Review Optical coherence tomography angiography: an overview of the technology and an assessment of applications for clinical research Andrew Koustenis Jr,1 Alon Harris,1 Josh Gross,1 Ingrida Januleviciene,2 Aaditya Shah,1 Brent Siesky1

1Department of ABSTRACT it is now widely accepted that vascular factors play , Indiana In recent years, ophthalmology has experienced a role in the risk of development and progression University School of , fi 2–6 Indianapolis, Indiana, USA signi cant developments with respect to imaging of the disease. Currently, other imaging modal- 2Eye Clinic of Lithuanian modalities. Optical coherence tomography angiography is ities are used to illustrate retinal capillary beds, University of Health Sciences, one such technology that seeks to improve diagnostics such as confocal scanning laser Doppler flowmetry Kaunas, for retinal diseases. Using standard structural ocular (Heidelberg retinal flowmetry (HRF), Heidelberg coherence tomography hardware, optical coherence Engineering, Heidelberg, Germany), which mea- Correspondence to Professor Alon Harris, tomography angiography demonstrates the ability to sures the amount of a vascular tissue and blood Research and non-invasively visualise the vasculature in the and flow within the peripapillary retinal capillary Diagnostic Center, Eugene and the with high resolution, allowing greater insight beds.7 In addition, fluorescein angiography (FA) is Marilyn Glick Institute, into retinal vascular . In addition, retinal and used to visualise the superficial retinal vascular beds Indiana University School of fl fl Medicine, Indianapolis, choroidal vessel density and blood ow can be and extravasation of vascular uid secondary to IN 46202, USA; quantified, offering potential to assist in the diagnosis of retinal . Other forms of OCT exist, such [email protected] a variety of retinal diseases. To date, numerous retinal as Doppler OCT, which assesses the retinal branch diseases, such as open-angle glaucoma, have been vessels for blood velocity and volumetric flow rate.8 Received 21 July 2016 found to possess a vascular component. Specifically, These imaging techniques, however, have found Revised 7 September 2016 fi Accepted 17 September 2016 ischaemia of the head and lamina cribrosa limitation in their ability to con rm pathological Published Online First has been theorised as a causative factor in ganglion cell vascular mechanisms in the ONH and lamina cri- 4 October 2016 death; however, confirmation of this mechanism has brosa that could be involved in the development been prohibited by the limitations of currently existing and progression of OAG. OCT angiography imaging modalities. Optical coherence tomography (OCT-A) is a retinal vascular imaging technology angiography provides clear imaging of these regions and that uses a novel algorithm to generate high- the possibility to elucidate further understanding of resolution images and quantify vessel density and vascular factors that contribute to glaucoma blood flow of the retina and choroid. Vessel density development and progression. Furthermore, this imaging changes in OAG, which account for significant modality may provide insight to neural pathologies with blindness globally,9 have been demonstrated with vascular components such as Alzheimer’s disease. OCT-A. In this review, the authors discuss the Herein, the authors discuss the theory of operation for OCT-A technology and current OAG findings. In optical coherence tomography angiography and the addition, the authors offer speculation on further current findings from pilot studies with a focus on open- applications of OCT-A to investigate retinal vascu- angle glaucoma. In addition, speculation is offered for lar abnormalities associated with neurodegenerative future applications of the technology to study other diseases such as Alzheimer’s disease (AD). diseases with microvascular contributions. TECHNOLOGY OVERVIEW: FUNCTION, ADVANTAGES AND DISADVANTAGES INTRODUCTION OCT-A uses laser light to produce a clear image of Evaluation of retinal pathology in vivo has made the retinal and choroidal microvasculature. An great improvements as the advancement of imaging overview schematic of the system’s operation is modalities has allowed for the quantification of depicted in figure 1. various tissues. Optical coherence tomography The device incorporates an infrared laser that (OCT) was developed in the early 1990s and has shines on the posterior retina and is reflected by been used both in research and in clinical practice the tissue. The reflected light is detected by the to image retinal structures such as the retinal nerve instrument and, once converted from an analogue fibre layer (RNFL), macula and optic nerve head to digital signal, is processed by the system’s com- (ONH). Specifically, retinal structural deficits asso- puter.10 During a scan of an individual’s retina, ciated with the development and progression of individual scans of layers of the retina are collected, open-angle glaucoma (OAG) have been identified. which are known as A-scans. A-scans are compiled For example, Hoh et al1 found that in patients into a B-scan, which is used to analyse both cross- with OAG, OCT showed RNFL and macula thin- sectional structural information and blood flow fl To cite: Koustenis A , ning with disease progression. However, one limita- information. The data collected from the re ected Harris A, Gross J, et al. Br J tion of OCT is that it does not provide information light are usually used to assess structural aspects of Ophthalmol 2017;101:16– pertaining to retinal blood flow. As the aetiology of the ONH, such as RNFL and macula thickness. 20. OAG has expanded to include multiple influences, However, in OCT-A, an additional algorithm is

