[Skip to Content]
[Skip to Content Landing]
Figure.
Optical Coherence Tomography Angiograms of the Superficial Vascular Complex From Healthy Participants and Participants With Diabetes
Optical Coherence Tomography Angiograms of the Superficial Vascular Complex From Healthy Participants and Participants With Diabetes

Projection-resolved optical coherence tomography angiograms with automatically detected nonperfusion areas in blue. DR indicates diabetic retinopathy; NPDR, nonproliferative DR; PDR, proliferative DR.

Table 1.  
Participant Characteristics
Participant Characteristics
Table 2.  
Avascular Areas and Commercial Vessel Density Measurements
Avascular Areas and Commercial Vessel Density Measurements
Table 3.  
Spearman Rank-order Correlation Coefficient
Spearman Rank-order Correlation Coefficient
Table 4.  
Diagnostic Accuracy of Nonperfusion Biomarkers and Commercial Vessel Density Measurements
Diagnostic Accuracy of Nonperfusion Biomarkers and Commercial Vessel Density Measurements
1.
Klein  BE.  Overview of epidemiologic studies of diabetic retinopathy.  Ophthalmic Epidemiol. 2007;14(4):179-183.PubMedGoogle ScholarCrossref
2.
Jia  Y, Bailey  ST, Hwang  TS,  et al.  Quantitative optical coherence tomography angiography of vascular abnormalities in the living human eye.  Proc Natl Acad Sci U S A. 2015;112(18):E2395-E2402.PubMedGoogle ScholarCrossref
3.
Hwang  TS, Gao  SS, Liu  L,  et al.  Automated quantification of capillary nonperfusion using optical coherence tomography angiography in diabetic retinopathy.  JAMA Ophthalmol. 2016;134(4):367-373.PubMedGoogle ScholarCrossref
4.
Hagag  AM, Gao  SS, Jia  Y, Huang  D.  Optical coherence tomography angiography: technical principles and clinical applications in ophthalmology.  Taiwan J Ophthalmol. 2017;7(3):115-129.PubMedGoogle ScholarCrossref
5.
Campbell  JP, Zhang  M, Hwang  TS,  et al.  Detailed vascular anatomy of the human retina by projection-resolved optical coherence tomography angiography.  Sci Rep. 2017;7:42201.PubMedGoogle ScholarCrossref
6.
Hwang  TS, Zhang  M, Bhavsar  K,  et al.  Visualization of 3 distinct retinal plexuses by projection-resolved optical coherence tomography angiography in diabetic retinopathy.  JAMA Ophthalmol. 2016;134(12):1411-1419.PubMedGoogle ScholarCrossref
7.
Spaide  RF, Fujimoto  JG, Waheed  NK.  Image artifacts in optical coherence tomography angiography.  Retina. 2015;35(11):2163-2180.PubMedGoogle ScholarCrossref
8.
Snodderly  DM, Weinhaus  RS, Choi  JC.  Neural-vascular relationships in central retina of macaque monkeys (Macaca fascicularis).  J Neurosci. 1992;12(4):1169-1193.PubMedGoogle ScholarCrossref
9.
Zhang  M, Hwang  TS, Campbell  JP,  et al.  Projection-resolved optical coherence tomographic angiography.  Biomed Opt Express. 2016;7(3):816-828.PubMedGoogle ScholarCrossref
10.
Zhang  M, Wang  J, Pechauer  AD,  et al.  Advanced image processing for optical coherence tomographic angiography of macular diseases.  Biomed Opt Express. 2015;6(12):4661-4675.PubMedGoogle ScholarCrossref
11.
Zhang  M, Hwang  TS, Dongye  C, Wilson  DJ, Huang  D, Jia  Y.  Automated quantification of nonperfusion in three retinal plexuses using projection-resolved optical coherence tomography angiography in diabetic retinopathy.  Invest Ophthalmol Vis Sci. 2016;57(13):5101-5106.PubMedGoogle ScholarCrossref
12.
Gao  SS, Jia  Y, Liu  L,  et al.  Compensation for reflectance variation in vessel density quantification by optical coherence tomography angiography.  Invest Ophthalmol Vis Sci. 2016;57(10):4485-4492.PubMedGoogle ScholarCrossref
13.
Kraus  MF, Liu  JJ, Schottenhamml  J,  et al.  Quantitative 3D-OCT motion correction with tilt and illumination correction, robust similarity measure and regularization.  Biomed Opt Express. 2014;5(8):2591-2613.PubMedGoogle ScholarCrossref
14.
Jia  Y, Tan  O, Tokayer  J,  et al.  Split-spectrum amplitude-decorrelation angiography with optical coherence tomography.  Opt Express. 2012;20(4):4710-4725.PubMedGoogle ScholarCrossref
15.
Holm  S.  A simple sequentially rejective multiple test procedure.  Scand J Stat. 1979;6(2):65-70.Google Scholar
16.
Simonett  JM, Scarinci  F, Picconi  F,  et al.  Early microvascular retinal changes in optical coherence tomography angiography in patients with type 1 diabetes mellitus.  Acta Ophthalmol. 2017;95(8):e751-e755.PubMedGoogle ScholarCrossref
17.
Carnevali  A, Sacconi  R, Corbelli  E,  et al.  Optical coherence tomography angiography analysis of retinal vascular plexuses and choriocapillaris in patients with type 1 diabetes without diabetic retinopathy.  Acta Diabetol. 2017;54(7):695-702.PubMedGoogle ScholarCrossref
18.
Dimitrova  G, Chihara  E, Takahashi  H, Amano  H, Okazaki  K.  Quantitative retinal optical coherence tomography angiography in patients with diabetes without diabetic retinopathy.  Invest Ophthalmol Vis Sci. 2017;58(1):190-196.PubMedGoogle ScholarCrossref
19.
Samara  WA, Shahlaee  A, Adam  MK,  et al.  Quantification of diabetic macular ischemia using optical coherence tomography angiography and its relationship with visual acuity.  Ophthalmology. 2017;124(2):235-244.PubMedGoogle ScholarCrossref
20.
Pedinielli  A, Bonnin  S, Sanharawi  ME,  et al.  Three different optical coherence tomography angiography measurement methods for assessing capillary density changes in diabetic retinopathy.  Ophthalmic Surg Lasers Imaging Retina. 2017;48(5):378-384.PubMedGoogle ScholarCrossref
21.
Ghasemi Falavarjani  K, Iafe  NA, Hubschman  JP, Tsui  I, Sadda  SR, Sarraf  D.  Optical coherence tomography angiography analysis of the foveal avascular zone and macular vessel density after anti-VEGF therapy in eyes with diabetic macular edema and retinal vein occlusion.  Invest Ophthalmol Vis Sci. 2017;58(1):30-34.PubMedGoogle ScholarCrossref
22.
Karst  SG, Deak  GG, Gerendas  BS,  et al.  Association of changes in macular perfusion with ranibizumab treatment for diabetic macular edema: a subanalysis of the RESTORE (extension) study.  JAMA Ophthalmol. 2018;136(4):315-321.PubMedGoogle ScholarCrossref
23.
Agemy  SA, Scripsema  NK, Shah  CM,  et al.  Retinal vascular perfusion density mapping using optical coherence tomography angiography in normals and diabetic retinopathy patients.  Retina. 2015;35(11):2353-2363.PubMedGoogle ScholarCrossref
24.
Zahid  S, Dolz-Marco  R, Freund  KB,  et al.  Fractal dimensional analysis of optical coherence tomography angiography in eyes with diabetic retinopathy.  Invest Ophthalmol Vis Sci. 2016;57(11):4940-4947.PubMedGoogle ScholarCrossref
Views 2,424
Citations 0
Original Investigation
June 14, 2018

