Article Text

Download PDFPDF

En face optical coherence tomography angiography for corneal neovascularisation
  1. Marcus Ang1,2,
  2. Yijun Cai2,
  3. Shahab Shahipasand2,
  4. Dawn A Sim2,
  5. Pearse A Keane2,3,
  6. Chelvin C A Sng1,2,4,
  7. Catherine A Egan2,
  8. Adnan Tufail2,
  9. Mark R Wilkins2
  1. 1Singapore National Eye Center, Singapore Eye Research Institute, Singapore, Singapore
  2. 2Moorfields Eye Hospital NHS Foundation Trust, London, UK
  3. 3Institute of Ophthalmology, University College London, London, UK
  4. 4Department of Ophthalmology, National University Hospital, Singapore, Singapore
  1. Correspondence to Dr Marcus Ang, Singapore National Eye Centre, Singapore Eye Research Institute, 11 Third Hospital Avenue, Singapore 168751, Singapore; marcus.ang.h.n{at}, marcusahn{at}


Background/aim Recently, there has been an increasing clinical need for objective evaluation of corneal neovascularisation, a condition which cause significant ocular morbidity. We describe the use of a rapid, non-invasive ‘en face’ optical coherence tomography angiography (OCTA) system for the assessment of corneal neovascularisation.

Methods Consecutive patients with abnormal corneal neovascularisation were scanned using a commercially available AngioVue OCTA system (Optovue, Fremont, California, USA) with the split-spectrum amplitude decorrelation angiography algorithm, using an anterior segment lens adapter. Each subject had four scans in each eye by a trained operator and two independent masked assessors analysed all images. Main outcome measures were scan quality (signal strength, image quality), area of neovascularisation and repeatability of corneal vascular grade.

Results We performed OCTA in 20 patients (11 men, 9 women, mean age 49.27±17.23 years) with abnormal corneal neovascularisation. The mean area of corneal neovascularisation was 0.57±0.30 mm2 with a mean neovascularisation grade of 3.5±0.2 in the OCTA scans. We found the OCTA to produce good quality images of the corneal vessels (signal strength: 36.95±13.97; image quality score 2.72±1.07) with good repeatability for assessing neovascularisation grade (κ=0.84).

Conclusions In this preliminary clinical study, we describe a method for acquiring angiography images with ‘en face’ views, using an OCTA system adapted for the evaluation of corneal neovascularisation. Further studies are required to compare the scans to other invasive angiography techniques for the quantitative evaluation of abnormal corneal vessels.

  • Cornea
  • Neovascularisation
  • Imaging
View Full Text

Statistics from


Corneal neovascularisation may develop from a wide variety of insults to the cornea—from chemical injury or contact lens overuse,1 to common infections such as herpes and trachoma.2 Visual impairment may occur secondary to the persistent inflammation, corneal oedema, lipid deposition and scarring.3 Further to this, morbidity may be caused by disruption to the immune privilege of the cornea, increasing the risk of rejection should corneal transplantation be required in these eyes.4 As new treatment modalities arise for treating corneal neovascularisation,5 the objective evaluation of abnormal corneal vessels is becoming increasingly important to evaluate these interventions.6

Fluorescein angiography (FA) and indocyanine green angiography (ICGA) techniques have already been described for the anterior segment.7 Clinical applications include preoperative localisation of corneal neovascularisation for targeted intervention,8 monitoring treatment response using vascularisation area9 or the diagnosis and prognostication of corneoscleral inflammation.10 ,11 However, these techniques are invasive, time-consuming and carry the risk for rare, but serious adverse reactions.12 ,13 Recently, advancements in optical coherence tomography (OCT) techniques have led to the ability to image blood vessels, by detecting phase variations or changes in reflectivity.14 At the same time, OCT systems are now able to rapidly acquire high-resolution scans over a three-dimensional (3D) volume to reconstruct coronal sections (C-scans), producing an ‘en face’ view of the scanned area.15

We had previously described how OCT angiography (OCTA) originally designed for the retina was adapted for normal anterior segment vasculature with substantial consistency.16 While OCTA techniques have been used to evaluate various retinal vascular pathologies,14 ,17 ,18 to our knowledge, its role has not been reported specifically for imaging corneal neovascularisation.19 ,20 Therefore, we conducted a pilot study to evaluate this novel split-spectrum amplitude decorrelation angiography technique to the image areas of abnormal corneal vasculature development secondary to various pathologies. We also studied the use of the ‘en face’ C-scan function to examine various corneal lesions and their associated aberrant vessels.

