December 2016
Volume 57, Issue 15
Open Access
Retina  |   December 2016
OCT-Leakage: A New Method to Identify and Locate Abnormal Fluid Accumulation in Diabetic Retinal Edema
Author Affiliations & Notes
  • José Cunha-Vaz
    Association for Innovation and Biomedical Research on Light and Image, Coimbra, Portugal
  • Torcato Santos
    Association for Innovation and Biomedical Research on Light and Image, Coimbra, Portugal
  • Luísa Ribeiro
    Association for Innovation and Biomedical Research on Light and Image, Coimbra, Portugal
  • Dalila Alves
    Association for Innovation and Biomedical Research on Light and Image, Coimbra, Portugal
  • Inês Marques
    Association for Innovation and Biomedical Research on Light and Image, Coimbra, Portugal
  • Morton Goldberg
    Wilmer Eye Institute, Baltimore, Maryland, United States
  • Correspondence: José Cunha-Vaz, AIBILI, Azinhaga de Santa Comba, Celas 3000-548, Coimbra, Portugal; cunhavaz@aibili.pt
Investigative Ophthalmology & Visual Science December 2016, Vol.57, 6776-6783. doi:10.1167/iovs.16-19999
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      José Cunha-Vaz, Torcato Santos, Luísa Ribeiro, Dalila Alves, Inês Marques, Morton Goldberg; OCT-Leakage: A New Method to Identify and Locate Abnormal Fluid Accumulation in Diabetic Retinal Edema. Invest. Ophthalmol. Vis. Sci. 2016;57(15):6776-6783. doi: 10.1167/iovs.16-19999.

      Download citation file:


      © 2017 Association for Research in Vision and Ophthalmology.

      ×
  • Supplements
Abstract

Purpose: To identify retinal extracellular fluid changes and their correlation with increased retinal thickness (RT) in eyes with subclinical and clinical macular edema in diabetes type 2.

Methods: A cohort of 48 eyes from 48 type 2 diabetic patients with mild or moderate nonproliferative diabetic retinopathy (Early Treatment Diabetic Retinopathy Study levels 20/35) were classified as having normal RT (10), subclinical macular edema (30), or clinical macular edema (8). They were examined with Cirrus spectral-domain optical coherence tomography (OCT) at baseline visits (ClinicalTrials.gov number, NCT01145599) in the Coimbra center. Results from automated analysis of the retinal extracellular space, using our OCT-Leakage algorithm to identify sites of low optical reflectivity, were compared with those from a control group of 25 healthy eyes.

Results: The highest increases in RT in the eyes with subclinical and clinical macular edema were found in the inner nuclear layer (INL). These increases were, on average, 49.9% in subclinical macular edema and 104.7% in clinical macular edema. Extracellular space increases in the INL that were identified with the OCT-Leakage algorithm showed a strong positive correlation with the increases in RT in the central subfield (r = 0.71, P < 0.001).

Conclusions: Increases in number of sites with lower optical reflectivity positively correlate with the increase in RT in the initial stages of macular edema in diabetes type 2. Diabetic macular edema is represented mainly by extracellular fluid accumulation that preferentially involves the INL of the retina.