Automated Quantification of Nonperfusion Areas in 3 Vascular Plexuses With Optical Coherence Tomography Angiography in Eyes of Patients With Diabetes

Author Affiliations
  • 1Casey Eye Institute, Oregon Health and Science University, Portland
  • 2Optovue Inc, Fremont, California
  • 3Georgetown University School of Medicine, Washington, DC
JAMA Ophthalmol. Published online June 14, 2018. doi:10.1001/jamaophthalmol.2018.2257
Key Points

Question  How useful are automatically quantified nonperfusion parameters from projection-resolved optical coherence tomography angiography in patients with diabetes?

Findings  In this cross-sectional study of 71 patients with diabetes and 39 healthy control individuals, avascular areas detected from projection-resolved optical coherence tomography angiography retinal plexuses distinguished between control individuals and various levels of diabetic retinopathy severity. These biomarkers were less dependent on age and signal strength and more sensitive than some commercially available vessel density measurements.

Meaning  Automatically quantified avascular areas from projection-resolved optical coherence tomography angiography of individual plexuses in patients with diabetes may offer a more reliable, sensitive method to evaluate diabetic vascular abnormalities than previous methods.

Abstract

Importance  Diabetic retinopathy (DR) is a leading cause of vision loss that is managed primarily through qualitative clinical examination of the retina. Optical coherence tomography angiography (OCTA) may offer an objective and quantitative method of evaluating DR.

