Article Text

Download PDFPDF

Wide-field en face swept-source optical coherence tomography angiography using extended field imaging in diabetic retinopathy
  1. Takao Hirano1,2,
  2. Shinji Kakihara2,
  3. Yuichi Toriyama2,
  4. Muneeswar Gupta Nittala1,
  5. Toshinori Murata2,
  6. Srinivas Sadda1,3
  1. 1 Doheny Image Reading Center, Doheny Eye Institute, Los Angeles, California, USA
  2. 2 Deparment of Ophthalmology, Shinshu University School of Medicine, Matsumoto, Nagano, Japan
  3. 3 Department of Ophthalmology, David Geffen School of Medicine at UCLA, Los Angeles, California, USA
  1. Correspondence to Dr Srinivas Sadda, Doheny Image Reading Center, Doheny Eye Institute, Los Angeles, California 90033, USA; ssadda{at}doheny.org

Abstract

Aims To examine the feasibility of wide-field en face swept-source optical coherence tomography angiography (SS-OCTA) with extended field imaging (EFI) for evaluation of the retinal vasculature in diabetic retinopathy (DR).

Methods This study included 37 eyes of 27 patients (age, 65±10 years; male patients, 18; female patients, 9) with DR. All patients underwent comprehensive ophthalmological examination, including OCTA and fluorescein angiography (FA). The imaging methods were compared for visible field of view, presence and extent of non-perfused areas (NPAs), presence and number of new blood vessels (NVs), vessel density (VD) and patient comfort level measured by Visual Analogue Scale.

Results SS-OCTA with EFI allowed capture of larger areas (by 1.80±0.18 times on average) of the fundus than SS-OCTA without EFI. Compared with FA, the sensitivities of SS-OCTA with EFI for detection of NPAs and NVs were 96% and 79%, respectively, with specificities of 100% and 96%, respectively. There was no significant difference in extent of NPAs (61.2±45.8 vs 61.5±55.0 disc areas, P=0.99) or number of NVs (1.5±3.3 vs 0.9±1.8, P=0.68) between FA and SS-OCTA with EFI. VD showed significantly lower values in EFI SS-OCTA than in those acquired without EFI (31.6%±4.3%  vs 34.2%±4.3%, P<0.001). Wide-field OCTA with EFI was significantly more comfortable for patients than FA (P<0.001).

Conclusions SS-OCTA with EFI allows acquisition of wide-field en face images of the retinal vasculature in patients with DR, with greater patient comfort than FA.

  • retina
  • imaging

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Diabetic retinopathy (DR) is the most common cause of vision loss among individuals with diabetes and the leading cause of vision impairment and blindness among adults of working age.1 The diagnosis, treatment planning and follow-up of DR are currently based on imaging findings obtained by fluorescein angiography (FA) and optical coherence tomography (OCT).2–4

FA has been used clinically in ophthalmology for nearly 50 years and is currently the gold standard for clinical evaluation of retinal vascular alterations in DR.5 However, the technique has two disadvantages. The first is that it is an invasive method requiring intravenous dye injection, which may cause nausea and, rarely but critically, may cause anaphylaxis, even in healthy subjects. Patients with severe DR often have systemic vascular complications such as associated renal and cardiovascular diseases.6 Therefore, caution is required during dye injection for FA in patients with severe DR.7 The other disadvantage of FA is that it produces two-dimensional images in which fluorescence signals of the superficial and deep capillary networks overlap and are difficult to distinguish.

In contrast, OCT angiography (OCTA), an imaging technique based on motion contrast, allows reconstruction of three-dimensional chorioretinal images without dye injection.8 9 It is a non-invasive, quick and repeatable method for assessment of retinal ischaemia and neovascularisation.10 11 However, one of its limitations is the size of scanning area, which is a concern since vascular alterations in DR extend beyond the posterior pole. Although the PLEX Elite 9000 (Carl Zeiss Meditec, Dublin, California, USA)—which received the first US Food and Drug Administration clearance for swept-source OCTA (SS-OCTA)—can scan a wide field (12×12 mm), its field of view does not compare to wide-field FA.

