The advent of optical coherence tomography angiography (OCTA) has provided a noninvasive imaging modality used to visualize retinal vasculature in detail. The visualization of both structural and vascular information in vivo potentiates the clinical evaluation of retinal conditions such as diabetic retinopathy, wet age-related macular degeneration, and retinal vascular occlusions.1–4 Image averaging, a technique widely used in OCT, has been shown to improve retinal visualization, revealing new pathophysiology details.5 Image averaging has been utilized in OCTA to provide higher-resolution images bearing a greater signal-to-noise ratio.6 Motion-correction software available on some OCTA devices further improves image quality through volumetric registration, reducing the level of artifact.7 OCTA image averaging was first conducted using ImageJ (NIH, Bethesda, MD) and was used by Mo et al. to obtain quantitative metrics such as capillary density, demonstrating significant enhancement of image quality, and remote procedure call parameters.8 Furthermore, multiple en face OCTA image averaging has been shown to improve the visualization of the choriocapillaris by reducing image granularity.9 This technique holds promise for improving the morphological visualization of retinal structures on OCTA.
Macular telangiectasia (MacTel) type 2 is a retinal vasculature disease characterized by abnormalities in the parafoveal capillaries.10,11 During its early stages, MacTel type 2 is asymptomatic and only detectable by grey coloration of the juxtafoveal area, due to diminished retinal transparency.10 As the disease progresses, the juxtafoveal vasculature develops telangiectasias and symptoms such as blurred vision, scotomas, and metamorphopsia.10 Historically, fundus fluorescein angiography has been the gold standard for diagnosing MacTel type 2.12 However, the arrival of OCTA has provided a noninvasive imaging modality ideal for examining vasculature in individual retinal layers in high resolution. To date, there have been no published OCTA image-averaging studies conducted using Adobe Photoshop (CC 2017; Adobe, San Jose, CA), a commercially available and widely accessible image-processing software. In the present study, we assessed the utility of this software as a practical method of averaging OCTA images.
Patients and Methods
This feasibility study was conducted with seven participants from a single ophthalmology center located in Toronto, Canada. Approval was obtained from IRB Services institutional review board, and the protocol was conducted in accordance with standards of the Declaration of Helsinki. Written informed consent was obtained from all study participants.
Fourteen eyes of seven patients with MacTel type 2 were recruited to take part in this study. Participants were not included if they were younger than 25 years of age, had ocular surgery in the previous 3 months, or had any of the following conditions: uncontrolled glaucoma, ocular infection, uncontrolled systemic hypertension, or any other retinal condition.
Optimal dilation was achieved using one drop of 0.5% proparacaine hydrochloride ophthalmic solution (Alcaine; Alcon, Fort Worth, TX), 1% tropicamide ophthalmic solution (Mydriacyl; Alcon, Fort Worth, TX), and 2.5% phenylephrine hydrochloride ophthalmic solution (Mydfrin; Alcon, Fort Worth, TX) administered in each eye 15 minutes prior diagnostic imaging. Three 3.00 mm × 3.00 mm OCTA images were obtained from each eye using three commercially available devices and their native software: Cirrus 5000 Angioplex (Zeiss, Oberkochen, Germany) software version 10.0.1.19039, AngioVue (Optos, Dunfermline, UK) using RTVue XR software version 2016.1.0.26, and the Topcon Triton (Tokyo, Japan) using FastMap software V10.11. Images of the superficial retinal capillary layer and deep retinal capillary layer, choriocapillaris, and the outer retina were obtained using the automated segmentation boundaries of each device. All images were exported as 1,024 × 1,024-pixel images. In order to reduce any inter-user variability, all images were acquired by the same individual. Exclusion criteria consisted of any noticeable movement artifact and a lack of image focus.
Captured OCTA images were exported as high-resolution TIFF files then imported into Adobe Photoshop CC 2017 as an image stack. The images were then aligned by creating a Smart Object and using the “Auto-Align Layers” option. This alignment option applies a perspective layout that designates one of the three source images as the reference image and transforms the other two images to align the overlapping content.13 Scans from each of the layers were processed independently.
Qualitative Image Assessment
Using an online questionnaire, three independent retinal specialists were asked to compare and rank the original and averaged images with respect to three factors: 1) clarity of the foveal avascular zone (FAZ), 2) clarity of blood vessel delineation, and 3) ability to identify abnormal vasculature. Each original image was presented alongside the respective averaged image for comparison.