Objective  To quantify capillary nonperfusion in 3 vascular plexuses in the macula of eyes patients with diabetes of various retinopathy severity using projection-resolved OCTA (PR-OCTA).

Design, Setting, and Participants  Cross-sectional study at a tertiary academic center comprising 1 eye each from healthy control individuals and patients with diabetes at different severity stages of retinopathy. Data were acquired and analyzed between January 2015 and December 2017.

Main Outcomes and Measures  Foveal avascular zone area, extrafoveal avascular area (EAA), and the sensitivity of detecting levels of retinopathy.

Results  The study included 39 control individuals (20 women [51%]; mean [SD] age, 43.41 [19.37] years); 16 patients with diabetes without retinopathy (8 women [50%]; mean [SD] age, 56.50 [12.43] years); 23 patients with mild to moderate nonproliferative DR (18 women [78%] ; mean [SD] age, 62.48 [10.55] years); and 32 patients with severe nonproliferative DR or proliferative DR (12 women [38%]; mean age, 53.41 [14.05] years). Mean (SD) foveal avascular zone area was 0.203 (0.103) mm2 for control individuals, 0.192 (0.084) mm2 for patients with diabetes without retinopathy, 0.243 [0.079] mm2 for mild to moderate nonproliferative DR, and 0.359 (0.275) mm2 for severe nonproliferative DR or proliferative DR. Mean (SD) EAA in whole inner retinal slab in these groups, respectively, were 0.020 (0.031) mm2, 0.034 (0.047) mm2, 0.038 (0.040) mm2, and 0.237 (0.235) mm2. The mean (SD) sum of EAA from 3 segmented plexuses in each of the respective groups were 0.103 (0.169) mm2, 0.213 (0.242) mm2, 0.451 (0.243) mm2, and 1.325 (1.140) mm2. With specificity fixed at 95%, using EAA in inner retinal slab, the sensitivity of detecting patients with diabetes from healthy control individuals was 28% (95% CI, 18%-40%), 31% for patients with DR (95% CI, 19%-45%), and 47% for patients with severe DR (95% CI, 29%-65%) from whole inner retinal EAA. With the sum of EAA from 3 individual plexuses, the sensitivities were 69% (95% CI, 57%-80%), 82% (95% CI, 70%-91%), and 97% (95% CI, 85%-100%), respectively. Avascular areas were not associated with signal strength index. The commercial vessel density from the 2-plexus scheme distinguished the groups with lower sensitivity and were dependent on SSI.

Conclusions and Relevance  Automatically quantified avascular areas from a 3-layer segmentation scheme using PR-OCTA distinguished levels of retinopathy with a greater sensitivity than avascular areas from unsegmented inner retinal slab or measurements from a commercially available vessel density in 2-layer scheme. Additional studies are needed to investigate the applicability of nonperfusion parameters in clinical settings.

Introduction

Diabetic retinopathy (DR) is a leading cause of vision loss in the United States.1 Prevention of vision loss related to DR depends on identifying qualitative features from clinical examination, fundus photographs, or fluorescein angiography. Optical coherence tomography angiography (OCTA), a noninvasive modality capable of capillary resolution imaging of retinal vasculature,2 is better suited for quantifying vascular changes related to diabetic retinopathy than conventional imaging and has opened the possibility of a quantitative biomarker to characterize DR.3

Unlike conventional modalities, OCTA is also able to image the retinal vasculature in 3 dimensions.4 The retinal vasculature is composed of 3 distinct layers in the macula: superficial vascular complex (SVC), intermediate capillary plexus (ICP), and deep capillary plexus (DCP).5 In DR, these layers can show incongruent areas of capillary nonperfusion that are not visible in whole inner retinal slabs,6 suggesting that an examination of the individual plexuses may reveal more vascular abnormalities and possibly increase the diagnostic accuracy of the biomarkers based on the angiograms.

However, projection artifacts in OCTA can confound the interpretation of the deeper plexus angiograms. The available commercial systems present the retinal vasculature in 2 plexuses: superficial and deep. They place the boundaries to include most of the ICP with the superficial layer, which makes the projection of the superficial vessels in the deep plexus less visible.7,8 Our group has developed an algorithm to reduce projection artifacts without moving the segmentation boundaries, and present the 3 macular plexuses distinctly according to known histologic studies.9 Using this algorithm and modern image processing techniques,10 we have previously shown that extrafoveal avascular areas (EAAs) from projection-resolved, segmented retinal angiograms perform better than unsegmented angiograms in distinguishing mild nonproliferative diabetic retinopathy (NPDR) eyes from control eyes.11 Avascular areas may have less age and signal strength dependence compared with vessel density (VD) measurements.12 In this study, we test how these OCTA-derived biomarkers perform across a wider spectrum of disease in a larger number of patients with diabetes and compare them with commercial VD measurements.

