Abstract
Purpose :
The Scottish Collaborative Optometry-Ophthalmology Network e-research (SCONe) is a large retinal image repository built from images captured in optometry practices across Scotland. Retinal images captured in the community were securely linked to longitudinal, routinely collected healthcare records inside the National Safe Haven. Phase 2 of the project aims to develop accurate prediction models for ocular disease using images which pre-date diagnoses.
Methods :
The SCONe retinal image repository currently contains 330k images for 30k patients collected from 11 optometry practices across Scotland. General ophthalmic and hospital inpatient data were used to identify patients with specific diseases such as macular degeneration or glaucoma. Control patients without the disease were identified and matched on demographics and relevant co-morbidity (including diabetes and heart disease). A machine learning model will identify features which differentiate retinas which do and do not go on to show signs of disease.
Results :
The current SCONe cohort closely reflects the sex balance in the Scottish 60+ population, it includes a slightly higher proportion of people of Asian ethnicity, and a higher proportion of people from a lower deprivation index. The size of the cohort and the extent of the longitudinal data, allows us to select carefully matched controls for a range of disease case groups. Among the 2,231 people with a diagnosis of macular degeneration in their linked data, SCONe holds 2,595 image pairs captured before their first diagnosis and 9,653 after. For 2,776 people with a glaucoma diagnosis we have 11,303 image pairs captured before and 10,052 after diagnosis. These images represent potential evidence of disease progression both before and after a diagnosis is made.
Conclusions :
Active engagement with the optometry community across Scotland has enabled the SCONe repository of retinal images to be built. Matching these retinal images to healthcare datasets has allowed us to accurately identify retinal images captured both pre- and post- diagnosis of a variety of ocular and systemic conditions.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.