Levene's test): satisfied. Friedman test, P < 0.001. Result: Hypothesis rejected, the difference in distribution of region-wise radius of curvature is statistically significant.
Intergroup Comparisons (Table 3B)
We also performed intergroup analysis comparing each region with corresponding region in the other two groups (for instance, nasal region in healthy eyes was compared to nasal region in CSCR and AMD eyes). 
CIB (Radius)
Superior region showed significant variations across three groups (P < 0.001). Dunn's post hoc pairwise comparison showed significant difference between AMD and CSCR (P < 0.001) and between CSCR and healthy eyes (P = 0.003). 
COB (Radius)
None of the regions showed a significant variation among three groups: 
  • 1. First row: Variable 1 (V1): nasal radius at CIB in healthy eyes (R healthy CIB nasal); V2: R CSCR CIB nasal; V3: R AMD CIB nasal. Hypothesis: There is no difference in distribution of nasal R across the three groups at CIB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene test): satisfied. Kruskal Wallis test, P = 0.21. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 2. Second row: V1: R healthy CIB inferior; V2: R CSCR CIB inferior; V3: R AMD CIB inferior. Hypothesis: There is no difference in distribution of inferior R across the three groups at CIB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene test): not satisfied. Kruskal-Wallis test, P = 0.11. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 3. Third row: V1: R healthy CIB temporal; V2: R CSCR CIB temporal; V3: R AMD CIB temporal. Hypothesis: There is no difference in distribution of temporal R across the three groups at CIB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Kruskal-Wallis test, P = 0.13. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 4. Fourth row: V1: R healthy CIB superior; V2: R CSCR CIB superior; V3: R AMD CIB superior. Hypothesis: There is no difference in distribution of superior R across the three groups at CIB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene test): not satisfied. Kruskal-Wallis test, P < 0.001. Result: Hypothesis is rejected, the difference in distribution of radius is statistically significant. Dunn's post hoc pairwise comparison showed AMD – CSCR (P < 0.001); AMD – Healthy (P = 0.440); CSCR – Healthy (P = 0.003).
  • 5. Fifth row: V1: R healthy CIB central; V2: R CSCR CIB central; V3: R AMD CIB central. Hypothesis: There is no difference in distribution of central R across the three groups at CIB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): not satisfied. Kruskal-Wallis test, P = 0.07. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 6. Sixth row: V1: R healthy COB nasal; V2: R CSCR COB nasal; V3: R AMD COB nasal. Hypothesis: There is no difference in distribution of nasal R across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Kruskal-Wallis test, P = 0.53. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 7. Seventh row: R healthy COB inferior; V2: R CSCR COB inferior; V3: R AMD COB inferior. Hypothesis: There is no difference in distribution of inferior R across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): not satisfied. Kruskal-Wallis test, P = 0.93. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 8. Eighth row: R healthy COB temporal; V2: R CSCR COB temporal; V3: R AMD COB temporal. Hypothesis: There is no difference in distribution of temporal R across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Kruskal-Wallis test, P = 0.12. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 9. Ninth row: R healthy COB superior; V2: R CSCR COB superior; V3: R AMD COB superior. Hypothesis: There is no difference in distribution of superior R across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Kruskal-Wallis test, P = 0.08. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
  • 10. Tenth row: R healthy COB central; V2: R CSCR COB central; V3: R AMD COB central. Hypothesis: There is no difference in distribution of central R across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): not satisfied. Kruskal-Wallis test, P = 0.69. Result: Hypothesis is accepted, the difference in distribution of radius is not statistically significant.
An age-matched analysis between healthy eyes < 55 years of age and CSCR eyes was also performed. Similarly, an age-matched analysis of healthy eyes ≥ 55 years of age with dry AMD eyes was performed (Supplementary Tables S6S10). 