Uji and Yoshimura reported a novel yet simple technique for extending the scan length of OCT—this extended field imaging (EFI) technique uses trial frames fitted with a +20 diopter lens.12 Although wide-field SS-OCTA using the EFI technique is assumed to be useful for wider and non-invasive evaluation of retinal vasculature, little research has been performed using spectral-domain OCTA (SD-OCTA) for quantifying retinal vasculature in patients with retinal vein occlusion,13 and no such study has been conducted in patients with DR.

In this study, we examined the feasibility of wide-field en face SS-OCTA using the EFI technique for evaluating the retinal microvasculature in eyes with DR.

Materials and methods

This prospective observational study included 37 eyes of 27 patients with DR who visited the outpatient clinic of the ophthalmology department of Shinshu University between November 2016 and January 2017. The study adhered to the tenets set forth in the Declaration of Helsinki. Written informed consent was obtained from all patients. All patients underwent comprehensive ophthalmological examination, including measurement of best-corrected visual acuity (BCVA), slit-lamp biomicroscopy, colour fundus photography, OCTA and FA with the Heidelberg Retina Angiograph 2 (HRA2; Heidelberg Engineering, Heidelberg, Germany). Data on age, sex and previous haemoglobin A1c (HbA1c; National Glycohemoglobin Standardization Program) levels were collected from the medical records. After examination with both FA and wide-field OCTA using the EFI technique, the testing comfort levels of the patients were assessed using a Visual Analogue Scale (VAS), with 0 indicating no discomfort and 10 indicating unbearable discomfort. Eyes with poor-quality OCTA images because of cataract, poor fixation and vitreous haemorrhage were excluded from the study.

OCTA with and without EFI

En face SS-OCTA images centred on the fovea with a scanning size of 12×12 mm were acquired with and without EFI using the PLEX Elite 9000 scanner (Carl Zeiss Meditec). EFI was performed in accordance with the method described in a previous report (online supplementary figure 1).12

Supplementary file 1

Evaluation of magnitude of extension and expansion of field of view

The magnitude of the extension of the horizontal and vertical field of view of SS-OCTA images acquired with EFI (EFI SS-OCTA) were calculated by a single non-masked observer (TH) and compared with those of images acquired without EFI. Horizontal and vertical distances between two landmark points (mainly the branching points of arcade vessels) were measured using ImageJ software (National Institutes of Health, Bethesda, Maryland, USA), and the corresponding extension ratio was calculated using the formula: distance (pixels) between two landmark points in SS-OCTA images acquired without EFI/Distance (pixels) between two landmark points in EFI SS-OCTA for both horizontal and vertical dimensions. The overall expansion of the field of view was then computed by multiplying the horizontal and vertical extension ratio values.

Evaluation of retinal vasculature

Two observers (YT and TM), who were masked to the clinical status, evaluated the retinal vasculature in each eye for the presence of ischaemia and neovascularisation using both FA and EFI SS-OCTA. Images were shuffled and presented to the graders in random order. In particular, ischaemia was determined by the presence of non-perfused areas (NPAs), which were measured manually in units of disc area (DA), with one DA=2.54 mm2. Neovascularisation was determined by the presence and number of new blood vessels (NVs).

Vessel density (VD) of EFI SS-OCTA was calculated by a single non-masked observer (TH) and compared with that of images acquired without EFI. VD was analysed using ImageJ for comparison between SS-OCTA images with and without EFI. For measurements, after importing the SS-OCTA image as an 8-bit image (1024×1024 pixels) into Image J, we cropped the upper part (1024×960 pixels) to exclude the watermark. This image was converted into a binary image using a modified version of the previously reported method.14 15 After processing with a top-hat filter, the image was separately processed to create two distinct binarised images: one was processed first by a Hessian filter, followed by global thresholding using Huang’s fuzzy thresholding method, and the other (duplicate) image was binarised through median local thresholding. Only pixels that were detected on both binarised versions were included in the final binarised image (figure 1).