Diabetic retinopathy is one of the leading causes of blindness in developed countries.1 The loss of vision is due to two major vision-threatening complications, proliferative retinopathy and center-involving macular edema.2 
With optical coherence tomography (OCT) it became possible to image the retina in vivo and to evaluate retinal edema by measuring retinal thickness (RT) with high accuracy. Recent work performed in a large cohort of patients with diabetes type 2 by the European Vision Institute Clinical Research Network (EVICR.net, in the public domain) showed that the increase in RT occurring in diabetic eyes with macular edema is predominately located in the inner nuclear layer (INL), extending to the neighboring retinal layers.3 This suggests that diabetic macular edema is mainly due to extracellular fluid accumulation, probably resulting from alteration of the blood–retinal barrier (BRB), one of the earliest changes occurring in the retina in diabetes.4,5 
We report here the application of a new algorithm, herein designated as OCT-Leakage. We developed this technique to perform automated analysis of the retinal extracellular space, using spectral-domain optical coherence tomography (SD-OCT) in a group of eyes with subclinical and clinical center-involving diabetic macular edema. 
Methods
Patients and Control Populations
This analysis was performed in the context of a prospective, multicenter, observational study, designed to follow eyes/patients with initial stages of nonproliferative diabetic retinopathy (NPDR). The results of this study and its procedures are described by Ribeiro et al.6 Informed consent was obtained from each patient after explanation of the nature of the study and before any study procedure. The tenets of the Declaration of Helsinki were followed, and approval was obtained from each institutional review board (ClinicalTrials.gov number, NCT01145599). 
Only the 48 eyes from 48 diabetic patients who were examined in the Coimbra Centre (Association for Innovation and Biomedical Research on Light and Image) with Cirrus SD-OCT (Carl Zeiss Meditec, Inc., Dublin, CA, USA) are reported here. 
To serve as a control population, we included 25 eyes with normal eye examinations from 25 nondiabetic patients, aged 39 to 55 years. 
Optical Coherence Tomography
We used the high-definition spectral-domain Cirrus 5000 OCT (Carl Zeiss Meditec, Inc.), which allows the acquisition of volumes of 200 × 200 × 1024 or 512 × 128 × 1024 voxels from a region of 6000 × 6000 × 2000 μm3. To identify eyes with increased RT in the central subfield (clinical and subclinical macular edema) and in the inner and outer rings, the reference values established by DRCR.net were used: 
For clinical macular edema (CME) (ClinicalTrials.gov number, NCT01909791)7
  •  
    RT ≥ 290 μm in women and ≥ 305 μm in men for Cirrus SD-OCT
  •  
    subclinical macular edema (SME)810:
  •  
    RT > 260 μm and < 290 μm in women and > 275 μm and < 305 μm in men for Cirrus SD-OCT
For the inner and outer rings (ClinicalTrials.gov number, NCT01331005)11
  •  
    Normal RT: no more than one area above the normal range (normal mean + 2 standard deviations) and no area 15 μm above the normal range
  •  
    Increased RT: at least two areas above the normal range and/or one area 15 μm above the normal range
Fulfilling these criteria, 30 eyes presented SME, 8 had CME, and 10 had NPDR without evidence of subclinical or clinical edema. 
OCT-Leakage: Analysis of Retinal Extracellular Fluid
Cell nuclei in the retina have higher reflectivity than water-filled areas. Therefore, it is considered that retinal extracellular fluid distribution will be identified by sites of low OCT optical reflectivity. To be able to identify abnormal increases in extracellular space, we first established a threshold value for optical reflectivity from A-scans obtained from a series of healthy control eyes using HD-OCT Cirrus 5000. The threshold value is established for the OCT device used, as it is dependent on the equipment (Fig. 1). The threshold was chosen considering the full retina scan and represents a value set below the mean plus 2 standard deviations of the value obtained in the healthy control eyes. We used the 8-bit reduced dataset of the Zeiss Cirrus 5000. The data were exported using a licensed feature of the Cirrus that exports 8-bit binary data for research purposes. Only scans with signal strength of 7 and above were considered. The data were processed by using the 8-bit reduced data set of the Zeiss Cirrus 5000. They were gathered using a licensed feature that exports 8-bit binary data for research purposes. 
Figure 1
 
Cirrus SD-OCT A-scan optical reflectivity profiles. (a) Full-length SD-OCT A-scan from a healthy subject. (b, c) Detail of SD-OCT A-scan in the retina for a healthy subject (bottom left) and for a NPDR patient with CME (bottom right). Spectral-domain OCT A-scan optical reflectivity from each of the segmented layers is differentiated by vertical dashed lines. Optical reflectivity threshold is shown as a horizontal line. In (c), note the decreased reflectivity registered in the GCL+IPL, in the INL, and in the ONL+IS.
Figure 1
 