16 Koustenis A , et al. Br J Ophthalmol 2017;101:16–20. doi:10.1136/bjophthalmol-2016-309389 Review

signal-to-noise ratio (SNR). This is achieved via two methods: considering the number of B-scans (N) and the number of band- width segmentations (M). While the full scan is relatively short at about 3 s, saccadic movements of the eye can cause a low SNR. Therefore, averaging the decorrelation across the number of B-scans increases the SNR. In addition, the equation averages the decorrelation value across M number of bandwidth segmen- tations.10 This process is depicted in figure 2. In addition to having a high SNR, the SSADA algorithm has a high level of connectivity in the image. This means that the vas- culature shown in the image appears smooth and continuous.11 This is important for allowing the image to be interpreted clearly. Once the algorithm has processed the data, the output of the OCT-A algorithm is an image of the retinal and choroidal vascu- Figure 1 Overall schematic of the optical coherence tomography lature, which can be segmented into four zones: the superficial 10 angiography system. Adapted with permission. plexus, the deep plexus, the outer retina and the choroid.12 13 This includes the full depth of the .11 Therefore, the vasculature from the superficial retina to the lamina cribrosa can be seen,14 including into the pores of the lamina cribrosa. From applied to calculate blood flow. The algorithm discussed here is this image, qualitative assessments can be made for pathologies, – known as the split-spectrum amplitude decorrelation angiog- such as glaucoma and diabetic .15 17 However, a raphy (SSADA) algorithm. This algorithm, which calculates the quantified assessment of blood flow can be obtained in the form 10 decorrelation in the reflected light, is shown in equation 1. of the flow index, as shown in equation 2, and the vessel density, which can be calculated as the per cent of an area con- NX1 XM Dðx; zÞ¼ 1 1 taining blood flow. 1 N M 1 n¼ 1 m¼1 A DVdA Anðx; zÞ Anþ ðx; zÞ dA 1 M ¼ ; N ¼ A 2 2 ( 4 8) ½ð1=2ÞAnðx; zÞ þð1=2ÞAnþ1ðx; zÞ Equation 2: The flow index, an average of the decorrelation Equation 1: The SSADA algorithm calculates the decorrelation values for a given area, is useful in quantifying the in reflected light at consecutive points, allowing for the vascula- microvasculature.11 ture to be visualised.10 The resulting outputs of OCT-A offer improvements over that The SSADA algorithm considers the fluctuation in the ampli- of alternative imaging modalities. Compared with HRF, OCT-A tude (A) of the reflected light between the consecutive B-scans offers insight to vessel density, whereas HRF only provides at each spatial location within the collected data. The static insight into blood flow velocity and the amount of avascular tissue would have a low decorrelation value, meaning that the zone present in a field of view.7 Another common imaging amplitude of reflected light does not fluctuate between B-scans. modality is FA. FA is an invasive process, as it requires an inject- Therefore, blood flow is represented by a high decorrelation able dye. In addition, time must be allowed for the dye to circu- value, specifically over a threshold of D=0.125.10 A crucial late throughout the body before images can be taken. advantage of the SSADA algorithm is maximising the Furthermore, the dye can illicit an anaphylactic reaction in some

Figure 2 Process of bandwidth segmentation, used by the split-spectrum amplitude decorrelation angiography algorithm to increase the signal-to-noise ratio. The bandwidth was segmented into four sections to maintain appropriate image resolution.10 Adapted with permission.

