Author: Tunde Peto (United Kingdom)
Co-authors: Barbra Hamill, Graders NetwORC UK, Savita Madhusudhan, Konstantinos Balaskas, UKBB Eyes and Vision Consortium, Paul Foster
Purpose
UK Biobank is a biomedical database comprising lifestyle, biomedical and genetic data from over 500,000 UK participants aged 40-69. Close to 70 000 participants had ophthalmic imaging of colour fundus photography and optical coherence tomography (OCT) at baseline. The purpose of this current abstract is to provide human grading results in order to describe the most common characteristic features of AMD on both colour and OCT imaging in UKBB. Imaging took place undilated during the participant’s baseline visit at the UKBB site. Altogether, 68527 participants’ image sets were available for grading by the NetwORC UK’s trained and accredited graders.
Setting/Venue
A single colour photograph of the optic disc and the macula were acquired using Topcon 3D OCT-1000 Mark II system, with a field angle of 45°, a digital zoom (2X, 4X), scanning range 6×6 mm, 4.5×4.5 mm, 3×3 mm and scanning speed of 18,000 A-scans per second. The horizontal and longitudinal (depth) resolutions were ≤20 µm and 5–6 µm, respectively. The same equipment obtained the spectral domain 3-dimensional OCT using a 3D macular volume scan (512 horizontal A-scans/B-scan; 128 B-scans in a 6×6 mm raster pattern).
Methods
Colour and OCT images were analysed by trained and certified NetwORC UK reading centre graders answering pre-determined questions to capture relevant AMD features visible on the images. These data were captured in a study specific grading form linked to the study database. Image sets of each participant were graded independently by one grader; the grader scrutinized both image modalities for features characteristic of AMD. If the images required second review, the grader requested adjudication of the encounter by a senior grader or clinician. Quality assurance (QA) was carried out by provision of rigorous training and certification of all graders, senior graders and clinicians before grading began; by double grading a subset of images; providing regular clinician review sessions and training exercises. In addition, quality assurance grading was undertaken on 1:20 randomly selected images that were re-introduced into the grading queue without any prior warning so exact agreement, intra- and intergrader reliability could be calculated. The UK Macula Society grant enabled this project.
Results
Approximately 99% of colour images and 99% of OCTs were available for the right eye and left eye. Of these available images, 90% of colour and 99% of OCT images were of gradable quality. Grading revealed approximately 10% of right eyes and 9% of left eyes had drusen only phenotype; an additional 0.5% of right eyes and 0.4% left eyes had drusen with pigmentary changes. Within these two categories, approx. 0.1% of right eyes and 0.13% of left eyes had reticular drusen identified on colour imaging. End-stage disease (geographic atrophy and late AMD - active or inactive) was seen in approximately 0.06% of right eyes and 0.07% of left eyes. In addition, 5% of right eyes and 6% of left eyes, with no apparent AMD features on colour imaging, were found to have at least one druse and /or sub-retinal drusenoid deposits on OCT imaging. The prevalence of features characteristic of AMD was less than 0.05% in both eyes.
Conlusions
The cohorts identified with characteristic changes for age related macular disease will form the basis of further analysis for its genetic and lifestyle determinants. The importance of the grading results presented here is that over 69000 participants’ image sets have been graded in details for characteristics of age related macular degeneration. As the UKBB reaches more subjects for follow-up and linkage, this set of grading will become an important dataset for research purposes, and might provide ground-truth for further analysis. The dataset provides an unparalleled opportunity to examine eye disease in this cohort and will provide the much needed baseline data for the follow-up studies that are to be conducted by UKBB.
Financial Disclosure
None relevant to current abstract. We thank the Macula Society for the grant that enabled this grading activity.
Comments
-