Cirrus SD-OCT A-scan optical reflectivity profiles. (a) Full-length SD-OCT A-scan from a healthy subject. (b, c) Detail of SD-OCT A-scan in the retina for a healthy subject (bottom left) and for a NPDR patient with CME (bottom right). Spectral-domain OCT A-scan optical reflectivity from each of the segmented layers is differentiated by vertical dashed lines. Optical reflectivity threshold is shown as a horizontal line. In (c), note the decreased reflectivity registered in the GCL+IPL, in the INL, and in the ONL+IS.
A graph theory segmentation algorithm based on Li et al.12 and Garvin et al.13 was implemented to automatically identify eight retinal interfaces. The retinal intralayer segments identified are retinal nerve fiber layer (RNFL), ganglion cell layer and inner plexiform layer (GCL+IPL), INL, outer plexiform layer (OPL), outer nuclear layer and photoreceptor inner segments (ONL+IS), photoreceptor outer segments (OS), and retinal pigment epithelium with Bruch's membrane (RPE) (Fig. 2). All segmented examinations were reviewed by experienced graders (Santos T, et al. IOVS 2015;56:ARVO E-Abstract 5953). 
Figure 2
 
Image with labeled intralayer segmentation. Eight surfaces and corresponding intralayers labeled from (A) to (G). (A) Retinal nerve fiber layer (RNFL); (B) ganglion cell layer and inner plexiform layer (GCL+IPL); (C) inner nuclear layer (INL); (D) outer plexiform layer (OPL); (E) outer nuclear layer and photoreceptor inner segments (ONL+IS); (F) photoreceptor outer segments (OS); (G) retinal pigment epithelium and Bruch's membrane (RPE).
Figure 2
 
Image with labeled intralayer segmentation. Eight surfaces and corresponding intralayers labeled from (A) to (G). (A) Retinal nerve fiber layer (RNFL); (B) ganglion cell layer and inner plexiform layer (GCL+IPL); (C) inner nuclear layer (INL); (D) outer plexiform layer (OPL); (E) outer nuclear layer and photoreceptor inner segments (ONL+IS); (F) photoreceptor outer segments (OS); (G) retinal pigment epithelium and Bruch's membrane (RPE).
Sites of low optical reflectivity (LOR) are depicted as two-dimensional images of the different areas of the eye by assigning a single representative value for each A-scan, which represents the existence of reflectivity values falling below the predefined threshold. The white areas depicted in the LOR maps represent the locations of A-scans having reflectivity values below the predefined threshold, while black areas are above the threshold (Figs. 3, 4, 5). Extracellular fluid distribution in a given area of the retina can be measured by the LOR area ratio, which stands for the number of A-scans with reflectivity values below the threshold divided by the total number of A-scans within the considered area. 
Figure 3
 
OCT-Leakage LOR maps for the right eye of a healthy subject. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer. Note that the highest value of LOR is in the ONL+IS.
Figure 3
 
OCT-Leakage LOR maps for the right eye of a healthy subject. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer. Note that the highest value of LOR is in the ONL+IS.
Figure 4
 
OCT-Leakage LOR maps for the right eye of a male diabetic patient with subclinical macular edema. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer, showing increased LOR ratios reflecting increases in the retinal extracellular space in the GPL+IPL, INL, OPL, and ONL+IS and extending to the OS layer.
Figure 4
 
OCT-Leakage LOR maps for the right eye of a male diabetic patient with subclinical macular edema. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer, showing increased LOR ratios reflecting increases in the retinal extracellular space in the GPL+IPL, INL, OPL, and ONL+IS and extending to the OS layer.
Figure 5
 
Spectral-domain OCT LOR maps for the INL at the central subfield: 500-μm radius area centered at the fovea of (a) a healthy subject, (b) a NPDR patient; (c) a SME patient, and (d) a CME patient. Areas of white represent low optical reflectivity.
Figure 5
 