Methods

The study team recruited healthy participants and participants with diabetes from Oregon Health and Science University and obtained informed written consent under a protocol in accordance with the Declaration of Helsinki and compliant with the Health Insurance Portability and Accountability Act of 1996. The institutional review board of Oregon Health and Science University approved the protocol. Each participant underwent a full clinical examination including an Early Treatment of Diabetic Retinopathy Study (ETDRS) protocol visual acuity. The ETDRS 7-field color photographs were obtained from all participants with diabetes. Eyes with other significant ocular pathology (including hypertensive retinopathy, retinal vascular occlusions, and uveitis) were excluded from the study. Other exclusion criteria included history of major intraocular surgery (including vitrectomy, cataract extraction, scleral buckle, and glaucoma surgery) within prior 4 months and poor-quality scans (signal strength index [SSI] less than 55 or significant motion artifacts). Patients with history of laser photocoagulation or intravitreal injections were not excluded. The clinician retinopathy severity grading guided the recruitment goals, but a masked grading of the color photographs assigned eyes into no retinopathy, mild, moderate, and severe NPDR and proliferative DR groups for final analysis. A commercial 70-kHz spectral-domain OCT system (RTVue, Optovue) obtained two 3 × 3 mm scans, with a depth of 2 mm centered at the fovea. The optical resolution in retina is 15 μm and 5 μm on transverse and axial directions, respectively. The digital sampling interval is 15 × 15 × 3 μm3/voxel. Two registered scans, 1 x-fast and 1 y-fast, were combined to form a single volume to reduce motion artifacts.13 When 2 eyes were eligible for the study, the eye with better-quality scans was selected.

A previous publication details the image processing steps used in this study.11 Briefly, a commercial version of split-spectrum amplitude decorrelation angiography algorithm detected blood flow.14 Projection-resolved (PR) OCTA algorithm then suppressed the projection artifacts by comparing the normalized flow signal strength based on the reflectance signal with the more superficial signal in the same transverse location.9 A semiautomated algorithm based on directional graph search segmented the volumes into SVC, ICP, and DCP. The SVC was defined as the inner 80% of ganglion cell complex, which includes the nerve fiber layer, the ganglion cell layer, and the inner plexiform layer with a minimum thickness of 61 μm. The ICP was defined as the outer 20% of the GCC plus the inner 50% of the inner nuclear layer. The DCP was defined as remaining slab internal to the OPL with the minimum thickness of 37 μm. The mix of proportional slab definition with a minimum thickness is to account for the thinning and merging of the layers at the foveal pit8 and to avoid arbitrary separation of the vessels into plexuses that are merging.5 Avascular areas were identified on the resulting PR-OCTA slabs and the whole inner retinal slab, which included all layers internal to OPL based on their vessel-distance map generated by Euclidean distance transform algorithm.11

We calculated the foveal avascular zone (FAZ) area from the whole inner retinal slab and EAA, which excludes the central 1-mm diameter circle, in the inner retinal slab and each of the individual 3 plexus angiograms. We excluded the central area a priori to reduce the effect of normal variation in the FAZ size that may regrade the diagnostic accuracy, especially in a 3 × 3-mm angiogram.11 The sum of AAs from individual plexuses were also calculated.

The eyes were then assigned into 4 predetermined groups: healthy control participants, patients with diabetes without retinopathy, patients with mild to moderate NPDR, and patients with severe NPDR and PDR based on fundus photograph grading. The Mann-Whitney U test evaluated the difference of the AAs between the groups. Spearman rank-order correlation analysis tested the correlation between the AAs and the increasing level of retinopathy severity and hemoglobin A1c level. P values were adjusted for multiple comparisons using the Holm-Bonferroni method.15 The P value level of significance was .05, and all P values were 2-sided. With the specificity fixed at 95%, the sensitivity of detecting patients with diabetes from healthy control individuals, patients with mild to moderate NPDR from those without retinopathy, and patients with severe NPDR or PDR from patients with less severe retinopathy were calculated for EAA of the inner retinal slab, each of the 3 individual plexuses, and the sum of the 3 plexuses. The diagnostic accuracy was further evaluated with area under the receiver operating characteristic curve between each of the severity groups. Pooled standard deviation verified the repeatability of these measurements by testing values from images obtained in the same visit but different sitting at the machine.