Koustenis A , et al. Br J Ophthalmol 2017;101:16–20. doi:10.1136/bjophthalmol-2016-309389 17 Review

patients.18 OCT-A eliminates these disadvantages, due to the short scan time and lack of a dye. In addition, the radial peripa- pillary capillaries (RPCs) that supply the RNFL can be seen with OCT-A, while not with FA.11 Furthermore, FA is limited in only providing a two-dimensional image, lacking depth information, while OCT-A allows for three-dimensional imaging.19 Finally, groups have shown OCT-A to have reasonable specificity and sensitivity when detecting pathologies such as choroidal neovas- cularisation CNV.20 While OCT-A has advantages over other imaging modalities, it is important to recognise limitations. For example, it must be acknowledged that OCT-A specifically analyses the posterior Figure 3 Optical coherence tomography angiography images retinal and choroidal microvasculature. While a specialised comparing the total optic nerve head vasculature in a healthy control adaptor to image capillaries in the anterior segment has been (left) and a patient with glaucoma (right).14 Adapted with permission. developed,21 not all of the relevant ocular vasculature is imaged, such as the retrobulbar vessels. Also, if one is interested in deeper vessels of the retina, one must be aware that the super- vessel density from a pilot study of patients with glaucoma and fi 10 cial vessels may obscure the layer of interest. This issue healthy controls. requires the investigator to consider the image with context, While retinal microvascular deficits in the total ONH are adding a level of subjectivity to the interpretation. Another chal- important to study, consideration should be given to specific lenge to the interpretation of the image is the presence of arte- regions of the retina, as well. When imaging the retinal vascula- fl facts. High velocity blood ow may create a void in the signal. ture in the periphery of the optic nerve in patients with OAG Retinal pigment epithelial detachment may cause artefacts to using OCT-A, Hollo et al showed that a decrease in the density present in the image, as well. An investigator should be aware of 22 of the RPC was correlated with a decrease in the thickness of such situations when evaluating an OCT-A image. In addition, the RNFL. The areas of RPC density decreases spatially and cor- if the results indicate a loss of vessel density, one cannot differ- responded with the areas of RNFL thinning, including early entiate the aetiology between tissue (capillary) loss and acute 30 11 RNFL thinning. Studies such as these are useful in elucidating ischaemia, necessitating contextual information to deduce the relationship between retinal vascular changes and retinal aetiology. For example, if one observed a decrease in RNFL structural changes seen in glaucoma. Yarmohammadi et al31 thickness in addition to a decrease in the density of the RPCs, used OCT-A to expand these findings, observing a gradation in one could infer that there is a chronic loss of capillaries, rather the decrease of vessel density relating to the severity of glau- than an acute ischaemic event. Furthermore, patient movement coma, in both peripapillary capillary density and total ONH can reduce the quality of the OCT-A image, requiring the vessel density. Furthermore, patients with less severe glaucoma patient to remain still and avoid blinking during the 23 were found to have a smaller decrease in vessel density from examination. In addition, while the SSADA algorithm does healthy controls compared with those with severe glaucoma improve the SNR compared with a full bandwidth variant, the progression, as shown in table 2.31 resolution of the image is compromised; OCT-A with the These findings highlight the impact of ocular vasculature defi- SSADA algorithm has a resolution of 18 mm, rather than 5 mm 24 cits in the pathophysiology of OAG. In addition, the group for a full bandwidth algorithm. Finally, very low amounts of noted that the visual field deficits in patients with glaucoma fl fi 25 blood ow near the de ned threshold may be undetected. were more strongly associated with decreases in vessel density than decreases in structural parameters such as RNFL thick- PRESENT FINDINGS WITH OCT-A IN GLAUCOMA AND ness.31 Liu et al32 reported supporting data, showing that OTHER DISEASES decreases in the peripapillary flow index were correlated with Glaucoma is a multifactorial, progressive, chronic optic neur- visual field deficits in glaucomatous . These findings validate opathy characterised by loss of retinal ganglion cells and their fi 26 the concept that de cits in retinal microvasculature may have an axons, resulting in visual field loss. It is the second leading fi 9 important role in the development of visual eld defects in cause of blindness worldwide. While several risk factors for patients with OAG. fi glaucoma have been identi ed, such as increased intraocular OCT-A is not limited to detecting deficits in vessel density, as pressure, the mechanisms underlying optic nerve damage fi 927 observed in glaucoma studies. CNV, identi ed as a pathological remain a matter of debate. Prior studies suggest that there increase in vessel density, can be detected with OCT-A as well. may be deficits in the ocular vasculature associated with glau- – Structural OCT is also not an ideal imaging method for CNV, as coma disease progression.2 6826With nearly 80 million people expected to suffer from the disease in 2020, and an increasing prevalence of disease with age, it is important to develop improved OAG diagnostic methods.28 29 The ability to quantify Table 1 Sample values of flow index and vessel density measured alterations in ocular blood flow via OCT-A has been demon- with optical coherence tomography angiography in patients with strated by several researchers. Using OCT-A, Leveque et al glaucoma demonstrated that there is a decrease in vessel density in patients with OAG compared with healthy controls. Furthermore, the Parameters Normal Open-angle glaucoma decreases in structural parameters, such as RNFL thickness, Flow index (dimensionless) 0.160 0.104 were directly correlated with the decreases in vessel density.14 Vessel density (%) 74.2 49.1 Figure 3 demonstrates this difference in vessel density of the ONH between healthy control subjects and patients with glau- Note that both parameters are markedly decreased in patients with glaucoma compared with healthy controls.11 coma. Table 1 shows example values of the flow index and