Spectral-domain OCT LOR maps for the INL at the central subfield: 500-μm radius area centered at the fovea of (a) a healthy subject, (b) a NPDR patient; (c) a SME patient, and (d) a CME patient. Areas of white represent low optical reflectivity.
Figure 6
 
Scatter plot and fitted values of % of increase of LOR area ratio and % of increase of INL thickness in central subfield.
Figure 6
 
Scatter plot and fitted values of % of increase of LOR area ratio and % of increase of INL thickness in central subfield.
Maps of LOR sites can therefore be obtained not only for the full retina scan but also layer by layer for en face images (Figs. 3, 4). 
To assess intraday reproducibility, a qualified technician acquired two consecutive scans of the same signal strength on 18 healthy eyes from subjects with age between 29 and 73 years (mean ± SD: 51.7 ± 16.7 years). The signal strength of the acquired data ranged from 7 to 10 (mean ± SD: 9.4 ± 1.1). The reproducibility for the LOR area ratio was found to be 0.97 with a 95% confidence interval of 0.92, 0.99. For the interoperator reproducibility, two qualified technicians acquired scans of the same signal strength on 17 healthy eyes from subjects with age between 29 and 47 years (mean ± SD: 39.1 ± 5.6 years). The reproducibility for the LOR area ratio was found to be 0.94 with a 95% confidence interval of 0.84, 0.98. 
Using the defined threshold, the values for the LOR ratios of the different retinal layers in the central subfield of the healthy eyes were RNFL: 0.17 ± 0.05; GCL+IPL: 0.42 ± 0.10; INL: 0.31 ± 0.08; OPL: 0.37 ± 0.10; ONL+IS: 0.96 ± 0.03; OS: 0.22 ± 0.09; RPE: 0.02 ± 0.01. In normal eyes, the highest value of extracellular space is located on the ONL+IS. 
Data Analysis
Categorical variables are summarized with frequencies and percentages, and numerical variables are summarized with mean and standard deviation (SD). 
Student's t-test was used to assess differences in retinal layer thickness and in LOR ratio between the healthy control group and the study groups (NPDR with normal RT, subclinical edema, and clinical edema). 
The Spearman correlation coefficient, r, was used to assess correlations between each layer's RT increase and the LOR area ratio increase. Since we performed multiple correlations, to protect from type I error, a Bonferroni correction was conducted. The new α will be the α-original (0.05) divided by the number of comparisons (7): 0.01. To determine if any of the seven correlations is statistically significant, the P value should be <0.01. The confidence interval determinations were similarly adjusted. 
Reproducibility of the LOR area ratio measurements was estimated using the intraclass correlation coefficient (ICC) determined on the basis of a 2-way mixed-effects model, absolute agreement. The ICC is commonly used to measure reliability, and a higher ICC indicates better reproducibility of the method. 
All statistical analyses were performed with Stata version 12.1 (StataCorp LP, College Station, TX, USA), and P values less than or equal to 0.007 were considered statistically significant results because the Bonferroni correction was used for multiple comparisons. 
Results
Baseline characteristics of the study eyes are presented in Table 1
Table 1
 
Baseline Characteristics of the Diabetic Eyes Included in This Analysis
Table 1
 
Baseline Characteristics of the Diabetic Eyes Included in This Analysis
We compared RT values in the central subfield from each of the different segmented retinal layers (RNFL, GCL+IPL, INL, OPL, ONL, IS/OS, RPE). Among the different groups of eyes examined (healthy controls, NPDR with normal RT, subclinical edema, and clinical edema), selective involvement of specific retinal layers is apparent (Table 2). 
Table 2
 
RT Values for the Central Subfield Obtained From the Different Segmented Retinal Layers (Layer Thickness, μm)
Table 2
 