Finally, we tested the correlation between the AAs and the SSI and the patient’s age using Spearman correlation analysis. These parameters were compared with the VD measurements produced by the default commercial software on the machine. The commercial software used a 2-layer scheme to divide the retinal vasculature into superficial and deep plexuses. The superficial plexus is defined as 15 μm below the ILM to 15 μm below inner plexiform layer, which includes the superficial portion of the inner nuclear layer and thus the ICP. The deep plexus is defined as 15 to 70 μm below the inner plexiform layer, which includes the deep capillary plexus and dips into the Henle fiber layer and outer nuclear layer centrally in the foveal pit. The commercial system did not have an algorithm to reduce the projection artifacts.

Results

The study included 1 eye each from 39 healthy control individuals, 16 patients with diabetes without retinopathy, 23 with mild to moderate NPDR, and 32 with more severe retinopathy. Table 1 summarizes the patient characteristics. Table 2 presents the automatically quantified AAs and commercial VD measurements, as well as the statistical differences between groups. Tables 3 and 4 present the correlation analyses and diagnostic accuracy of the nonperfuison metrics.

Mean (SD) foveal avascular zone area was 0.203 (0.103) mm2 for control individuals, 0.192 (0.084) mm2 for patients with diabetes without retinopathy, 0.243 [0.079] mm2 for mild to moderate NPDR, and 0.359 (0.275) mm2 for severe NPDR or PDR. Mean (SD) EAA in the whole inner retinal slab in these groups, respectively, were 0.020 (0.031) mm2, 0.034 (0.047) mm2, 0.038 (0.040) mm2, and 0.237 (0.235) mm2. The mean (SD) sum of EAA from 3 segmented plexuses in each of the respective groups were 0.103 (0.169) mm2, 0.213 (0.242) mm2, 0.451 (0.243) mm2, and 1.325 (1.140) mm2. With specificity fixed at 95%, using EAA in inner retinal slab, the sensitivity of detecting patients with diabetes from healthy control individuals was 28% (95% CI, 18%-40%), 31% for patients with DR (95% CI, 19%-45%), and 47% for patients with severe DR (95% CI, 29%-65%) from whole inner retinal EAA. With the sum of EAA from 3 individual plexuses, the sensitivities were 69% (95% CI, 57%-80%), 82% (95% CI, 70%-91%), and 97% (95% CI, 85%-100%), respectively.

In general, AAs were greater and VDs were less in eyes with more severe DR (Figure; eFigure in the Supplement). All parameters distinguished the severe NPDR to PDR group from control group with statistical significance. The FAZ size correlated significantly with increasing retinopathy severity (Tables 2 and 3) but did not distinguish between control, patients with diabetes without retinopathy, or the mild to moderate NPDR group. The EAA from SVC angiogram and the sum of EAA from the 3 plexuses distinguished with statistical significance between all groups. The commercial VD distinguished between all groups with statistical significance except between patients with diabetes without retinopathy and mild to moderate NPDR groups.

With the specificity fixed at 95%, EAA from the SVC was the most sensitive biomarker to detect all levels of diabetic retinopathy (70%, 84%, and 100%), with the sum of EAA from all 3 plexuses performing similarly (69%, 82%, and 97%). They were more sensitive than the commercial VD measurements for either superficial (36%, 43%, 66%) or deep (61%, 70%, 94%) plexuses, especially for the lower level of retinopathy (Table 4).

All vessel parameters were significantly correlated with ETDRS vision scores and retinopathy severity groups (Table 3). In the control group, the projection-resolved AA measurements had minimal association with age and no significant association with SSI, whereas the commercial VD measurements were strongly associated with age and SSI (Table 3). There was a strong association between SSI and retinopathy severity group with Spearman ρ of −0.625 (P < .001).

The area under the receiving operating curve evaluating the diagnostic accuracy between each of the groups was best for SVC EAA and the sum of EAA of all 3 plexuses (Table 4). The commercial VD measurements were nearly as good as these EAA measurements and performed better than ICP or DCP EAA measurements. The repeatability of AA, as measured by pooled SD, were 0.018 mm2 for FAZ and 0.026 mm2, 0.057 mm2, 0.021 mm2, and 0.048 mm2 for EAA of the inner retina, SVC, ICP, and DCP, respectively.

Discussion

We have demonstrated that projection-resolved OCTA derived AA with a 3-layer segmentation scheme that correspond to the known anatomic boundaries of retinal vasculature can distinguish between control individuals, patients with diabetes without retinopathy, patients with mild to moderate NPDR, and severe NPDR to PDR groups. The sensitivity of detecting retinopathy groups was greater when the vascular plexuses were examined individually, and the AA from the superficial vascular complex was the most sensitive. All AA measurements had good repeatability as measured by pooled standard deviation. This demonstrates the potential of OCTA to objectively and reliably evaluate DR.