Materials and methods

We conducted a cross-sectional, observational study in subjects with corneal neovascularisation at Moorfields Eye Hospital from 1 December 2014 to 30 April 2015. Our study followed the principles of the Declaration of Helsinki, with approval obtained from our local ethics committee. All patients were first clinically evaluated with a digital slit lamp (Topcon ATE-600, Topcon) and had the areas of corneal neovascularisation identified (MA, MRW), with photographs taken using standard diffuse illumination (Nikon D1x digital camera, Nikon). A trained operator then performed all OCTA scans using the AngioVue system (Optovue, Fremont, California, USA), which uses the split-spectrum amplitude decorrelation angiography algorithm.21 For this study, we used the long corneal adaptor module to produce four 6×6 mm volume scans in each area of corneal neovascularisation. Each scan was performed with axial resolution of 5 μm, transverse resolution of 15 μm, beam width of 22 μm with a light source centred on 840 nm at up to a depth of 2.3 mm. B-scans were obtained comprising 304×304 A-scans captured at 70 000 scans per second that constructs a 3D scan cube in approximately 3–4 s.20 Consecutive B-scans captured at a fixed position were registered and compared with the next location to calculate the decorrelation for detection of blood flow.

All scans were automatically processed to reduce motion artefacts by the internal software (ReVue V.2014.2.0.15). Next, we analysed OCT B-scans (conventional, transverse sections) and C-scans (‘en face’ scan in the coronal plane) with their accompanying angiograms. We used the built-in ‘Angioflow’ function to highlight the region of interest (default radius 1.25 mm, threshold 0.05). The quality of the scans were then assessed by two independent masked evaluators (YC, SS) using a recognised score, that is, 0: no vessel discernible; 1: poor vessel delineation; 2: good vessel delineation; 3: very good vessel delineation; 4: excellent vessel delineation.7 All images with good delineation were also graded the corneal neovascularisation based on the extent of the corneal neovascularisation modified from a previously described system,22 that is, 1: engorged limbal vessels not extending onto the cornea; 2: new vessels extend into cornea but <1 mm from the limbus; 3: new vessels extend ≥1 mm beyond the limbus but not involving the paracentral cornea (<3 mm from the limbus); 4: new vessels extend ≥3 mm from the limbus and into the paracentral cornea. Images were then exported from the system as a portable network graphics image file into the Image J software (V.1.49p, National Institutes of Health, Bethesda, Maryland, USA).23 The Image J software ‘adjust threshold’ function was applied to highlight blood vessels and reduce the surrounding noise, before the area of neovascularisation was calculated adapted from a previously described method (figure 1).24 We also recorded the signal strength of all OCTA scans to ensure that all scans were of ‘adequate’ quality as determined by the AngioVue system (Optovue).

Figure 1

Illustration of optical coherence tomography angiography (OCTA) image analysis. (A) The area of corneal neovascularisation in the inferotemporal aspect of the cornea was first identified and photographed with the slit-lamp camera. (B) The OCTA scans were performed and analysed using the built-in ‘Angioflow’ function (ReVue V.2014.2.0.15) to highlight the abnormal vessels in the region of interest (default radius 1.25 mm, threshold 0.05). (C) The images were then exported from the system as a portable network graphics image file into the Image J software (V.1.49p, Wayne Rasband; National Institutes of Health, Bethesda, Maryland, USA). (D) The ‘adjust threshold’ function in Image J was used to highlight the blood vessels and reduce the surrounding noise to export the resulting vascular tree as a binary image for analysis.

Statistical analysis

We analysed all scan images obtained for repeatability for image quality scores and vascular grade using the kappa coefficient (κ) value, where κ≤0.2, was considered slight, 0.21–0.40 weak, 0.41–0.6 moderate, 0.61–0.8 substantial and 0.81–1.0 considered ‘almost perfect’ in agreement, respectively.25 Statistical analysis also included descriptive statistics, where the mean and SD (±SD) were calculated for the continuous variables and outcome measures. As this was a pilot study, no formal sample size calculations were performed, as our main aim was to assess the feasibility of using this novel imaging system for corneal neovascularisation in a reliable manner. A p value <0.05 was considered statistically significant for comparisons between two quadrants. Statistical Package for the Social Sciences V.17.0 (SPSS, Chicago, Illinois, USA) was used to analyse the data.