RT Values for the Central Subfield Obtained From the Different Segmented Retinal Layers (Layer Thickness, μm)
The RNFL and GC+IPL show thinning in the eyes with NPDR without edema. Retinal thickness increases are apparent in NPDR eyes with SME and clinical edema in the GCL+IPL, INL, OPL and ONL+IS layers. 
Low optical reflectivity ratios show significant increases in the INL in SME and in the GCL+IPL and INL in CME (Table 3). 
Table 3
 
LOR Ratio Values for the Central Subfield Obtained From the Different Segmented Retinal Layers
Table 3
 
LOR Ratio Values for the Central Subfield Obtained From the Different Segmented Retinal Layers
The healthy control group was used to determine retinal layer thickness increases and LOR ratio increases in the study groups. 
In the central subfield, the INL shows the largest increases in RT. These increases are 49.9% in SME and 104.7% in CME. There is also a marked increase in the OPL in SME (29.2%) and in CME (47.3%), with RT increases identified in other retinal layers in clinical macular edema. The preferential involvement of the INL in RT increases is also present in the inner and outer ring areas (Table 4). 
Table 4
 
Comparison Between Percentage of SD-OCT LOR Area Ratio and Percentage of Change of Layer Thickness From Normative Values for SME and CME Eyes in the Central Subfield
Table 4
 