Our automated algorithm for detecting AA was designed to mimic the human grader’s detection of nonperfusion areas, identifying only the contiguous and pathologic areas of capillary dropout while ignoring increased intercapillary spaces. By using the mean reflectance in the ganglion cell layer and inner plexiform layer to compensate for variation in signal strength within an image, we improved the accuracy of classifying vessel vs static tissue on en face OCTA. The use of vessel-distance map on vascular binary image avoided exaggerated contribution of large vessels in determining VD abnormality and eliminates the false-positive detection along large arterioles. We have previously demonstrated high repeatability and high detection accuracy with this approach.10 Based on the observation that the VD measurements are dependent on SSI and age, we hypothesized that our approach would detect capillary nonperfusion caused by disease but would be less likely to detect artifactitious changes in perfusion index caused by image quality or normal changes in capillary density with aging. We have demonstrated that the AA measurements are independent from SSI and correlate minimally with age, potentially improving their value in detecting diabetic retinopathy.

Another unique feature of this study is the projection-resolved algorithm and segmentation of the macular vasculature into 3 plexuses respecting the true known boundaries. We hypothesized that (1) projection artifacts would make detection of nonperfusion more difficult in the deeper plexuses, (2) the inclusion of the ICP into the superficial plexus as done in many commercial segmentation schemes would decrease the sensitivity of nonperfusion detection, and (3) even if the deep plexus included the ICP and DCP together without projection artifacts, an en face analysis of this layer would only reveal congruous AA between the 2 layers. We expected that our approach would lead to more sensitive and specific detection of microvasculopathy, and in turn, better correlation with clinical grading of diabetic retinopathy.

The results demonstrate that 3-layer segmentation with projection resolution indeed detects more nonperfusion than unsegmented inner retinal angiogram and is better able to distinguish between retinopathy severity groups. However, the diagnostic accuracy as measured by area under the receiving operator characteristic curve did not show a clearly superior performance compared with the 2-layer commercial VD measurements in this study. This may be owing to the association between the SSI and the retinopathy groups in this study (Spearman ρ, −0.568; P < .001). The SSI may be associated with retinopathy severity for various factors that are not associated with microangiopathy of diabetes. Because these factors are likely not be consistently present in different populations, an SSI-dependent nonperfusion parameter may give less reproducible study results. Furthermore, all scans included in the study had high SSI (mean [SD], 71.35 [7.45]; range, 59-88) and it is possible that in the situations with more variable scan qualities, VD may be less useful in distinguishing between the groups.

Other groups have reported on the differences in nonperfusion parameters from OCTA and patients with diabetes. Simonett et al,16 in comparing patients with type 1 diabetes without retinopathy with healthy control eyes, reported a significant difference in the deep plexus parafoveal VD but not in the superficial plexus.16 Carnevali et al17 also compared VD in patients with type 1 diabetes without retinopathy with healthy control individuals and similarly found a significant difference in the deep but not in the superficial plexus.17 In contrast, Dimitrova et al,18 using similar methods, reported a significantly reduced VDs in both superficial and deep plexuses in diabetic patients without retinopathy compared to controls.

Samara et al19 examined control, mild NPDR, moderate to severe NPDR, and PDR eyes with VD measurements and found that deep plexus VD measurements distinguished between the diabetic retinopathy groups better than superficial plexus. Both plexuses showed significant difference between the control group and the retinopathy groups of all levels, and none of the measurements had significant difference between PDR and moderate-to-severe NPDR groups.19 In contrast, Pedinielli et al20 did not show a statistically significant difference between patients with diabetes without retinopathy, patients with mild NPDR, patients with moderate NPDR, or patients with severe NPDR vs the control eyes in deep or superficial plexus VD.

These studies had different patient populations and different analysis methods. Their divergent conclusions and the lack of reproducibility of some of the findings with respect to the differences in superficial and deep plexus VD highlight the challenges of quantification using OCTA. Many steps are required to process the images, and there are numerous potential confounders. Image acquisition and quality, segmentation of the layers, projection artifacts, strategies for motion artifact reduction, and image processing can all influence the results. In particular, the role of image quality as measured by SSI may be particularly important. The disagreements in the conclusions from various studies may be because the SSI and retinopathy severity are variably related quantities. The exclusion of scans with low SSI does not appear to eliminate the influence of SSI on VD measurements.