18 Koustenis A , et al. Br J Ophthalmol 2017;101:16–20. doi:10.1136/bjophthalmol-2016-309389 Review

Table 2 Sample values of vessel density in healthy controls and patients with various degrees of glaucoma Healthy Glaucoma Mild glaucoma Moderate–severe glaucoma

Circumpapillary vessel density (%) 65.1 61.1 57.5 49.6 Whole image vessel density (%) 56.7 51.6 48.3 41.7 Average retinal nerve fibre layer thickness (μm) 99.0 87.6 78.6 65.2 Note the gradation in vessel density decreases between different severities of glaucoma.31

CNV appears similar to drusen.13 The dye-labelled fluid that structural assessments were made using structural OCT, OCT-A leaks from the newly formed immature capillaries may obscure has not been used as an imaging modality specifically in AD the presence of CNV when viewed with FA.33 OCT-A allows studies. The effectiveness of retinal laser Doppler flowmetry is CNV to be visualised clearly and quantified. For example, CNV limited to detecting deoxygenated blood and blood flowing has been demonstrated in age-related through larger vessels, making the retinal veins the more con- (AMD) with OCT-A. Compared with FA, the visualisation of the venient vessels to analyse.41 Therefore, present studies assessing retinal microvasculature was clearer, as haemorrhaging in wet retinal blood flow in patients with AD are inherently limited. AMD did not obscure the image with OCT-A.13 34 CNV has also Current findings in the retinal and choroidal microvasculature been quantified with OCT-A in other diseases, such as in multiple of the retina in patients with AD may be expanded with evanescent white dot syndrome (MEWDS), a condition where OCT-A, especially considering the documented deficits in cere- yellow-white lesions form on the retinal pigment epithelium, bral vasculature in AD and the close relationship between the – causing blurry vision.33 35 Furthermore, researchers have demon- cerebral and retinal circulations.37 46 48 Patients with AD also strated the quantification of small amounts of CNV with OCT-A, experience visual field deficits;42 a study using OCT-A may offer as low as 0.05 mm2.13 Detecting small amounts of CNV is useful insight into the aetiology of this phenomenon. for monitoring disease progression or treatment response. In addition to AD, other neurological disorders have been shown to have structural deficits in the retina compared with healthy subjects. Parkinson’s disease, schizophrenia and FUTURE APPLICATIONS OF OCT-A TO multiple sclerosis have all been shown to have some degree of – NEURODEGENERATIVE DISEASES RNFL thinning.48 52 Given that vascular deficits are seen in AD, The quantification of retinal microvasculature has been demon- it is reasonable to speculate that using OCT-A to investigate strated with OCT-A in glaucoma studies. Certain neurodegen- vascular changes in these diseases could expand our current erative diseases with vascular components have the potential to knowledge. be studied with OCT-A, as well. In general, there is value to studying the retina for neurodegenerative diseases, as the retina, CONCLUSION due to the lack of myelin, is a convenient structure to study neu- Given the high prevalence of neurodegenerative diseases such as 36 rodegeneration. Diseases such as AD have vascular patho- glaucoma and the importance of understanding the role of the 37 logical components, including within the retina. OCT-A vasculature in the pathophysiology of the disease, the quantifica- imaging of the retina of a patient with AD, as well as other neu- tion of blood flow in the retina and choroid is an important rodegenerative diseases, may provide a further understanding of development. Studies have demonstrated the presence of vascu- fi how vascular de cits play a role in the pathophysiology of the lar deficits in the ONH in patients with OAG with OCT-A. In disease. addition, quantified CNV in MEWDS and AMD has been In recent studies, researchers have used structural OCT to shown. While most of the results thus far are pilot data, which study structural changes in the retina in patients with AD. must be properly vetted and supported, OCT-A has contributed Studies have demonstrated a thinning of the RNFL, as well as a to improved imaging modalities. By providing high-resolution 38–40 thickening of the macula. These alterations have been images and vessel density quantification of the retina and the shown to occur with some gradation, with greater alterations optic nerve extending into the lamina cribrosa, understanding of occurring in patients with AD than in those with a mild cogni- the retinal vascular component in neurodegenerative diseases 41 tive impairment. In addition, groups have documented may be furthered with OCT-A. Advancing the understanding of changes in the retinal vasculature in patients with AD in studies the pathophysiology of diseases with retinal vascular compo- fl using retinal laser Doppler owmetry. Notably, the retinal veins nents allows for more effective treatment of diseases and man- 42–44 have been shown to have a decrease in lumen diameter. agement of vision and health. This observation is hypothesised to be the result of amyloid-β deposits on the internal vessel wall, thus causing the lumen to narrow.41 43 44 This theory, however, is debated.45 In response Acknowledgements Dr AH would like to disclose that he receives remuneration to the decrease in lumen diameter, the volumetric blood flow from Stemnion, Biolight, Nano Retina, AdOM, Science Based Health, Isarna 41 42 Therapeutics and Ono Pharmaceuticals for serving as a consultant. AH also holds an rate in the ocular venous return has been shown to decrease. ownership interest in AdOM, Nano Retina and Oxymap. All relationships listed Similar to glaucoma studies using OCT-A, these findings demon- above are pursuant to Indiana University’s policy on outside activities. IJ would like strate a potential relationship between deficits in the retinal vas- to disclose that she receive remuneration from Alcon, Allergan, Pfizer, Santen, culature and retinal structure. Understanding the relationships Origmed and Vittamed. between retinal vascular abnormalities and retinal neurodegen- Contributors All of the authors listed had substantial contributions to concept and eration may be important to understanding the pathophysio- design of this review, drafting and revision, as well as final approval of the paper for logical process of AD. publication. Thus far, vasculature investigations of the retina in patients Competing interests None declared. with AD have used retinal laser Doppler flowmetry.41 While the Provenance and peer review Not commissioned; externally peer reviewed.