A separate sub analysis including comparison of differences in CIB/COB (radius) parameters in each region (nasal, inferior, temporal, superior) with the central region was performed among the three groups (i.e., nasal – central; inferior – central; temporal – central; superior – central). Only the CIB radius (superior – central) showed significant difference among the three groups (P = 0.004) as shown in Table 4. On the other hand, the COB radius differential in two regions (temporal – central, P < 0.001; superior – central, P = 0.01) showed statically significant results (Table 4). 
  • 1. First row: variable 1 (V1): radius of curvature in nasal region minus radius of curvature in central region at CIB in healthy eyes (R [nasal – central] healthy CIB); V2: R (nasal – central) CSCR CIB; V3: R (nasal – central) AMD CIB. Hypothesis: The radius of curvature difference between Central and Nasal regions is not different from each other across the 3 groups (healthy, CSCR and AMD). Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Test: Kruskal-Wallis, P = 0.21. Result: Hypothesis accepted.
  • 2. Second row: V1: R (inferior – central) healthy CIB; V2: R (inferior – central) CSCR CIB; V3: R (inferior – central) AMD CIB. Hypothesis: The radius of curvature difference between the central and inferior regions is not different from each other across the three groups (healthy, CSCR, and AMD). Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): not satisfied. Test: Kruskal-Wallis, P = 0.26. Result: Hypothesis accepted.
  • 3. Third row: V1: R (temporal – central) healthy CIB; V2: R (temporal – central) CSCR CIB; V3: R (temporal – central) AMD CIB. Hypothesis: The radius of curvature difference between Central and temporal regions is not different from each other across the three groups (healthy, CSCR, and AMD). Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Test: Kruskal-Wallis, P = 0.65. Result: Hypothesis accepted.
  • 4. Fourth row: V1: R (superior – central) healthy CIB; V2: R (superior – central) CSCR CIB; V3: R (superior – central) AMD CIB. Hypothesis: The radius of curvature difference between the central and superior regions is not different from each other across the three groups (healthy, CSCR and AMD). Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Test: Kruskal-Wallis, P = 0.004. Result: Hypothesis rejected, the difference in regional radius between central and superior region is significantly different across the three groups. Dunn's post hoc pairwise comparison showed AMD – CSCR (P = 0.021); AMD – Healthy (P = 1.000); CSCR – Healthy (P = 0.007)
  • 5. Fifth row: V1: R (nasal – central) healthy COB; V2: R (nasal – central) CSCR COB; V3: R (nasal – central) AMD COB. Hypothesis: The radius of curvature difference between the central and nasal regions is not different from each other across the three groups (healthy, CSCR, and AMD). Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Test: Kruskal-Wallis, P = 0.61. Result: Hypothesis accepted.
  • 6. Sixth row: V1: R (inferior – central) healthy COB; V2: R (inferior – central) CSCR COB; V3: R (inferior – central) AMD COB. Hypothesis: The radius of curvature difference between the central and inferior regions is not different from each other across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Test: Kruskal-Wallis, P = 0.29. Result: Hypothesis accepted.
  • 7. Seventh row: V1: R (temporal – central) healthy COB; V2: R (temporal – central) CSCR COB; V3: R (temporal – central) AMD COB. Hypothesis: The radius of curvature difference between the central and temporal regions is not different from each other across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Test: Kruskal-Wallis, P < 0.001. Result: Hypothesis rejected, the difference in regional radius between central and temporal region is significantly different across the three groups. Dunn's post hoc pairwise comparison showed AMD – CSCR (P < 0.001); AMD – Healthy (P < 0.001); CSCR – Healthy (P = 0.823)
  • 8. Eighth row: R (superior – central) healthy COB; V2: R (superior – central) CSCR COB; V3: R (superior – central) AMD COB. Hypothesis: The radius of curvature difference between the central and superior regions is not different from each other across the three groups at COB. Normality (Shapiro Wilk test): not satisfied, Variance (Levene's test): satisfied. Test: Kruskal-Wallis, P = 0.01. Result: Hypothesis rejected, the difference in regional radius between central and superior region is significantly different across the three groups. Dunn's post hoc pairwise comparison showed AMD – CSCR (P = 0.019); AMD – Healthy (P = 1.000); CSCR – Healthy (P = 0.045)
Table 4.
 