Figure 1

Swept-source optical coherence tomography angiography (SS-OCTA) images of patients with diabetic retinopathy with quantitative image outputs. Upper left: SS-OCTA image acquired without extended field imaging (EFI) (12×12 mm, 1024×1024 pixel). Upper middle: The 1024×960 pixel rectangular area was cropped from the 12×12 mm SS-OCTA images without EFI (outlined in yellow). Upper right: Binarised image of the cropped area. Lower left: SS-OCTA image acquired with EFI (12×12 mm, 1024×1024 pixel). Lower middle: The 1024×960 pixel rectangular area was cropped from the 12×12 mm SS-OCTA images with EFI (outlined in yellow). Lower right: Binarised image of the cropped area.

Statistical analyses

Statistical analyses were performed using Statistical Packages for Social Sciences V.22.0 (IBM). Continuous variables were expressed as mean values±SD or median and IQR. Intraclass correlation coefficients (ICCs) were used to estimate the agreement between individual measurements of the two readers. Since the ICC between the two readers was consistently >0.8, comparison of continuous variables between FA and OCTA images was performed using non-parametric Wilcoxon matched-pairs signed-rank test with measurements recorded by just one of the readers. P values <0.05 were judged to be statistically significant.

Results

Patient characteristics

Of a total of 54 eyes imaged for this study, 17 eyes (31%) were excluded from the final analysis cohort due to image artefacts, primarily related to poor fixation. Thus, 37 eyes from 27 patients (age, 65±10 years; male patients, 18; female patients, 9) were ultimately included in this analysis. All patients had type 2 diabetes, with a mean HbA1c level of 7.9%±1.9%. In terms of severity of DR, 4 eyes had mild non-proliferative DR (NPDR); 9, moderate NPDR; 11, severe NPDR and 13, proliferative DR. Of the 37 eyes, 19 had already undergone panretinal photocoagulation and 21 had diabetic macular oedema. The mean logarithm of the minimum angle of resolution BCVA was 0.07 (approximate Snellen equivalent 20/23)±0.22 (range, −0.18 to 0.70, approximate Snellen equivalent 20/13 to 20/100).

Extension and expansion ratios

The mean extension ratios in the horizontal and vertical directions were 1.34±0.07 (range, 1.22–1.48) and 1.33±0.07 (range, 1.23–1.47), respectively. The extension ratios in the two dimensions were not significantly different (P=0.26, Wilcoxon matched-pairs signed-rank test) and showed a strong positive correlation (correlation coefficient=0.79, P<0.0001, Spearman correlation test). SS-OCTA images acquired with EFI captured larger areas of the fundus than did SS-OCTA images acquired without EFI (larger by an average ratio of 1.80±0.18 (range, 1.50–2.18)) or FA images, which captured a 55° field using the HRA2 (figure 2).

Figure 2

Representative case of a 64-year-old patient with proliferative diabetic retinopathy. Left: Swept-source optical coherence tomography angiography (SS-OCTA) image acquired without extended field imaging (EFI) (12×12 mm). Middle: SS-OCTA image (12×12 mm) acquired with EFI. Right: Fluorescein angiogram (FA) image obtained by Heidelberg Retina Angiograph 2 (HRA2) (55°). Both white and black arrowheads represent the same points in each image. White dot square represents the same area with SS-OCTA image acquired without EFI (12×12 mm). SS-OCTA image acquired with EFI captured 1.75 times as large as SS-OCTA images acquired without EFI. SS-OCTA image acquired with EFI can capture even more than the 55° field using the HRA2.

Evaluation of retinal vasculature

The specificity of EFI SS-OCTA for detection of NPAs was 100%, with a sensitivity of 96%; the corresponding agreement (kappa statistic) with FA findings was 0.874. The Wilcoxon signed-rank test did not identify a significant difference in NPAs between FA images and EFI SS-OCTA images (61.2±45.8 vs 61.5±55.0 DA, P=0.99).

The specificity of EFI SS-OCTA for detection of NVs was 96%, with a sensitivity of 79%; the corresponding agreement (kappa statistic) with FA findings was 0.763. The Wilcoxon signed-rank test did not identify a significant difference in the number of NVs between FA images and EFI SS-OCTA images (1.5±3.3 vs 0.9±1.8, P=0.68).