Koustenis A , et al. Br J Ophthalmol 2017;101:16–20. doi:10.1136/bjophthalmol-2016-309389 19 Review

REFERENCES 27 Siaudvytyte L, Januleviciene I, Daveckaite A, et al. Literature review and 1 Hoh ST, Greenfield DS, Mistlberger A, et al. Optical coherence tomography and meta-analysis of translaminar pressure difference in open-angle glaucoma. Eye scanning laser polarimetry in normal, ocular hypertensive, and glaucomatous eyes. 2015;29:1242–50. Am J Ophthalmol 2000;129:129–35. 28 Quigley HA, Broman AT. The number of people with glaucoma worldwide in 2010 2 Galassi F, Sodi A, Ucci F, et al. Ocular and glaucoma prognosis: a and 2020. Br J Ophthalmol 2006;90:262–7. color Doppler imaging study. Arch Ophthalmol 2003;121:1711–15. 29 Varma R, Lee PP, Goldberg I, et al. An assessment of the health and economic 3 Galassi F, Giambene B, Menchini U. Ocular perfusion Pressure and retrobulbar burdens of glaucoma. Am J Ophthalmol 2011;152:515–22. haemodynamics in pseudoexfoliative glaucoma. Graefes Arch Clin Exp Ophthalmol 30 Holló G. Vessel density calculated from OCT angiography in 3 peripapillary sectors 2008;246:411–16. in normal, ocular hypertensive, and glaucoma eyes. Eur J Ophthalmol 2016;26: 4 Martinez A, Sanchez M. Retrobulbar hemodynamic parameters in pseudoexfoliation e42–5. syndrome and pseudoexfoliative glaucoma. Graefes Arch Clin Exp Ophthalmol 31 Yarmohammadi A, Zangwill LM, Diniz-Filho A, et al. Optical coherence tomography 2008;246:1341–9. angiography vessel density in healthy glaucoma suspects, and glaucoma eyes. Invest 5 Martínez A, Sánchez M. Predictive value of colour Doppler imaging in a prospective Ophthalmol Vis Sci 2016;57:451–9. study of visual field progression in primary open-angle glaucoma. Acta Ophthalmol 32 Liu L, Jia Y, Takusagawa HL, et al. Optical coherence tomography Scand 2005;83:716–22. angiography of the peripapillary retina in glaucoma. JAMA Ophthalmol 6 Januleviciene I, Sliesoraityte I, Siesky B, et al. Diagnostic compatibility of structural 2015;133:1045–52. and haemodynamic parameters in open-angle glaucoma patients. Acta Ophthalmol 33 Nozaki M, Hamada MD, Kimura M, et al. Value of OCT angiography in the 2008;86:552–7. diagnosis of choroidal neovascularization complicating multiple evanescent white 7 Zion IB, Harris A, Moore D, et al. Interobserver repeatability of Heidelberg retinal dot syndrome. Ophthalmic Surg Lasers Imaging Retina 2016;47:580–4. flowmetry using pixel-by-pixel analysis. J Glaucoma 2009;18:280–3. 34 Coscas GJ, Lupidi M, Coscas F, et al. Optical coherence tomography angiography 8 Weinreb RN, Harris A. Ocular blood flow in glaucoma. Amsterdam: Kugler versus traditional multimodal imaging in assessing the activity of exudative age Publications, 2009. related macular degeneration: a new diagnostic challenge. Retina (Philadelphia, Pa) 9 Leske MC, Heijl A, Hyman L, et al. Predictors of long-term progression in the early 2015;35:2219–28. manifest glaucoma trial. Ophthalmology 2007;114:1965–72. 