Analysis Showing Difference in Mean R Between Each Region and Central Region Compared Across Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes) at CIB and COB
Table 4.
 
Analysis Showing Difference in Mean R Between Each Region and Central Region Compared Across Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes) at CIB and COB
Correlation Analysis
Pearson/ Spearman correlation coefficient was calculated between R CIB and R COB in the three groups (Table 5). Normality of the data was assessed using Shapiro-Wilk Test. For normally distributed variables, Pearson's correlation was used, and when the Shapiro-Wilk test for normal distribution was not satisfied, the Spearman rank correlation test was used. R CIB and R COB had a strong correlation in all the three groups (Healthy: r = 0.91, P < 0.001; CSCR: r = 0.68, P < 0.001; AMD: r = 0.84, P < 0.001). 
Table 5.
 
Correlation Between Various Parameters Within the Group
Table 5.
 
Correlation Between Various Parameters Within the Group
Similarly, we also performed correlation analysis of CIB and COB parameters with SFCT (Table 5). There was moderate positive correlation of R CIB and R COB with SFCT in healthy eyes (r = 0.62 and 0.43, respectively; P < 0.001 and 0.01). In contrast, CSCR (r = −0.02 and −0.24; P = 0.91 and 0.19) and AMD eyes (r = −0.06, −0.05; P = 0.76 and 0.77, respectively) showed weak or negligible negative correlation of R CIB and R COB with SFCT. 
Discussion
Our results show that overall the contour of choroid was the flattest in CSCR and steepest in AMD. The R was highest for CSCR at COB, as well as CIB. The reverse was true for AMD. CSCR measurements were 28.2% higher than AMD and 18.8% higher than healthy eyes. On analysis of different regions in CIB and COB (nasal, inferior, temporal, superior, and central) within each group (intragroup); the radius differed significantly between each region in all the three groups. The CIB is flatter than the COB in healthy eyes, CSCR eyes, as well as AMD eyes. This was significant in the overall radius (en-face) (R at CIB > R at COB) in all the three groups as well as across all regions (except nasal region in healthy and AMD eyes where statistical significance was not reached). The difference between CIB and COB was most pronounced in CSCR eyes (9% higher than healthy eyes). 
Intragroup regional analysis showed that the radius of curvature was the minimum for the central region in all groups at CIB and COB. In CSCR at COB, this difference was drastic because the radius of curvature in the central region was 3.6 times lower than in the temporal and superior regions. 
On performing the regional analysis across the three groups (intergroup) at CIB, the difference across the three groups was significant in superior region, with CSCR having the highest values and AMD having the lowest values. On intergroup analysis at COB, the difference across the three groups was not significant in any region. This is notable because at COB, the overall radius across the three groups is significantly different from each other, but on evaluation of individual regions, this significance value is lost. 
As mentioned before, in COB and CIB, the radius of curvature is the maximum in CSCR followed by healthy eyes and then AMD eyes (RCSCR> Rhealthy >RAMD). This implies that the surface is most convex for AMD and flattest for CSCR. This is contrary to an article published recently by Izumi et al.22 However, it is important to understand two facts: (i) choroidal contour topography in 3D has never been studied before, and so the inferences have been based on the choroidal thickness measurements22 and (ii) error of image aspect ratio.23 An anterior-posterior stretching of OCT images is used in many commercial OCT machines, for example in Spectralis SD-OCT (Heidelberg Engineering, Heidelberg, Germany) the OCT images are stretched three times in the axial direction by default. This is indicated on the display and device output by scale bars. OCT image can also be stretched because of OCT sampling density being higher in the axial direction as compared to the transverse direction. When an image is “non-stretched,” the pixel aspect ratio is 1:1. In Spectralis SD-OCT (Heidelberg Engineering), it is stretched axially such that the pixel aspect ratio is 3:1. Measurements using such images without making the correction are erroneous. Another issue is that the lines that are perpendicular in a stretched image are likely not perpendicular in a non-stretched 1:1 image, and thus the measurements may be inaccurate.23 Unfortunately, most of the published studies do not mention the image aspect ratio and whether it was considered and corrections made during the measurements. It may not be possible for the journal reviewer or other readers to detect this error. In our study, we used the raw images of OCT B -scan. Figure 6 shows images in 1:1 aspect ratio and 3:1 aspect ratio. Finally, subsequent reports24 published on wide-field CT in CSCR show that the choroid is thickened not only in the subfoveal area but also in the periphery, and this is also contrary to the report of Izumi et al.22 
Figure 6.
 