Interobserver agreement of NPA measurements for FA images and EFI SS-OCTA images, as assessed by ICC were 0.83 (95% CI 0.68 to 0.91) and 0.91 (95% CI 0.83 to 0.96), respectively; the corresponding ICCs for measurements of NVs were 0.97 (95% CI 0.94 to 0.99) and 0.82 (95% CI 0.65 to 0.91), respectively.

VD showed significantly lower values in EFI SS-OCTA than in SS-OCTA images acquired without EFI (31.6%±4.3% vs 34.2%±4.3%, P<0.001) (figure 3).

Figure 3

Vessel density (VD) of swept-source optical coherence tomography angiography (SS-OCTA) images with and without extended field imaging (EFI). VD showed significantly lower values in SS-OCTA images with EFI than in SS-OCTA images without EFI (31.6%±4.3% vs 34.2%±4.3%, ***P<0.001).

Intraexamination comfort level

Wide-field OCTA with EFI (mean VAS score, 1.5±1.3; median (25%–75% IQR), 11 2) was more comfortable for patients than FA (mean VAS score, 4.1±1.9; median (25%–75% IQR) 43–5; P<0.0001, Wilcoxon matched-pairs signed-rank test).

Discussion

OCTA is a non-invasive, rapid and reproducible method for assessment of the chorioretinal vasculature without dye injection. However, one of its limitations is the field of view or the size of the scanning area. Although the PLEX Elite 9000 captures the widest OCTA images among commercially available OCTA devices, its maximum scan size (12×12 mm) may not be sufficient for evaluation of retinal diseases which cause significant alterations beyond the posterior pole, such as DR.

In clinical investigations and research studies, montage images may be constructed manually from multiple 6×6 mm SD-OCTA scans. This may not be practical, however, in the context of a busy clinical practice if there is a desire to use OCTA data for therapeutic decision-making in eyes with DR. Recently, some commercial OCTA devices have featured algorithms that can semiautomatically create montage images of OCTA scans to simulate a wider field of view.16 These montage-based methods, however, necessarily require: (1) extra time to obtain multiple scans (with overlapping areas), (2) an extra step to perform the montage operation and (3) inaccuracies due to subtle misalignments.

The EFI technique is a novel yet simple method for extending the scan length of SD-OCT B-scan images.12 It involves imaging of the posterior pole through trial frames fitted with +20 D lenses. In this study, we applied this technique to acquire en face SS-OCTA images for evaluation of DR. Our results indicate that en face SS-OCTA with EFI can capture larger areas (by an average ratio of 1.80±0.18) of the fundus than SS-OCTA without EFI in approximately 10 s, while maintaining a similar aspect ratio. The scan size of SS-OCTA with EFI in our study was actually larger than that of an FA, captured using 55° field of view setting.

We also performed qualitative and quantitative comparison of EFI SS-OCTA and FA images of the same region in order to determine whether EFI SS-OCTA was useful and sufficient for assessment of the retinal vasculature in DR. The results of the quantitative comparison of the two imaging techniques in terms of presence or absence of NPA and NVs revealed the performance of EFI SS-OCTA to be satisfactory, as evident from its sensitivity, specificity and kappa coefficients. Moreover, on quantitative comparison in terms of extent of NPAs and number of NVs, no significant differences were observed between FA and EFI SS-OCTA. We also performed quantitative comparison of SS-OCTA images acquired with and without EFI in connection with VD. It should be noted that VD showed significantly lower values in EFI SS-OCTA than in SS-OCTA images acquired without EFI. This is not surprising as we have seen lower VD values in 6×6 mm field images compare to 3×3 mm field images in a prior study—this was thought to be due to differences in transverse resolution between difference scan patterns.17 One would expect EFI scans that cover a larger territory (with the same number of A-scans) would have an effective lower resolution, which would translate to a lower VD. Another possible explanation or contributor to the difference is that the wider field SS-OCTA EFI images may have included additional areas of non-perfusion near the borders of the scan area.