Comparison Between Percentage of SD-OCT LOR Area Ratio and Percentage of Change of Layer Thickness From Normative Values for SME and CME Eyes in the Central Subfield
Sites of SD-OCT LOR are identified in white, indicating locations of increased extracellular space (Fig. 5). 
Automated analysis of the retinal extracellular space shows a positive correlation with the observed increases in RT for SME and CME eyes (Table 4). 
This correlation between the percentage of increase of the LOR area ratio and the percentage of increase of layer thickness, in the central subfield, was particularly strong in the INL (Fig. 6). 
Significant positive correlations were also found in the inner and outer ring areas for eyes with increased RT in these areas. For the inner ring, a strong correlation was found in the ONL+IS in the nasal area (r = 0.72 (0.00, 0.95), P = 0.013), and a moderate correlation was present in the OPL in the temporal area (r = 0.53 (−0.15, 0.87), P = 0.041). 
In the outer ring, a strong correlation was found in the OPL in the inferior area (r = 0.63 (−0.17, 0.93), P = 0.039) and in the GCL+IPL in the temporal area (r = 0.76 (0.02, 0.96), P = 0.011). Very strong correlations were found in the superior area of the outer ring in RNFL, INL, and ONL+IS (r = 0.88 (0.22, 0.99), P = 0.004; r = 0.90 (0.31, 0.99), P = 0.002; r = 0.93 (0.47, 0.99), P = 0.001, respectively). 
Discussion
This study was a post hoc analysis, including data from only one participating center of a large observational study in eyes of patients with diabetes type 2 and mild NPDR (Early Treatment Diabetic Retinopathy Study [ETDRS] levels 20 and 35) and good visual acuity. In a previous report it was possible to show that the increase in RT in diabetic eyes with macular edema is predominantly located in the INL, extending to the neighboring retinal cell layers, suggesting that it may be due to extracellular fluid accumulation.3 
In the present report the data from one of the participating centers were examined using a novel algorithm to identify the location of the sites of LOR (OCT-Leakage). In this way, automated analysis and mapping of the retinal extracellular spaces (i.e., extracellular fluid) could be accomplished. 
A strong correlation was found between the increase in the retinal extracellular space and the increase in RT in the INL in both subclinical and clinical macular edema. This correlation was also observed in other retinal layers, validating the concept that increases in RT in the initial stages of diabetic retinal disease are mainly associated with increases in the retinal extracellular space. Furthermore, it demonstrates that the changes in the retinal extracellular space in diabetic macular edema occur in different amounts in different layers of the retina and are apparently largest in the INL. 
Previous attempts to identify sites of LOR in the OCT, with changes in the spaces between cells, and to correlate them with sites of BRB breakdown have demonstrated only limited utility.14 The new approach presented here for the first time permits the identification, location, and mapping of areas of fluid accumulation and their correlation with changes in RT. Other authors have previously focused on the optical reflectivity of cysts in different retinal diseases, demonstrating that the optical reflectivity could show heterogeneity of reflectivity, especially in DME.15 Their observations do not oppose our findings. The method proposed here identifies only the sites that have lower than normal optical reflectivity, suggesting that water-based fluid is present in these sites. It is possible that fluid with different densities is present in different eyes with shorter or longer evolution of disease. It is also likely that fluid with high optical reflectivity, such as blood, may occur in the retina, and is not detected by our method. 
In diabetic macular edema the extravasation of fluid is considered to be the result of vascular injury and possibly RPE dysfunction, leading to breakdown of the BRB and accumulation of excess extracellular fluid.16 
The method described here detects increases in fluid accumulation in eyes with increased RT due to subclinical and clinical macular edema. Protein-free fluid accumulation is expected to occur before protein leakage. It is, therefore, likely that OCT-Leakage may be at least as sensitive as fluorescein angiography. Importantly, unlike the situation with fluorescein angiography, the amount of fluid can be quantified. Our group is pursuing this line of research and looking at the specificity and sensitivity of automated analysis of the retinal extracellular fluid to identify and quantify the sites of leakage demonstrated by fluorescein angiography. 
Automated analysis of the retinal extracellular space appears, therefore, to be a useful tool to evaluate fluid accumulation in retinal edema and may offer a demonstration of the location and severity of the breakdown of the BRB that is responsible for the increases in abnormal extracellular fluid. 