35 Toussaint, Brain. Multiple Evanescent White Dot Syndrome. American Academy of 10 Jia Y, Tan O, Tokayer J, et al. Split-spectrum amplitude-decorrelation angiography Ophthalmology EyeWiki. Online. Updated August 25, 2015. (accessed 23 Jun with optical coherence tomography. Opt Express 2012;20:4710–25. 2015). 11 Jia Y, Morrison JC, Tokayer J, et al. Quantitative OCT angiography of optic nerve 36 Frohman EM, Fujimoto JG, Frohman TC, et al. Optical coherence tomography: a head blood flow. Biomed Opt Express 2012;3:3127–37. window into the mechanisms of multiple sclerosis. Nat Clin Pract Neurol 12 Spaide RF, Klancnik J Jr, Cooney MJ. Retinal vascular layers imaged by fluorescein 2008;4:664–75. angiography and optical coherence tomography angiography. JAMA Ophthal 37 Fischer VW, Siddiqu A, Ysufaly Y. Altered angioarchitecture in selected areas of 2015;133:45–50. brains with Alzheimer’s disease. Acta Neuropathol 1990;79:672–9. 13 Jia Y, Baily ST, Wilson DJ, et al. Quantitative optical coherence tomography 38 Ascaso FJ, Cruz N, Modrego PJ, et al. Retinal alterations in mild cognitive angiography of choroidal neovascularization in age-related macular degeneration. impairment and Alzheimer’s disease: an optical coherence tomography study. Ophthalmology 2014;121:1435–44. J Neurol 2014;261:1522–30. 14 Lévêque PM, Zéboulon P, Brasnu E, et al. Optic disc vascularization in glaucoma: 39 He XF, Liu YT, Peng C, et al. Optical coherence tomography assessed retinal nerve value of spectral-domain optical coherence tomography angiography. J Ophthalmol fiber layer thickness in patients with Alzheimer’s disease: a meta-analysis. Int 2016:6956717. J Ophthalmol 2012;5:401–5. 15 Hwang TS, Jia Y, Gao SS, et al. Optical coherence tomography angiography features 40 Coppola G, Di Renzo A, Ziccardi L. Optical coherence tomography in Alzheimer’s of . Retina (Philadelphia, Pa) 2015;35:2371–6. disease: a meta-analysis. PLoS ONE 2015;10:e0134750. 16 Takase N, Nozaki M, Kato A, et al. Enlargement of foveal avascular zone in diabetic 41 Feke GT, Hyman BT, Stern RA, et al. Retinal blood flow in mild cognitive eyes evaluated by en face optical coherence tomography angiography. Retina impairment and Alzheimer’s disease. Alzheimers Dement (Amst) 2015;1:144–51. (Philadelphia, Pa) 2015;35:2377–8. 42 Berisha F, Feke GT, Trempe CL, et al. Retinal abnormalities in early Alzheimer’s 17 Agemy SA, Scripsema NK, Shah CM, et al. Retinal vascular perfusion density disease. Invest Ophthalmol Vis Sci 2007;48:2285–9. mapping using optical coherence tomography angiography in normals and diabetic 43 Dorr A, Sahota B, Chinta LV, et al. Amyloid-β-dependent compromise of retinopathy patients. Retina (Philadelphia, Pa) 2015;35:2353–63. microvascular structure and function in a model of Alzheimer’s disease. Brain 18 Bearelly S, Rao S, Fekrat S. Anaphylaxis following intravenous fluorescein angiography 2012;135:3039–50. in a vitreoretinal clinic: report of 4 cases. Can J Ophthalmol 2009;44:444–5. 