Despite scanning a larger field, the OCTA images in this study were relatively free of motion artefact, which may be a reflection of the acquisition technology. Both SD-OCTA and SS-OCTA use Fourier domain detection techniques; however, SD-OCTA devices use a broadband near-infrared super luminescent diode as a light source, while SS-OCTA devices use a tunable swept laser and a single photo detector.16 Because of these differences, SS-OCTA can capture fundus images of the same area more quickly than SD-OCTA. This higher speed of scanning may have contributed to the relative lack of motion or other artefacts in the images captured for the study. We also assessed patient comfort level during FA and SS-OCTA with EFI. Not surprisingly given the lack of need for dye injection and a shorter acquisition time, EFI SS-OCTA was significantly more comfortable than FA.

There are some limitations to our study which should be considered. Because EFI does not change the hardware capabilities of the OCT system, it causes a necessary decrease in image resolution. Since the EFI technique simply magnifies each pixel, subtle vascular details could potentially be missed in EFI SS-OCTA images. For this reason, the suitability of EFI may need to be assessed for each disease application. As an example, the efficiency of EFI SS-OCTA to detect choroidal neovascular lesions is yet to be determined. On the other hand, EFI SS-OCTA does appear to be useful in the setting of DR, where retinal NVs and areas of non-perfusion are well seen and quantified by EFI SS-OCTA.

Although no statistically significant differences were observed in this study, the number of NVs measured on EFI SS-OCTA images tended to be smaller than those measured on FA images. This is likely due to the fact that OCTA does not depict dye leakage from blood vessels. Thus, whereas neovascular lesions on FA may have appeared larger due to early adjacent leakage, OCTA may have better depicted their true physical extent. Alternatively, since OCTA only depicts vessels with flow within a certain velocity range, it is possible that some portions of the NV revealed by FA were not visualised by OCTA. Another possible contributor to smaller number of NVs on OCTA images is segmentation error. Since OCTA has superior depth resolution to FA and areas of NV may span a significant axial distance (above the retinal surface), selection of appropriate segmentation boundaries would be essential in order to ensure that the slab of interest included the entire region of NVs. Figure 4 shows the use of the instrument predefined vitreoretinal slab (which extends anteriorly from the automatically detected retinal surface) which appears to be ideally suited for displaying retinal NVs.

Figure 4

Swept-source optical coherence tomography angiography (SS-OCTA) images with or without extended field imaging (EFI) and fluorescein angiogram (FA) image. Upper: Each slab (vitreoretinal interface (VRI), superficial, deep, whole retina) of SS-OCTA images without EFI. Lower: Each slab of SS-OCTA images with EFI. Right: FA image of the same area of the 12×12 mm SS-OCTA images with EFI. Three new extraretinal vessels were clearly depicted in the VRI images, similar to FA image (blue dot circle).

Another limitation of the present technique is that there is learning curve before good quality EFI SS-OCTA images can be reliably obtained. In this initial study, it took approximately 1 month for the investigators to gain the ability to acquire images of adequate quality for this study. However, even afterwards, approximately 30% of cases were excluded from analysis because of image artefacts, primarily due to poor fixation. Another possible cause was positional instability of +20 D lenses between the eye and device, which may have produced various optical artefacts. In future studies, the appropriate lens position may need to be determined and fixed for each patient.

Future technological developments can be expected to lead to the advent of OCTA devices capable of scanning larger areas of the retina than is now possible. Such devices, however, may be associated with a higher cost and may not be broadly available for some time. In the interim, the EFI technique may prove to be a useful approach to expand the field of view of OCTA imaging, which may facilitate the assessment of diseases which affect the vasculature in more peripheral regions of the retina.

References

Footnotes

  • Contributors TH, TM and SS: study conception, design, analysis, interpretation of data, drafting and revising, final approval. SK, YT and MGN: data acquisition, drafting and revising, final approval.

  • Funding SS is a consultant for and receives research support from Optosand Carl Zeiss Meditec, and serves as a consultant for Centervue, and has access to research instruments provided by Heidelberg Engineering, Topcon Medical Systems, Optos, Carl Zeiss Meditec, Nidek, and Centervue.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Ethics committee of the Shinshu University School of Medicine (approval number 2044).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost B Jonas