Statistics from Altmetric.com
Optical coherence tomography (OCT) has recently been adapted to enable rapid, non-contact detailed imaging of the anterior segment of the eye. Anterior segment optical coherence tomography (AS-OCT) was originally developed for use in refractive surgery but the ability to provide cross-sectional images of the angle and its key anatomical landmarks makes it potentially useful in glaucoma research and clinical management of primary angle closure. Objective recordings of the anterior chamber angle can then be analysed by qualitative or quantitative means to provide an assessment of whether the angle is open or closed and degree of angle width/opening.
Currently qualitative categorical outcomes (angle open or closed) are used to define the presence or absence of angle-closure on AS-OCT images. Quantitative measurements of angle width are useful for comparing angle configuration before and after interventions including laser iridotomy or cataract surgery and for reporting these outcomes in a quantifiable way. Parameters that are used for measuring angle width on anterior segment images include the angle opening distance (AOD) at a fixed distance anterior to the scleral spur, the angle recess area (ARA) and the trabecular iris space area (TISA).
In this edition of the journal, Console et al1 report the use of new image analysis software for AS-OCT (see page 1612). This new software, like the existing inbuilt image analysis software packages, requires identification of the scleral spur as the starting-point for taking measurements. Accurate identification of scleral spur is turning out to be a significant problem for image analysis and limits the applications of the current devices. In a recent study by Sakata et al, it was not possible to identify the scleral spur in 28% eyes on images taken using the Visante AS-OCT.2 In the paper published here, the authors report the effect of interobserver reproducibility in scleral spur identification on variation in quantitative measures of angle width. They estimate that up to 50% of variation between observers in measurements of angle area (ARA, TISA) and 10% variation in linear measurements (AOD, anterior chamber angle) is due to poor reproducibility of the scleral spur location. They also report (as did Sakata’s paper) that in patients with narrow angles, the scleral spur is even more difficult to identify than in those with open angles. Obviously this is the group of glaucoma patients in which AS-OCT imaging is going to be most useful, and so future developments and refinements of image quality will be essential for better performance of this technology.
Studies comparing AS-OCT with gonioscopy for diagnosing appositional angle-closure have shown AS-OCT to overclassify individuals as having closed angles.3 4 Possible explanations for these findings include the effect of light on the pupil and therefore angle widening, when doing gonioscopy, artefacts of angle viewing through the gonio-lens and inadvertent corneal indentation. Potentially the most important reason for the disparity between the two methods of angle examination is that the definitions of angle-closure using AS-OCT and gonioscopy differ. A closed angle is defined by gonioscopy as non-visibility of posterior trabecular meshwork. When using the currently available AS-OCT devices it is not possible to confidently identify the trabecular meshwork and Schwalbe line. The definition of closure using AS-OCT is any contact between peripheral iris and angle wall anterior to scleral spur. Thus, even if there is only a short area of contact between iris and angle wall, these cases are defined as having closed angles. This probably goes some way to explaining the poor specificity (ability to correctly identify open angles as open) of the AS-OCT (62.9%) when compared with the scanning peripheral anterior chamber analyser (76.6%) and intraocular lens-master (77.7%) in evaluating these different methods for community screening for angle closure.4 Incorporating analysis of the peripheral iris profile and physiological response to dark–light stimuli5 with AS-OCT may improve the specificity of AS-OCT in future angle-closure screening studies.
Currently, difficulty identifying angle structures is due to poor image quality (low signal-to-noise ratios). Factors such as scan resolution, motion artefacts and scan speed can limit visibility of key anatomical landmarks. Image-averaging software designed for retinal imaging has recently been employed in AS-OCT image analysis to try and improve image quality and overcome some of the problems discussed in Console et al’s paper. By averaging three to eight images, a higher-resolution image is obtained, enabling easier identification of scleral spur. When used in the slit-lamp OCT (Heidleberg Engineering), the interobserver agreement for localisation of scleral spur improves significantly (kappa improves from 0.55 to 0.73).6 If this and other improvements are made to image quality and analysis software, it should be easier to localise fundamental landmarks like scleral spur and maybe other structures such as the Schwalbe line.
One of the challenges of AS-OCT is devising a classification for angle closure or risk of angle closure using this technology. At present, the definitions of closure are those described above with their limitations. There is no current consensus on what defines a narrow angle (ie, with no apposition but little space between iris and angle wall). A linear or area-based measurement parameter such as AOD or TISA could be used to define those at risk of or with angle-closure who would potentially benefit from treatment. Arriving at appropriate cut-off measurements of these parameters to define narrow or closed angles will require comparison with the current reference standard of gonioscopy, analysis of cross-sectional data to look at the association between measures of angle width and evidence of disease (raised intraocular pressure, peripheral anterior synechiae or acute angle closure) and longitudinal follow-up of patients to determine the natural history. Console et al’s paper published here suggests that a linear measure such as the AOD at 500 µm anterior to scleral spur may be subject to less interobserver variability than other measures and may be the most appropriate parameter to use for angle measurement. However, before getting to this stage, image quality and analysis, which is dependent on localisation of angle landmarks by either an observer or the software, has to improve sufficiently to allow reliable and reproducible measurements of angle morphology. High-resolution imaging such as high-definition OCT where the trabecular meshwork and Schwalbe line are visible may be the solution once it is adapted to be user-friendly for examining patients.7 Maybe then, AS-OCT can fulfil its potential for detecting angle closure in at-risk populations and monitoring the effect of interventions.
Competing interests: None.
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.