In this observational study, we performed OCTA scans in 20 patients (11 men, 9 women, mean age 49.27±17.23 years) in the affected eyes (right eye=10) with abnormal corneal neovascularisation. In order to increase the external validity of this pilot study, we included patients with a variety clinical conditions who developed abnormal corneal and limbal vasculature that are related to previous herpetic keratitis (n=10), corneal graft, that is, penetrating keratoplasty and deep anterior lamellar keratoplasty (n=4), bacterial keratitis (n=3) and limbal stem cell deficiency (n=3). The mean area of abnormal corneal vessel development within each region of interest was 0.57±0.30 mm2 with a mean neovascularisation grade of 3.5±0.2. We found that the OCTA scans in our study were of high quality (mean signal strength: 36.95±13.97; mean image quality score of 2.72±1.07) and produced with good repeatability (κ=0.84) and interobserver agreement (κ=0.95).

Figure 2A,B illustrates a 66-year-old female who had suffered from recurrent herpetic keratitis for 3 years and developed lipid keratopathy. The OCTA system was able to identify the larger ‘feeder vessels’, as well as the smaller vessels that were not visible on slit-lamp examination, which were deep within the stroma and obscured by the lipid deposition and scar. Abnormal vessels are visible in the localised area of limbal stem cell deficiency in the right eye of a 64-year-old male with ocular rosacea and marginal keratitis (figure 2C). The OCTA provided a measure of the size and depth of lesion as well as the area of neovascularisation (figure 2D). The scans also demonstrated destruction of the normal limbal vasculature architecture adjacent to the area of corneal scarring, suggesting that the area of damage is more extensive than on clinical examination.

Figure 2

Clinical applications of optical coherence tomography angiography. (A) Lipid keratopathy from previous herpetic keratitis. (B) Abnormal ‘feeder’ vessels and smaller capillaries visualised in the optical coherence tomography angiography images. Note that of the two larger vessels the one of the left has a brighter signal suggesting greater flow in the vessel. (C) Marginal keratitis and limbal stem cell deficiency from ocular surface disease. (D) Corneal neovascularisation and adjacent destruction of limbal vasculature delineated on optical coherence tomography angiography.

We used the ‘en face’ function that provides information in several planes at a specific location on the cornea to assess the location, depth and size of the corneal pathology with the abnormal corneal vessels. Figure 3 shows the ‘en face’ OCTA images of the eye with lipid keratopathy photographed in figure 2A. These are sequential scans taken at 50 µm intervals, illustrating that the deep stromal scar and lipid deposition extend to the Descemet membrane. The associated feeder vessel penetrates from superficial to mid-stroma. Smaller capillaries extend deeper into the area of lipid deposition in the cornea. OCTA imaging provides valuable information for preoperative planning, such as the location and depth of the feeder corneal vessel for diathermy, as well as the depth and size of the corneal lesion in preparation for anterior lamellar keratoplasty.

Figure 3

En face optical coherence tomography angiography views simultaneously demonstrate the corneal pathology and the extent of abnormal vasculature. The C-scans were automatically determined by two boundaries parallel to the surface, where the default thickness (50 μm) between the boundaries may be manually modified. (A) Epithelium-50 μm. (B) 150–200 μm. (C) 250–300 μm. (D) 450–500 μm.

In our series, we performed OCTA in eyes with bacterial keratitis to study its potential clinical use. A 34-year-old contact lens wearer developed bacterial keratitis near an area of previous scarring, and developed secondary neovascularisation (figure 4A). The associated abnormal vessels within the infiltrate are clearly demonstrated by the OCTA (figure 4B). Of note, there is also a brighter signal from the limbal vessels adjacent to the infiltrate as seen on the OCTA scan. Figure 4C shows a slit-lamp photograph of a 30-year-old male who underwent a deep anterior lamellar keratoplasty for keratoconus and defaulted topical steroid treatment for 3 months, subsequently presenting with graft rejection and extensive corneal neovascularisation. The stromal vessels are evident with extensive infiltration deep into the donor cornea stroma detected by the OCTA scans and have a brighter signal in the thick central vessels compared with thinner vessels in the periphery (figure 4D).

Figure 4

Flow within vessels may be detected by the optical coherence tomography. (A) Bacterial keratitis with secondary neovascularisation. (B) A brighter signal from the limbal vessels adjacent to the infiltrate may indicate active inflammation. (C) Graft rejection with extensive vascularisation of the cornea. (D) Vessels invading donor stroma with a bright signal indicate active flow within the vessels.