Limitations and Strengths

Limitations of this study include the cross-sectional design and a modest number of participants. Furthermore, hypertension, laser therapy, and anti–vascular endothelial growth factorare confounders that might influence retinal vasculature. Participants with any ocular pathology that might interfere with diabetic vasculopathy, including hypertensive retinopathy, were excluded from the the study. In addition, 2017 and 2018 studies by Ghasemi Falavarjani et al21 and Karst et al22 reported that no significant changes were observed in retinal perfusion after anti–vascular endothelial growth factor treatment in patients with DR. Another limitation of this study is that we did not fully explore other methods of vascular quantification methods, such as vessel length density, vascular perfusion density mapping, automated perfusion density, and fractal dimension analysis, to compare with our PR-OCTA AA.23,24 The strengths of this study were the consistently high-quality images, advanced image processing algorithms, rigorous grading of the retinopathy levels by masked graders from 7 field ETDRS photographs, and the visual acuity assessment using ETDRS protocol vision.

Conclusions

Automatically quantified AA detected in 3-plexus segmentation using PR-OCTA can detect differences between control individuals, patients with diabetes without retinopathy, patients with mild to moderate NPDR, and patients with severe NPDR to PDR. With PR-OCTA, 3 macular plexuses were examined independently according to the known anatomic boundaries. These biomarkers were not dependent on SSI and may be less likely to be confounded by other covariate factors that are not the result of microvasculopathy of diabetes.

Back to top
Article Information

Corresponding Author: Yali Jia, PhD, Casey Eye Institute, Oregon Health & Science University, 3375 SW Terwilliger Blvd, Portland, OR 97239 (jiaya@ohsu.edu).

Accepted for Publication: April 16, 2018.

Published Online: June 14, 2018. doi:10.1001/jamaophthalmol.2018.2257

Author Contributions: Dr Jia had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Hwang, Hagag, Wilson, David, Jia.

Acquisition, analysis, or interpretation of data: Hwang, Hagag, Wang, Zhang, Smith, Wilson, Jia.

Drafting of the manuscript: Hwang.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Hwang, Hagag, Wang.

Obtained funding: Hwang, Jia.

Administrative, technical, or material support: Wang, Zhang, Smith, Wilson, David, Jia.

Supervision: Hwang, Wilson, David, Jia.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Drs Jia and Huang have a significant financial interest in Optovue Inc, a company that may have a commercial interest in the results of this research and technology. Dr Zhang is employed by Optovue Inc. These potential conflicts of interest have been reviewed and managed by Oregon Health & Science University. No other disclosures were reported.

Funding/Support: This work was supported by grants R01 EY027833, DP3 DK104397, R01 EY024544, and P30 EY010572 from the National Institutes of Health, an unrestricted departmental funding grant and William & Mary Greve Special Scholar Award from Research to Prevent Blindness, New York, New York.

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Previous Presentation: This paper was presented at the Association for Research in Vision and Ophthalmology Annual Meeting; May 3, 2017; Baltimore, Maryland.

