Article Text

Is there a role for computerised image analysis in glaucoma management?
  1. Department of Ophthalmology, The Mater Hospital, Dublin

    Statistics from

    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.

    The key elements in glaucoma diagnosis and follow up are measurement of the intraocular pressure, assessment of the optic disc, and examination of the visual field. In the 1920s, Ransom Pickard wrote about the benefits of drawing the optic disc to outline the disc and cup boundaries, placing emphasis also on the depth of the cup.1 2 In addition, he used a grid to measure the size of the cup relative to the disc, and his results of serial drawings over many years show an increase in cupping with the passage of time.3

    Armaly4 popularised the use of the cup:disc ratio in the 1960s, particularly in epidemiology studies, but the term was later adopted by clinicians and to this day is the most commonly used clinical method of describing the glaucomatous optic disc. In a classic paper by Lichter in 1976,5 the poor agreement in cup:disc ratio estimation by a number of leading glaucoma specialists clearly illustrated that it is an “inexact method of recording the status of the disc” and “is probably not reliable in checking for small disc changes”. The level of within observer agreement is considerably greater than the degree of between observer agreement at describing optic disc features.6 The weakness of this form of disc assessment is the subjective estimation by the observer of the boundary of the cup edge relative to the edge of the disc. Secondly, the size of the optic disc determines the physiological size of the cup.

    The usefulness of perimetry in glaucoma is somewhat limited by the subjective nature of the patient’s response when presented with a test target in the field of vision. This weakness is true of both manual and automated methods and for static and kinetic test strategies. The subjective aspect of the test invariably results in a measurement error which can on occasions be so large as to invalidate the test result. Consequently the signal to noise ratio can be so small that it becomes difficult to identify real change over time. Steps are being taken to reduce this source of error which will lead to more reliable results.

    As a result, long term glaucoma management is undermined by the clinician’s subjective assessment of the appearance of the optic disc and by the subjective nature of a patient’s response when performing perimetry. There is a need to have more objective methods to help make more realistic decisions about progression of the disease. The measurement of optic disc topography with image analysis techniques (initially described by Nagin and Schwartz7) offers the potential of an objective test. The recent introduction of confocal laser scanning ophthalmoscopy (CLSO) represents one of the latest development in imaging technology in glaucoma.

    However, before any computer image analysis system can be introduced into the clinical setting, it is essential that certain requirements are fulfilled. These include that the image acquired and the topography measurements are accurate, clinically meaningful, and reproducible. By accuracy is meant that the resolution of the imaging system produces an accurate contour of the structure under investigation. It is important too that the topographic variables quantified (for example, neuroretinal rim area, cup shape, retinal nerve fibre layer thickness) provide useful clinical information about the disease. Ideally we need to identify those topographic features which predict subsequent change in the visual field. It is pointless acquiring volumes of quantitative data if the measurements have little or no bearing on the disease.

    The issue of reproducibility with CLSO has two key elements—namely: (1) the ability of the imaging system to acquire the same topography contour at independent imaging sessions when no anatomical change has occurred8 9 (which is related to the issue of accuracy of image acquisition described above); and (2) the reliability of the operator and imaging system to generate the same topographic measurements on repeated measurements of the “same image”.10-12 In this issue of the journal (p 14), Geyer and colleagues examine reproducibility of topography measurements with the TopSS CLSO in a group of glaucoma patients whose optic discs were imaged at two independent sessions 30 minutes apart. Using a number of statistical tests to examine reproducibility, the study identified three (out of 14) disc variables with a high degree of reproducibility which were suggested to be sufficiently robust topography features to be useful for long term monitoring of the optic disc. These features were cup depth, cup volume, and cup area halfway between the surface of the disc and the floor of the cup. We do not know if these same variables are equally reproducible in non-glaucomatous optic discs—for example, in ocular hypertension, or over longer time intervals such as 6 months apart. We do know that previous work shows a significant correlation between cup volume and visual field loss in glaucoma,13 14 and therefore at least one of these variables might provide valuable clinical information. Secondly, cup shape is capable of detecting early glaucomatous visual field loss with a high degree of sensitivity and specificity.15

    In addition to the above prerequisites, instruments such as the CLSO (and other glaucoma image analysis techniques such as nerve fibre layer polarimetry16 and optical coherence tomography17) should generate data which discriminate between two or more groups, and measure change over time. The challenge to those involved in this area of clinical research is to demonstrate the usefulness of this type of objective assessment of the optic disc and retinal fibre layer during long term follow up over 5 to 10 years. It remains to be seen whether any of the markers of cup shape identified by Geyer et al turns out to be sufficiently robust in terms of accuracy and reproducibility, and also clinically meaningful in detecting disc change before field progression during glaucoma follow up. Only then can we consider the use of such expensive equipment in routine glaucoma management.


    Linked Articles