In this preliminary study, we describe a novel application of OCTA technology designed for retinal scans, with the aid of an anterior segment lens adapter, to achieve non-invasive imaging of corneal neovascularisation. Our technique produced OCTA scans with good signal strength and repeatability in terms of image quality (κ=0.84). Furthermore, we described an adapted technique using the built-in software within the OCTA system (designed for retinal blood flow analysis) to obtain measurements of the corneal vessels, which could allow users of the same system to reproduce this analysis for comparative studies in the future. As OCTA is still in its infancy in the realm of clinical use, and initially intended for retina vasculature assessment, the purpose of this pilot clinical study was to design a standardised scanning protocol and demonstrate its repeatability in terms of image quality. We also showed the potential of OCTA to successfully delineate pathological corneal vessels in various conditions such as infectious keratitis, lipid keratopathy and corneal graft vascularisation. In a busy clinical setting, the speed of acquisition was a key advantage of the OCTA system that made it a useful tool for acquiring images of corneal neovascularisation. Each non-contact scan had a rapid acquisition time of 3–4 s that produced a 3D reconstruction of the area of interest in the cornea with its associated abnormal vessels.

Our study described another useful function of this current OCTA system—its ability to optically ‘dissect’ various layers of the cornea with high-resolution scans and visualise vessel flow within each layer. The ‘en face’ mode produces C-scans that are orientated to the frontal plane and gives transverse slices of a corneal lesion and neovascularisation—which is more intuitive to untrained operators, as opposed to the traditional cross-sectional OCT ‘B-scans’ that resemble histology slides, and are more useful for studying anterior segment anatomy.26 ,27 The coronal reconstruction allows for an instant overview of the corneal pathology and changes to the length, calibre and area of corneal neovascularisation, which was previously not possible with B-scan OCT views.28 The ‘en face’ scans give information on the depth of invading corneal vessels as well as the associated scar or infiltrate, which is valuable when planning for surgical procedures such as diathermy or anterior lamellar keratoplasty. Moreover, the simultaneous view of ‘en face’ and OCTA modes allows for the potential for objective localisation and monitoring or the corneal pathology by pinpointing the area of interest and varying the depth of the coronal section.

As with every new imaging system, clinicians should understand its limitations and the confines of scan interpretation. First, image distortions may occur due to patient movement, inclinations of the scanning plane relative due to the corneal surface, with fragmentation as deeper scans are obtained. Thus, we recommend performing multiple scans to obtain a good quality image, which patients are usually able to tolerate, as each non-contact scan only requires 3–4 s to complete. Second, though the current OCTA system used in this study comes with an in-built motion correction for ocular saccades or micro-movements, it does not carry an eye-tracking system with registration, which is required for comparisons in follow-up scans. Finally, there is currently a limited image resolution with a relatively small field of view (up to 8×8 mm in this current system). Moreover, OCTA may not detect all vessels with minimal flow or blocked signal from corneal opacities, and is unable to demonstrate vessel leakage or sequence of flow within vessels. Nonetheless, our initial experience with the AngioVue OCTA system revealed a relatively easy learning curve, and further software development specifically for the anterior segment may increase the field of view and quality of scans.

Our observations from this early work suggests that OCTA may provide a promising non-invasive imaging alternative to FA or ICGA, both of which have been well-described for detecting areas of vascularisation in the anterior segment.11 ,29 While invasive angiography allows for the detection of dye leakage that suggests inflammation or activity,9 it does not allow for the appreciation of the depth of the vessels in relation to the associated pathology or the cornea.30 The advantage of this OCTA system is its ability to swiftly obtain scans that outline the area of cornea neovascularisation, and simultaneously determines its relationship to the associated corneal pathology. As this is a pilot clinical study describing a relatively new imaging system adapted for the anterior segment, we recognise the limitations of our observational study in a small number of patients with corneal neovascularisation. Nevertheless, we present promising results suggesting that this non-invasive OCTA system has the potential to produce consistent images of the abnormal corneal vessels, which may have useful clinical applications in the future.31 ,32 Further prospective studies are required to directly compare OCTA with current angiography techniques for objectively visualising and quantifying corneal vessels for the diagnosis and monitoring of treatment in patients with abnormal corneal neovascularisation.


We acknowledge Becky MacPhee (Moorfields Eye Hospital Imaging Specialist), who performed all the imaging, for her contribution to this study.


View Abstract


  • Twitter Follow Mark Wilkins at @wilkoman

  • Contributors All authors met the ICJME criteria: substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; drafting the article or revising it critically for important intellectual content; and final approval of the version to be published.

  • Funding PAK, DAS, CAE, AT and MRW have received a proportion of their funding from the Department of Health's NIHR Biomedical Research Centre for Ophthalmology at Moorfields Eye Hospital and UCL Institute of Ophthalmology.

  • Competing interests AT is a member of the Optovue Inc. users group that received non-financial, technical support.

  • Patient consent Obtained.

  • Ethics approval Moorfields Eye Hospital Ethics Review Board.

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

  • Data sharing statement Any additional unpublished data may be obtained from the Corresponding Author.

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.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost Jonas