44 Kur J, Newman EA, Chan-Ling T. Cellular and physiological mechanisms underlying 19 Cole ED, Novais EA, Louzada RN, et al. Contemporary retinal imaging techniques in blood flow regulation in the retina choroid in health and disease. Prog Retin Eye diabetic retinopathy: a review. Clin Experiment Ophthalmol 2016;44:289–99. Res 2012;31:377–406. 20 Gong J, Yu S, Gong Y, et al. The diagnostic accuracy of optical coherence 45 Ho CY, Troncoso JC, Knox D, et al. Beta-amyloid, phospho-tau and alpha-synuclein tomography angiography for neovascular age-related macular degeneration: a deposits similar to those in the brain are not identified in the eyes of Alzheimer’s comparison with fundus fluorescein angiography. J Ophthalmol 2016;2016:1–8. and Parkinson’s disease patients. Brain Pathol 2014;24:25–32. 21 Ang M, Sim DA, Keane PA, et al. Optical coherence tomography angiography for 46 Brown WR, Thore CR. Review: cerebral microvascular pathology in ageing and anterior segment vasculature imaging. Ophthalmology 2015;122:1740–7. neurodegeneration. Neuropathol Appl Neurobiol 2011;37:56–74. 22 Chen FK, Viljoen RD, Bukowska DM. Classification of image artefacts in optical 47 Nakahara T, Mori A, Kurauchi Y, et al. Neurovascular interactions in the retina: coherence tomography angiography of the choroid in macular diseases. Clin physiological and pathological roles. J Pharmacol Sci 2013;123:79–84. Experiment Ophthalmol 2016;44:388–99. 48 Shih YY, De La Garza BH, Huang S, et al. Comparison of retinal and cerebral blood 23 Spaide RF, Fujimoto JG, Waheed NK. Image artifacts in optical coherence flow between continuous arterial spin labeling MRI and fluorescent microsphere tomography angiography. Retina (Philadelphia, Pa) 2015;35:2163–80. techniques. J Magn Reson Imaging 2014;40:609–15. 24 Gao SS, Liu G, Huang D, et al. Optimization of the split-spectrum 49 Hajee ME, March WF, Lazzaro DR, et al. Inner retinal layer thinning in Parkinson amplitude-decorrelation angiography algorithm on a spectral optical coherence disease. Arch Ophthalmol 2009;127:737–41. tomography system. Opt Lett 2015;40:2305–8. 50 Yu JG, Feng YF, Xiang Y, et al. Retinal fiber nerve layer thickness changes in 25 Kuehlewein L, Bansal M, Lenis TL, et al. Optical coherence tomography angiography Parkinson disease: a meta-analysis. PLoS ONE 2014;9:e85718. of Type 1 neovascularization in age-related macular degeneration. Am J Ophthalmol 51 Moreno-Ramos T, Benito-León J, Villarejo A, et al. Retinal nerve fiber layer thinning 2015;160:739–48. in dementia associated with Parkinson’s disease, dementia with Lewy bodies, and 26 Moore NA, Harris A, Wentz S, et al. Baseline retrobulbar blood flow is associated Alzheimer’s disease. J Alzheimers Dis 2013;34:659–64. with both functional and structural glaucomatous progression after 4 years. 52 Ascaso FJ, Cabezón L, Quintanilla MA, et al. Retinal nerve fiber layer thickness Br J Ophthalmol Published Online First: 13 Jun 2016. doi:10.1136/bjophthalmol- measured by optical coherence tomography in patients with schizophrenia: a short 2016-308460 report. Eur J 2010;24:227–35.

20 Koustenis A , et al. Br J Ophthalmol 2017;101:16–20. doi:10.1136/bjophthalmol-2016-309389