The results presented here need to be replicated by other centers with larger patient cohorts. However, they open new perspectives for improving the detection, localization, quantification, and management of retinal edema. Quantification and mapping of the changes occurring in the retinal extracellular fluid allow noninvasive evaluation of alterations of the BRB. Such evaluations are particularly promising in following a specific eye's response to treatment. Using this methodology in conjunction with information obtained through other new, noninvasive imaging methodologies is expected to offer a wider spectrum of information than is currently available. The OCT analysis of diabetic macular edema proposed by Soliman, Sander, and Jorgensen17 can now be tested using OCT-Leakage with improved identification of the location of abnormal retinal fluid layer by layer. 
OCT-Leakage is able to identify the location of the major increases of the retinal extracellular space occurring in cases of macular edema and in the different layers of the retina. This technique is thus able to identify the retinal layers more affected in a specific eye. This information may offer new insights into the progression of and recovery from macular edema. 
Finally, location and quantification of retinal edema determined by our method of OCT-Leakage may complement OCT microangiography, thus offering the possibility of obtaining valuable information on breakdown of the BRB and extracellular fluid accumulation, together with improved visualization of capillary dropout and vascular morphology, using only noninvasive OCT-based methodologies. 
Acknowledgments
Disclosure: J. Cunha-Vaz, Alimera Sciences (C), Allergan (C), Bayer (C), Fovea Pharmaceuticals (C), GeneSignal (C), Novartis (C), OM Pharma (C), Pfizer (C), Roche (C), Zeiss (C), P; T. Santos, P; L. Ribeiro, None; D. Alves, None; I. Marques, None; M. Goldberg, Eyegate (C) 
References
Fong DS, Aiello LP, Ferris FL, Klein R. Diabetic retinopathy. Diabetes Care. 2004; 27: 2540–2553.
Aiello LP, Gardner TW, King GL, et al. Diabetic retinopathy. Diabetes Care. 1998; 21: 143–156.
Bandello F, Tejerina AN, Vujosevic S, et al. Retinal layer location of increased retinal thickness in eyes with subclinical and clinical macular edema in diabetes type 2. Ophthalmic Res. In press.
Cunha-Vaz J, Faria de Abreu JR, Campos AJ. Early breakdown of the blood-retinal barrier in diabetes. Br J Ophthalmol. 1975; 59: 649–656.
Cunha-Vaz JG, Goldberg MF, Vygantas C, Noth J. Early detection of retinal involvement in diabetes by vitreous fluorophotometry. Ophthalmology. 1979; 86: 264–275.
Ribeiro L, Bandello F, Tejerina AN, et al. Characterization of retinal disease progression in a 1-year longitudinal study of eyes with mild nonproliferative retinopathy in diabetes type 2. Invest Ophthalmol Vis Sci. 2015; 56: 5698–5705.
Chalam KV, Bressler SB, Edwards AR, et al. Retinal thickness in people with diabetes and minimal or no diabetic retinopathy: Heidelberg Spectralis optical coherence tomography. Invest Ophthalmol Vis Sci. 2012; 53: 8154–8161.
Bressler NM, Miller KM, Beck RW, et al. Observational study of subclinical diabetic macular edema. Eye (Lond). 2012; 26: 833–840.
Browning DJ, Fraser CM. The predictive value of patient and eye characteristics on the course of subclinical diabetic macular edema. Am J Ophthalmol. 2008; 145: 149–154.
Pires I, Santos AR, Nunes S, Lobo C, Cunha-Vaz J. Subclinical macular edema as a predictor of progression to clinically significant macular edema in type 2 diabetes. Ophthalmologica. 2013; 230: 201–206.
Friedman SM, Almukhtar TH, Baker CW, et al. Topical nepafenec in eyes with noncentral diabetic macular edema. Retina. 2015; 35: 944–956.
Li K, Wu X, Chen DZ, Sonka M. Optimal surface segmentation in volumetric images--a graph-theoretic approach. IEEE Trans Pattern Anal Mach Intell. 2006; 28: 119–134.
Garvin MK, Abramoff MD, Kardon R, Russell SR, Wu X, Sonka M. Intraretinal layer segmentation of macular optical coherence tomography images using optimal 3-D graph search. IEEE Trans Med Imaging. 2008; 27: 1495–1505.
Bernardes R, Santos T, Cunha-Vaz J. Evaluation of blood-retinal barrier function from Fourier domain high-definition optical coherence tomography. 2009: 316–319.
Horii T, Murakami T, Nishijima K, et al. Relationship between fluorescein pooling and optical coherence tomographic reflectivity of cystoid spaces in diabetic macular edema. Ophthalmology. 2012; 119: 1047–1055.
Miller JW, Adamis AP, Shima DT, et al. Vascular endothelial growth factor/vascular permeability factor is temporally and spatially correlated with ocular angiogenesis in a primate model. Am J Pathol. 1994; 145: 574–584.
Soliman W, Sander B, Jorgensen TM. Enhanced optical coherence patterns of diabetic macular oedema and their correlation with the pathophysiology. Acta Ophthalmol Scand. 2007; 85: 613–617.
Figure 1
 