OCT B scan of an eye with CSCR with COB radius = 43.67 and CIB radius 55.16 (A) in a 1:1 aspect ratio and (B) in a 3:1 aspect ratio. OCT B scan of a healthy eye with COB radius = 31.47 and CIB radius 33.61 (C) in a 1:1 aspect ratio (D) in a 3:1 aspect ratio. OCT B scan of an eye with AMD with COB radius = 21.09 and CIB radius 21.91 (E) in a 1:1 aspect ratio and (F) in a 3:1 aspect ratio.
Figure 6.
 
OCT B scan of an eye with CSCR with COB radius = 43.67 and CIB radius 55.16 (A) in a 1:1 aspect ratio and (B) in a 3:1 aspect ratio. OCT B scan of a healthy eye with COB radius = 31.47 and CIB radius 33.61 (C) in a 1:1 aspect ratio (D) in a 3:1 aspect ratio. OCT B scan of an eye with AMD with COB radius = 21.09 and CIB radius 21.91 (E) in a 1:1 aspect ratio and (F) in a 3:1 aspect ratio.
It has been well established that the CT increases in CSCR and that there is vascular engorgement. When the CT increases, there is dilation of the Haller layer vessels and attenuation of the inner choroid (Sattler's layer and choriocapillaries).25 This could be because of limited available space for expansion of choroid. The choroid is thickened not only in the subfoveal area but also in the periphery.24 Thus the choroidal contour flattens in the CSCR. Wide-field CT analysis in healthy eyes has revealed that CT is the maximum in the subfoveal area and decreases toward the periphery.26 This may explain a steeper contour of healthy eyes as compared to eyes with CSCR. 
Correlation between CIB and COB parameters was significant in all the three groups. However, the correlation of CIB radius and COB radius with SFCT was moderate (positive) in healthy eyes, and negligible/weak in AMD eyes and CSCR eyes. This may be indicative that choroidal contour may not be easily predictable based on SFCT, which is the most used choroidal measurement in various diseases. It is important to obtain peripheral CT to determine the relation between choroidal contour and wide field CT. 
So far, we have discussed the overall contour at CIB and COB in the three groups. The local variations in the contour in each of the three groups were notable too. The center had the lowest radius of curvature at CIB, as well as COB in each of the three groups. In other words, the center was the steepest among all the regions within a group. Furthermore, the difference in the change of contour was also significant within the group in all three groups (at CIB: the difference between the superior region and the central region was significant; at COB: the central region was significantly different from the inferior, temporal, and superior regions). 
In this study, an automated methodology of delineation of CIB and COB followed by contour mapping was performed. In particular, choroid layer segmentation was achieved by combining our previously reported machine learning method and volumetric smoothing methods. Limitations of this study include its retrospective nature and limited sample size. We did not obtain wide-field peripheral CT in our cases and thus were unable to determine the correlation between choroidal contour and peripheral CT. Because of the small sample size in each group, we were unable to evaluate how age influenced choroidal contour in healthy eyes and how active or resolved CSCR influenced choroidal contour. There have been studies27,28 pointing toward the mismatch of posterior segment shape when seen on current-generation OCT machines versus when seen on magnetic resonance imaging or ultrasound scanning. Accounting for the factors leading to such discrepancy or incorporation of algorithms meant for the correction of the retinal OCT shape would help in the development of more sophisticated image-based morphological measurements. 
In conclusion, our novel approach to evaluate choroidal surface demonstrates that the contour of the choroid at CIB and COB was the flattest in CSCR and steepest in AMD. The central region was the steepest among all regions in both healthy and diseased eyes. In future studies, we plan to evaluate the choroidal contour in various other diseases including high myopia, as well as correlate the choroidal contour with visual function. 
Acknowledgments
Supported by NIH CORE Grant P30 EY08098 to the Department of Ophthalmology, the Eye and Ear Foundation of Pittsburgh, and from an unrestricted grant from Research to Prevent Blindness, New York, NY. 
Disclosure: S. Arora, None; S.R. Singh, None; S.C. Vupparaboina, None; B. Rosario, None; M.N. Ibrahim, None; A. Selvam, None; A. Zarnegar, None; S. Harihar, None; V. Sant, None; J.A.Sahel, Pixium Vision (F), GenSight Biologics (F), Sparing Vision (F), Prophesee (F), Chronolife (F); K.K. Vupparaboina, None; J. Chhablani, Salutaris Medical Devices (C), Allergan (C), OD-OS (C) 
References
Xie R, Qiu B, Chhablani J, Zhang X. Evaluation of choroidal thickness using optical coherent tomography: a review. Front Med (Lausanne). 2021; 8: 783519. [CrossRef] [PubMed]