OCTA scans from two individuals were excluded due to significant image distortion as a result of movement artifact. Individual scans from either the superficial or deep capillary layers that were unable to be averaged were also excluded. The remaining scans were aligned and averaged using Adobe Photoshop CC 2017. Successful alignment and averaging were contingent on the scans containing a minimum 40% overlap. No images from Zeiss 5000 Angioplex were eligible for image averaging. A total of 25 sets of images (13 from Optovue Angiovue and 12 from Topcon Triton) were included in the image grading, bringing the overall questionnaire total to 75 questions. Following averaging, there was a noted qualitative improvement in image clarity and a reduction in the amount of noise (Figures 1 and 2). Masked image graders found the averaged images to be 87%, 89%, and 69% slightly or definitely preferable to the original, with respect to the clarity of the FAZ, clarity of blood vessel delineation, and the ability to identify abnormal vasculature, respectively (Figure 3). The image graders demonstrated intergrader consistency, with all three graders agreeing that the images are slightly or definitely preferable 75%, 87%, and 25% of the time in the context of the three criteria listed above. With respect to the clarity of the FAZ and the ability to identify abnormal vasculature, there was no significant difference between the devices and the graders' preference for the averaged image. However, on a t-test, the averaged images were preferred by graders more frequently on the Topcon Triton when assessing the clarity of blood vessel delineation (P = .0465).
(A–C) 3.0 mm × 3.0 mm spectral-domain optical coherence tomography angiograms of the deep capillary plexus scan taken using the Topcon Triton in a patient with macular telangiectasia (MacTel) type 2. (D) Photoshop averaged image of 3.0 mm × 3 .0 mm scans of the same patient with MacTel type 2 illustrating improved clarity of the deep plexus.
(A–C) 3.0 mm × 3.0 mm spectral-domain optical coherence tomography angiograms of the superficial capillary plexus scan taken using the Optovue Angiovue in a patient with macular telangiectasia (MacTel) type 2. (D) Photoshop averaged image of 3.0 mm × 3 .0 mm scans of the same patient with MacTel type 2 illustrating improved detail of the telangiectatic vessels of the superficial plexus.
Preferences of image graders comparing original optical coherence tomography angiography images to those averaged using Photoshop. The averaged image was slightly or definitely preferred over the original images 87%, 89%, or 69% of the time with respect to the clarity of foveal avascular zone, blood vessel delineation, and ability to identify abnormal vasculature, respectively.
In the present study, 25 triads of OCTA images, obtained from either the superficial or deep retinal capillary layer patients with MacTel type 2, were successfully averaged using Adobe Photoshop. Compared to the original images, the averaged images are qualitatively of greater resolution and were preferred by retinal specialists with respect to three criteria: 1) clarity of the FAZ, 2) clarity of blood vessel delineation, and 3) ability to identify abnormal vasculature. Despite the technique producing preferable images across all criteria, the graders had the lowest levels of agreeability with respect to the ability to identify abnormal vasculature. Although 69% of the averaged images were found to be superior, respondents were only in unanimous agreement that they were indeed preferable 25% of the time. This may in part be due to high original image resolution as opposed to a deficiency in processing. These results may have implications for the technique's utility, with greater potential for research uses as opposed to diagnostic or clinical use in MacTel Type 2. The technique consistently improved the clarity of the FAZ and blood vessel delineation and may be used for both clinical research and diagnostic assessment in retinal conditions such as diabetic retinopathy and retinal venous occlusion.14,4 The graders preference for averaged images taken by the Topcon Triton with respect to blood vessel delineation may indicate that this technique may be optimized for use with swept source as opposed to spectral domain OCTA. This may in part be due to the ability of swept-source OCTA to further penetrate the vitreous using a longer wavelength than that of conventional spectral domain OCTA and offer a higher resolution image.15 Further studies will be required to elucidate this effect and the optimal image parameters for this method of averaging.
Overall, the major benefit of this technique lies in its ease and simplicity of use. The described technique involves two major steps focused on the creation of a Smart Object. This is in contrast to previously described methodologies, which require several registration steps involving the installation of multiple plugins.9,16
A major limitation of this technique is the inability of the software to average images with less than 40% overlap, restricting its use to scans that exhibit very minimal artifact or motion. This directly resulted in the mandatory exclusion of the OCTA images obtained from the Zeiss 5000 Angioplex, whose images were not successfully registered and averaged due to significant motion artifact and low image overlap in all eyes in our sample. The study was also limited by a small sample size and strict exclusion criteria, reducing the number of original image sets.
In conclusion, the technique of image averaging provides higher-resolution images and is valuable for improving the morphological visualization of retinal structures on OCTA.6 Despite image averaging being commonplace, research investigating various averaging methods has been limited. This was the first study to examine the use of a commercially available photo editing software as a potential OCTA image averaging tool. Adobe Photoshop is an easily accessible and adaptable software with potential for use in OCTA averaging and morphologic assessment of retinal vasculature. Future studies can utilize this technique to optimize the acquisition of various imaging parameters, increasing OCTA efficiency, and improving patient comfort.