References
1.
Klein  BE.  Overview of epidemiologic studies of diabetic retinopathy.  Ophthalmic Epidemiol. 2007;14(4):179-183.PubMedGoogle ScholarCrossref
2.
Jia  Y, Bailey  ST, Hwang  TS,  et al.  Quantitative optical coherence tomography angiography of vascular abnormalities in the living human eye.  Proc Natl Acad Sci U S A. 2015;112(18):E2395-E2402.PubMedGoogle ScholarCrossref
3.
Hwang  TS, Gao  SS, Liu  L,  et al.  Automated quantification of capillary nonperfusion using optical coherence tomography angiography in diabetic retinopathy.  JAMA Ophthalmol. 2016;134(4):367-373.PubMedGoogle ScholarCrossref
4.
Hagag  AM, Gao  SS, Jia  Y, Huang  D.  Optical coherence tomography angiography: technical principles and clinical applications in ophthalmology.  Taiwan J Ophthalmol. 2017;7(3):115-129.PubMedGoogle ScholarCrossref
5.
Campbell  JP, Zhang  M, Hwang  TS,  et al.  Detailed vascular anatomy of the human retina by projection-resolved optical coherence tomography angiography.  Sci Rep. 2017;7:42201.PubMedGoogle ScholarCrossref
6.
Hwang  TS, Zhang  M, Bhavsar  K,  et al.  Visualization of 3 distinct retinal plexuses by projection-resolved optical coherence tomography angiography in diabetic retinopathy.  JAMA Ophthalmol. 2016;134(12):1411-1419.PubMedGoogle ScholarCrossref
7.
Spaide  RF, Fujimoto  JG, Waheed  NK.  Image artifacts in optical coherence tomography angiography.  Retina. 2015;35(11):2163-2180.PubMedGoogle ScholarCrossref
8.
Snodderly  DM, Weinhaus  RS, Choi  JC.  Neural-vascular relationships in central retina of macaque monkeys (Macaca fascicularis).  J Neurosci. 1992;12(4):1169-1193.PubMedGoogle ScholarCrossref
9.
Zhang  M, Hwang  TS, Campbell  JP,  et al.  Projection-resolved optical coherence tomographic angiography.  Biomed Opt Express. 2016;7(3):816-828.PubMedGoogle ScholarCrossref
10.
Zhang  M, Wang  J, Pechauer  AD,  et al.  Advanced image processing for optical coherence tomographic angiography of macular diseases.  Biomed Opt Express. 2015;6(12):4661-4675.PubMedGoogle ScholarCrossref
11.
Zhang  M, Hwang  TS, Dongye  C, Wilson  DJ, Huang  D, Jia  Y.  Automated quantification of nonperfusion in three retinal plexuses using projection-resolved optical coherence tomography angiography in diabetic retinopathy.  Invest Ophthalmol Vis Sci. 2016;57(13):5101-5106.PubMedGoogle ScholarCrossref
12.
Gao  SS, Jia  Y, Liu  L,  et al.  Compensation for reflectance variation in vessel density quantification by optical coherence tomography angiography.  Invest Ophthalmol Vis Sci. 2016;57(10):4485-4492.PubMedGoogle ScholarCrossref
13.
Kraus  MF, Liu  JJ, Schottenhamml  J,  et al.  Quantitative 3D-OCT motion correction with tilt and illumination correction, robust similarity measure and regularization.  Biomed Opt Express. 2014;5(8):2591-2613.PubMedGoogle ScholarCrossref
14.
Jia  Y, Tan  O, Tokayer  J,  et al.  Split-spectrum amplitude-decorrelation angiography with optical coherence tomography.  Opt Express. 2012;20(4):4710-4725.PubMedGoogle ScholarCrossref
15.
Holm  S.  A simple sequentially rejective multiple test procedure.  Scand J Stat. 1979;6(2):65-70.Google Scholar
16.
Simonett  JM, Scarinci  F, Picconi  F,  et al.  Early microvascular retinal changes in optical coherence tomography angiography in patients with type 1 diabetes mellitus.  Acta Ophthalmol. 2017;95(8):e751-e755.PubMedGoogle ScholarCrossref
17.
Carnevali  A, Sacconi  R, Corbelli  E,  et al.  Optical coherence tomography angiography analysis of retinal vascular plexuses and choriocapillaris in patients with type 1 diabetes without diabetic retinopathy.  Acta Diabetol. 2017;54(7):695-702.PubMedGoogle ScholarCrossref
18.
Dimitrova  G, Chihara  E, Takahashi  H, Amano  H, Okazaki  K.  Quantitative retinal optical coherence tomography angiography in patients with diabetes without diabetic retinopathy.  Invest Ophthalmol Vis Sci. 2017;58(1):190-196.PubMedGoogle ScholarCrossref
19.
Samara  WA, Shahlaee  A, Adam  MK,  et al.  Quantification of diabetic macular ischemia using optical coherence tomography angiography and its relationship with visual acuity.  Ophthalmology. 2017;124(2):235-244.PubMedGoogle ScholarCrossref
20.
Pedinielli  A, Bonnin  S, Sanharawi  ME,  et al.  Three different optical coherence tomography angiography measurement methods for assessing capillary density changes in diabetic retinopathy.  Ophthalmic Surg Lasers Imaging Retina. 2017;48(5):378-384.PubMedGoogle ScholarCrossref
21.
Ghasemi Falavarjani  K, Iafe  NA, Hubschman  JP, Tsui  I, Sadda  SR, Sarraf  D.  Optical coherence tomography angiography analysis of the foveal avascular zone and macular vessel density after anti-VEGF therapy in eyes with diabetic macular edema and retinal vein occlusion.  Invest Ophthalmol Vis Sci. 2017;58(1):30-34.PubMedGoogle ScholarCrossref
22.
Karst  SG, Deak  GG, Gerendas  BS,  et al.  Association of changes in macular perfusion with ranibizumab treatment for diabetic macular edema: a subanalysis of the RESTORE (extension) study.  JAMA Ophthalmol. 2018;136(4):315-321.PubMedGoogle ScholarCrossref
23.
Agemy  SA, Scripsema  NK, Shah  CM,  et al.  Retinal vascular perfusion density mapping using optical coherence tomography angiography in normals and diabetic retinopathy patients.  Retina. 2015;35(11):2353-2363.PubMedGoogle ScholarCrossref
24.
Zahid  S, Dolz-Marco  R, Freund  KB,  et al.  Fractal dimensional analysis of optical coherence tomography angiography in eyes with diabetic retinopathy.  Invest Ophthalmol Vis Sci. 2016;57(11):4940-4947.PubMedGoogle ScholarCrossref
×