Agrawal R, Chhablani J, Tan KA, Shah S, Sarvaiya C, Banker A. Choroidal vascularity index in central serous chorioretinopathy. Retina. 2016; 36: 1646–1651. [CrossRef] [PubMed]
Kim RY, Chung DH, Kim M, Park YH. Use of choroidal vascularity index for choroidal structural evaluation in central serous chorioretinopathy with choroidal neovascularization. Retina. 2020; 40: 1395–1402. [CrossRef] [PubMed]
Sacconi R, Vella G, Battista M, et al. Choroidal vascularity index in different cohorts of dry age-related macular degeneration. Transl Vis Sci Technol. 2021; 10(12): 26. [CrossRef] [PubMed]
Wei X, Ting DSW, Ng WY, Khandelwal N, Agrawal R, Cheung CMG. Choroidal vascularity index: a novel optical coherence tomography based parameter in patients with exudative age-related macular degeneration. Retina. 2017; 37: 1120–1125. [CrossRef] [PubMed]
Arora S, Rosario B, Mohammed AR, et al. Regression patterns of central serous chorioretinopathy using en face optical coherence tomography. Graefes Arch Clin Exp Ophthalmol. 2022; 260: 2475–2481. [PubMed]
Goud A, Singh SR, Sahoo NK, et al. New insights on choroidal vascularity: a comprehensive topographic approach. Invest Ophthalmol Vis Sci. 2019; 60: 3563–3569. [CrossRef] [PubMed]
Arora S, Singh SR, Rosario B, et al. Three-dimensional choroidal contour mapping in healthy population. Sci Rep. 2024; 14(1): 6210. [CrossRef] [PubMed]
Nkrumah G, Maltsev DS, Manuel PA, et al. Current choroidal imaging findings in central serous chorioretinopathy. Vision (Basel). 2020; 4(4): 44. [CrossRef] [PubMed]
Koh LHL, Agrawal R, Khandelwal N, Sai Charan L, Chhablani J. Choroidal vascular changes in age-related macular degeneration. Acta Ophthalmol. 2017; 95(7): e597–e601. [CrossRef] [PubMed]
Lejoyeux R, Atia R, Vupparaboina KK, et al. En-face analysis of short posterior ciliary arteries crossing the sclera to choroid using wide-field swept-source optical coherence tomography. Sci Rep. 2021; 11(1): 8732. [CrossRef] [PubMed]
Ibrahim MN, Bashar SB, Rasheed MA, et al. Volumetric quantification of choroid and Haller's sublayer using OCT scans: an accurate and unified approach based on stratified smoothing. Comput Med Imaging Graph. 2022; 99: 102086. [CrossRef] [PubMed]
Vupparaboina KK, Selvam A, Suthaharan S, et al. Automated choroid layer segmentation based on wide-field ss-oct images using deep residual encoder-decoder architecture. Invest Ophthalmol Vis Sci. 2021; 62: 2162.
Cleaveland WS . Robust locally weighted regression and smoothening scatterplots. J Am Stat Assoc. 1979; 74: 829–836. [CrossRef]
Medioni GTCK, Lee MS. Tensor voting: theory and applications. In: Proceedings of RFIA. 2000.
Vupparaboina KK, Nizampatnam S, Chhablani J, Richhariya A, Jana S. Automated estimation of choroidal thickness distribution and volume based on OCT images of posterior visual section. Comput Med Imaging Graph. 2015; 46 Pt 3: 315–327. [PubMed]
Claxton D . Surface Curvature, MATLAB Central File Exchange [Available from: https://www.mathworks.com/matlabcentral/fileexchange/11168-surface-curvature. Accessed September 4, 2023.
Jennings A . Sphere Fit (least squared) MATLAB Central File Exchange. 2022 [Available from: https://www.mathworks.com/matlabcentral/fileexchange/34129-sphere-fit-least-squared. Accessed September 4, 2023.
Kasa I . A circle fitting procedure and its error analysis. IEEE Trans Instrument Meas. 1976; IM-25: 8–14. [CrossRef]
Umbach D, Jones KN. A few methods for fitting circles to data. IEEE Trans Instrument Meas. 2003; 52: 1881–1885. [CrossRef]
Schober P, Boer C, Schwarte LA. Correlation coefficients: appropriate use and interpretation. Anesth Analg. 2018; 126: 1763–1768. [CrossRef] [PubMed]
Izumi T, Maruko I, Kawano T, Sakaihara M, Iida T. Morphological differences of choroid in central serous chorioretinopathy determined by ultra-widefield optical coherence tomography. Graefes Arch Clin Exp Ophthalmol. 2022; 260: 295–301. [CrossRef] [PubMed]
Sigal IA, Schuman JS, Ishikawa H, Kagemann L, Wollstein G. A problem of proportions in OCT-based morphometry and a proposed solution. Invest Ophthalmol Vis Sci. 2016; 57: 484–485. [CrossRef] [PubMed]
Ishikura M, Muraoka Y, Nishigori N, et al. Widefield choroidal thickness of eyes with central serous chorioretinopathy examined by swept-source OCT. Ophthalmol Retina. 2022; 6: 949–956. [CrossRef] [PubMed]
Cheung CMG, Lee WK, Koizumi H, Dansingani K, Lai TYY, Freund KB. Pachychoroid disease. Eye (Lond). 2019; 33: 14–33. [CrossRef] [PubMed]
Rasheed MA, Singh SR, Invernizzi A, et al. Wide-field choroidal thickness profile in healthy eyes. Sci Rep. 2018; 8(1): 17166. [CrossRef] [PubMed]
Kuo AN, McNabb RP, Chiu SJ, et al. Correction of ocular shape in retinal optical coherence tomography and effect on current clinical measures. Am J Ophthalmol. 2013; 156: 304–311. [CrossRef] [PubMed]
Kuo AN, Verkicharla PK, McNabb RP, et al. Posterior eye shape measurement with retinal OCT compared to MRI. Invest Ophthalmol Vis Sci. 2016; 57(9): Oct196–Oct203. [CrossRef] [PubMed]
Figure 1.
 