- Chalam KV, Sambhav K. Optical coherence tomography Angiography in retinal diseases. J Ophthalmic Vis Res. 2016;11(1):84–92. https://doi.org/10.4103/2008-322X.180709 PMID: doi:10.4103/2008-322X.180709 [CrossRef]27195091
- de Carlo TE, Chin AT, Bonini Filho MA, et al. Detection of microvascular changes in eyes of patients with diabetes but not clinical diabetic retinopathy using optical coherence tomography angiography. Retina. 2015;35(11):2364–2370. https://doi.org/10.1097/IAE.0000000000000882 PMID: doi:10.1097/IAE.0000000000000882 [CrossRef]26469537
- Palejwala NV, Jia Y, Gao SS, et al. Detection of nonexudative choroidal neovascularization in age-related macular degeneration with optical coherence tomography angiography. Retina. 2015;35(11):2204–2211. https://doi.org/10.1097/IAE.0000000000000867 PMID: doi:10.1097/IAE.0000000000000867 [CrossRef]26469533
- Kashani AH, Lee SY, Moshfeghi A, Durbin MK, Puliafito CA. Optical coherence tomography angiography of retinal venous occlusion. Retina. 2015;35(11):2323–2331. https://doi.org/10.1097/IAE.0000000000000811 PMID: doi:10.1097/IAE.0000000000000811 [CrossRef]26457395
- Sander B, Larsen M, Thrane L, Hougaard JL, Jørgensen TM. Enhanced optical coherence tomography imaging by multiple scan averaging. Br J Ophthalmol. 2005;89(2):207–212. https://doi.org/10.1136/bjo.2004.045989 PMID: doi:10.1136/bjo.2004.045989 [CrossRef]15665354
- Waheed NK, Duker JS. Image averaging, a powerful tool in optical coherence tomography and optical coherence tomography angiography. JAMA Ophthalmol. 2017;135(11):1204–1205. https://doi.org/10.1001/jamaophthalmol.2017.4015 PMID: doi:10.1001/jamaophthalmol.2017.4015 [CrossRef]28983576
- Camino A, Zhang M, Gao SS, et al. Evaluation of artifact reduction in optical coherence tomography angiography with real-time tracking and motion correction technology. Biomed Opt Express. 2016;7(10):3905–3915. https://doi.org/10.1364/BOE.7.003905 PMID: doi:10.1364/BOE.7.003905 [CrossRef]27867702
- Mo S, Phillips E, Krawitz BD, et al. Visualization of radial peripapillary capillaries using optical coherence tomography angiography: the effect of image averaging. PLoS One. 2017;12(1):e0169385. doi:10.1371/journal.pone.0169385 [CrossRef]. PMID: 28068370
- Uji A, Balasubramanian S, Lei J, Baghdasaryan E, Al-Sheikh M, Sadda SR. Choriocapillaris imaging using multiple en face optical coherence tomography angiography image averaging. JAMA Ophthalmol. 2017;135(11):1197–1204. https://doi.org/10.1001/jamaophthalmol.2017.3904 PMID: doi:10.1001/jamaophthalmol.2017.3904 [CrossRef]28983552
- Roisman L, Rosenfeld PJ. Optical Coherence Tomography Angiography of Macular Telangiectasia Type 2. Dev Ophthalmol. 2016;56:146–158. https://doi.org/10.1159/000442807 PMID: doi:10.1159/000442807 [CrossRef]27022942
- Chidambara L, Gadde SGK, Yadav NK, et al. Characteristics and quantification of vascular changes in macular telangiectasia type 2 on optical coherence tomography angiography. Br J Ophthalmol. 2016;100(11):1482–1488. https://doi.org/10.1136/bjophthalmol-2015-307941 PMID: doi:10.1136/bjophthalmol-2015-307941 [CrossRef]26823394
- Toto L, Di Antonio L, Mastropasqua R, et al. Multimodal Imaging of Macular Telangiectasia Type 2: Focus on Vascular Changes Using Optical Coherence Tomography Angiography. Invest Ophthalmol Vis Sci. 2016;57(9):OCT268–OCT276. https://doi.org/10.1167/iovs.15-18872 PMID: doi:10.1167/iovs.15-18872 [CrossRef]27409482
- Align and distribute layers. Adobe website. https://helpx.adobe.com/photoshop/using/aligning-layers.html#automatically_align_image_layers. Published 2018. Accessed March 3, 2019.
- Freiberg FJ, Pfau M, Wons J, Wirth MA, Becker MD, Michels S. Optical coherence tomography angiography of the foveal avascular zone in diabetic retinopathy. Graefes Arch Clin Exp Ophthalmol. 2016;254(6):1051–1058. https://doi.org/10.1007/s00417-015-3148-2 PMID: doi:10.1007/s00417-015-3148-2 [CrossRef]
- Unterhuber A, Povazay B, Hermann B, Sattmann H, Chavez-Pirson A, Drexler W. In vivo retinal optical coherence tomography at 1040 nm - enhanced penetration into the choroid. Opt Express. 2005;13(9):3252–3258. https://doi.org/10.1364/OPEX.13.003252 PMID: doi:10.1364/OPEX.13.003252 [CrossRef]19495226
- Uji A, Balasubramanian S, Lei J, et al. Multiple enface image averaging for enhanced optical coherence tomography angiography imaging. Acta Ophthalmol. 2018;96(7):e820–e827. https://doi.org/10.1111/aos.13740 PMID: doi:10.1111/aos.13740 [CrossRef]29855147