Purchase this article with an account.
Nicola Bernadette Quinn, Asma Aslam, Imre Lengyel, Tunde Peto, Ruth E Hogg; Refining the ultra-wide field colour fundus photography grading scheme for use in epidemiological studies. Invest. Ophthalmol. Vis. Sci. 2014;55(13):4818.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To develop a grading scheme to classify peripheral retinal lesions suitable for use in a large epidemiological study.
A grading scheme was developed based on a combination of the Moorfield’s grid (which is derived from the International Classification for age-related macular degeneration (AMD)) and the Wisconsin age-related maculopathy grading scheme (WARMGS) grid. These macula centred grids were extended with further two zones in order to capture information from the peripheral retina. Both peripheral zones were divided into eight sub-fields. A systematic review of potential retinal lesions identified the following lesions for presence/absence evaluations: choroidal neovascular membrane, geographic atrophy, floaters, naevi, retinal tears, white without pressure and haemorrhages. In this study drusen and RPE changes were evaluated. Validation and inter-grader repeatability was undertaken by two independent graders using images from the Reykjavik Eye Study. Hundred OPTOMAP P200C AF ultra-wide field laser scanning ophthalmoscope colour images of fifty participants were selected randomly from the 12-year follow-up of the Reykjavik Eye Study. Percentage of lesions located in the various zones and segments was calculated to inform on optimal design of a peripheral retinal grid.
Agreement for identification of peripheral lesions ranged from 60%-100% (Kappa 0.48-1.0). Agreement was highest for presence or absence of individual retinal lesions (97-100%, Kappa 0.92-1.0). Greatest variability was for presence of drusen in the various zones (60%-99%, Kappa 0.47-0.70). Drusen were most frequently located superiorly (68%). RPE changes were also seen most commonly in the superior sub fields (63%), in the most peripheral zones (81%) and in those sub fields closest to the mid-line (75%).
In developing a grid for grading peripheral lesions, our findings suggest that the area beyond the traditional macula grids should include zones to separate the mid periphery from the far periphery and also include sub-fields that respect the mid-line. Therefore we propose a grid that includes 16 peripheral sub-fields. This grading scheme showed low variability between graders and therefore should prove suitable for use in a large epidemiological studies.
This PDF is available to Subscribers Only