Flowchart depicting categorization of eyes into three groups and inclusion and exclusion criteria.
Figure 1.
 
Flowchart depicting categorization of eyes into three groups and inclusion and exclusion criteria.
Figure 2.
 
Comparison of choroid segmentations by manual and proposed automated methods. Two scans of a healthy eye, two scans of an eye with CSCR, and two scans of an eye with AMD have been shown to compare manual versus automated segmentation. For each eye sequential OCT B scans were used to generate OCT volume data.
Figure 2.
 
Comparison of choroid segmentations by manual and proposed automated methods. Two scans of a healthy eye, two scans of an eye with CSCR, and two scans of an eye with AMD have been shown to compare manual versus automated segmentation. For each eye sequential OCT B scans were used to generate OCT volume data.
Figure 3.
 
The division into five regions (central, nasal, temporal, superior, inferior) to facilitate the study of region-wise statistics. In the GUI, the user selects the foveal center on the RPE en face image. The user also selects two more points: any point nasal to the fovea (based on the optic disc) and any point superior to the fovea (based on optic disc and fovea). Based on the user input, a predefined grid mask of size 1500 × 1500 pixels (much larger than the en-face image size of 1024 × 750 pixels) consisting of five regions is overlaid on the RPE en-face image with the grid's center matching with the foveal center. More specifically, the grid mask has a circle of radius 1 mm (around ∼85 pixels) at its center and the remaining region is four equal regions separated by lines positioned at 45°, 135°, 225°, and 315°. The circle of radius 1 mm around the foveal center constitutes the central region. Further, based on the user input, nasal and superior regions are identified. The region opposite to the nasal corresponds to the temporal region, and, similarly, the region opposite to the superior region is the inferior region.
Figure 3.
 