Cirrus SD-OCT A-scan optical reflectivity profiles. (a) Full-length SD-OCT A-scan from a healthy subject. (b, c) Detail of SD-OCT A-scan in the retina for a healthy subject (bottom left) and for a NPDR patient with CME (bottom right). Spectral-domain OCT A-scan optical reflectivity from each of the segmented layers is differentiated by vertical dashed lines. Optical reflectivity threshold is shown as a horizontal line. In (c), note the decreased reflectivity registered in the GCL+IPL, in the INL, and in the ONL+IS.
Figure 1
 
Cirrus SD-OCT A-scan optical reflectivity profiles. (a) Full-length SD-OCT A-scan from a healthy subject. (b, c) Detail of SD-OCT A-scan in the retina for a healthy subject (bottom left) and for a NPDR patient with CME (bottom right). Spectral-domain OCT A-scan optical reflectivity from each of the segmented layers is differentiated by vertical dashed lines. Optical reflectivity threshold is shown as a horizontal line. In (c), note the decreased reflectivity registered in the GCL+IPL, in the INL, and in the ONL+IS.
Figure 2
 
Image with labeled intralayer segmentation. Eight surfaces and corresponding intralayers labeled from (A) to (G). (A) Retinal nerve fiber layer (RNFL); (B) ganglion cell layer and inner plexiform layer (GCL+IPL); (C) inner nuclear layer (INL); (D) outer plexiform layer (OPL); (E) outer nuclear layer and photoreceptor inner segments (ONL+IS); (F) photoreceptor outer segments (OS); (G) retinal pigment epithelium and Bruch's membrane (RPE).
Figure 2
 
Image with labeled intralayer segmentation. Eight surfaces and corresponding intralayers labeled from (A) to (G). (A) Retinal nerve fiber layer (RNFL); (B) ganglion cell layer and inner plexiform layer (GCL+IPL); (C) inner nuclear layer (INL); (D) outer plexiform layer (OPL); (E) outer nuclear layer and photoreceptor inner segments (ONL+IS); (F) photoreceptor outer segments (OS); (G) retinal pigment epithelium and Bruch's membrane (RPE).
Figure 3
 
OCT-Leakage LOR maps for the right eye of a healthy subject. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer. Note that the highest value of LOR is in the ONL+IS.
Figure 3
 
OCT-Leakage LOR maps for the right eye of a healthy subject. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer. Note that the highest value of LOR is in the ONL+IS.
Figure 4
 
OCT-Leakage LOR maps for the right eye of a male diabetic patient with subclinical macular edema. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer, showing increased LOR ratios reflecting increases in the retinal extracellular space in the GPL+IPL, INL, OPL, and ONL+IS and extending to the OS layer.
Figure 4
 
OCT-Leakage LOR maps for the right eye of a male diabetic patient with subclinical macular edema. (a) Full retina scan LOR map; (b) ETDRS grid map; (c) B-scan centered on the fovea; (d–i) LOR maps layer by layer, showing increased LOR ratios reflecting increases in the retinal extracellular space in the GPL+IPL, INL, OPL, and ONL+IS and extending to the OS layer.
Figure 5
 
Spectral-domain OCT LOR maps for the INL at the central subfield: 500-μm radius area centered at the fovea of (a) a healthy subject, (b) a NPDR patient; (c) a SME patient, and (d) a CME patient. Areas of white represent low optical reflectivity.
Figure 5
 
Spectral-domain OCT LOR maps for the INL at the central subfield: 500-μm radius area centered at the fovea of (a) a healthy subject, (b) a NPDR patient; (c) a SME patient, and (d) a CME patient. Areas of white represent low optical reflectivity.
Figure 6
 
Scatter plot and fitted values of % of increase of LOR area ratio and % of increase of INL thickness in central subfield.
Figure 6
 
Scatter plot and fitted values of % of increase of LOR area ratio and % of increase of INL thickness in central subfield.
Table 1
 
Baseline Characteristics of the Diabetic Eyes Included in This Analysis
Table 1
 
Baseline Characteristics of the Diabetic Eyes Included in This Analysis
Table 2
 
RT Values for the Central Subfield Obtained From the Different Segmented Retinal Layers (Layer Thickness, μm)
Table 2
 
RT Values for the Central Subfield Obtained From the Different Segmented Retinal Layers (Layer Thickness, μm)
Table 3
 
LOR Ratio Values for the Central Subfield Obtained From the Different Segmented Retinal Layers
Table 3
 
LOR Ratio Values for the Central Subfield Obtained From the Different Segmented Retinal Layers
Table 4
 
Comparison Between Percentage of SD-OCT LOR Area Ratio and Percentage of Change of Layer Thickness From Normative Values for SME and CME Eyes in the Central Subfield
Table 4
 
Comparison Between Percentage of SD-OCT LOR Area Ratio and Percentage of Change of Layer Thickness From Normative Values for SME and CME Eyes in the Central Subfield
×
×

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.

×