The division into five regions (central, nasal, temporal, superior, inferior) to facilitate the study of region-wise statistics. In the GUI, the user selects the foveal center on the RPE en face image. The user also selects two more points: any point nasal to the fovea (based on the optic disc) and any point superior to the fovea (based on optic disc and fovea). Based on the user input, a predefined grid mask of size 1500 × 1500 pixels (much larger than the en-face image size of 1024 × 750 pixels) consisting of five regions is overlaid on the RPE en-face image with the grid's center matching with the foveal center. More specifically, the grid mask has a circle of radius 1 mm (around ∼85 pixels) at its center and the remaining region is four equal regions separated by lines positioned at 45°, 135°, 225°, and 315°. The circle of radius 1 mm around the foveal center constitutes the central region. Further, based on the user input, nasal and superior regions are identified. The region opposite to the nasal corresponds to the temporal region, and, similarly, the region opposite to the superior region is the inferior region.
Figure 4.
 
Sequential OCT B scans were used to generate OCT volume. Choroidal contour on CIB and COB and choroidal thickness map have been demonstrated in representative cases of AMD eye, healthy eye, and eye with CSCR.
Figure 4.
 
Sequential OCT B scans were used to generate OCT volume. Choroidal contour on CIB and COB and choroidal thickness map have been demonstrated in representative cases of AMD eye, healthy eye, and eye with CSCR.
Figure 5.
 
Box and whisker plots for parameters: radius (R) for (A) CIB and (B) COB.
Figure 5.
 
Box and whisker plots for parameters: radius (R) for (A) CIB and (B) COB.
Figure 6.
 
OCT B scan of an eye with CSCR with COB radius = 43.67 and CIB radius 55.16 (A) in a 1:1 aspect ratio and (B) in a 3:1 aspect ratio. OCT B scan of a healthy eye with COB radius = 31.47 and CIB radius 33.61 (C) in a 1:1 aspect ratio (D) in a 3:1 aspect ratio. OCT B scan of an eye with AMD with COB radius = 21.09 and CIB radius 21.91 (E) in a 1:1 aspect ratio and (F) in a 3:1 aspect ratio.
Figure 6.
 
OCT B scan of an eye with CSCR with COB radius = 43.67 and CIB radius 55.16 (A) in a 1:1 aspect ratio and (B) in a 3:1 aspect ratio. OCT B scan of a healthy eye with COB radius = 31.47 and CIB radius 33.61 (C) in a 1:1 aspect ratio (D) in a 3:1 aspect ratio. OCT B scan of an eye with AMD with COB radius = 21.09 and CIB radius 21.91 (E) in a 1:1 aspect ratio and (F) in a 3:1 aspect ratio.
Table 1.
 
Demographic Features Including Healthy Eyes, CSCR Eyes, and Dry AMD Eyes
Table 1.
 
Demographic Features Including Healthy Eyes, CSCR Eyes, and Dry AMD Eyes
Table 2.
 
Overall Features (En-Face Region) at CIB and COB in the Three Groups: Healthy, CSCR, and Dry AMD Eyes
Table 2.
 
Overall Features (En-Face Region) at CIB and COB in the Three Groups: Healthy, CSCR, and Dry AMD Eyes
Table 3A.
 
Analysis of Radius of Curvature (R) Including Comparison of Different Regions (Nasal, Inferior, Temporal, Superior, and Central) Within a Group (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes): Intragroup Analysis at CIB and COB
Table 3A.
 
Analysis of Radius of Curvature (R) Including Comparison of Different Regions (Nasal, Inferior, Temporal, Superior, and Central) Within a Group (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes): Intragroup Analysis at CIB and COB
Table 3B.
 
Regional Mean Radius of Curvature (R) Compared Between the Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes: Intergroup Analysis at CIB and COB
Table 3B.
 
Regional Mean Radius of Curvature (R) Compared Between the Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes: Intergroup Analysis at CIB and COB
Table 4.
 
Analysis Showing Difference in Mean R Between Each Region and Central Region Compared Across Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes) at CIB and COB
Table 4.
 
Analysis Showing Difference in Mean R Between Each Region and Central Region Compared Across Three Groups (Healthy Eyes, CSCR Eyes, and Dry AMD Eyes) at CIB and COB
Table 5.
 
Correlation Between Various Parameters Within the Group
Table 5.
 
Correlation Between Various Parameters Within the Group
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×