Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

background

The recently completed LENS trial demonstrated that fenofibrate therapy reduced progression of diabetic retinopathy compared to placebo in participants with mild diabetic eye disease (https://evidence.nejm.org/doi/full/10.1056/EVIDoa2400179). This result was based on NHS Scotland’s Diabetic Eye Screening programme’s ordinal grading of retinal images. The trial included 1151 participants for whom the LENS investigators, based in the Clinical Trial Service Unit (CTSU), hold baseline retinal images and post-randomisation retinal images for 95% of trial participants. The availability of these images provides the opportunity to gain insights into the effects of fenofibrate therapy on the retina.

Identification of ocular biomarkers from multimodal imaging, termed oculomics (https://doi.org/10.1167/tvst.9.2.6), and the growth of Artificial Intelligence (AI) methods to automate extraction of ocular biomarkers has opened opportunities to seek associations between ocular characteristics and disease (including systemic diseases). More recently, AI foundation models for retinal imaging (https://doi.org/10.1038/s41586-023-06555-x) have been shown to identify signs of health conditions (both eye diseases and systemic disorders) in retinal images and to expedite diagnoses of disease by learning generalisable representations from retinal images. This project seeks to leverage these AI discoveries to propose AI-derived endpoints for application in clinical trials. 

RESEARCH EXPERIENCE, RESEARCH METHODS AND TRAINING

A selection of LENS retinal images will be labelled with features of retinopathy and maculopathy (e.g. micro-aneurysms, exudates), a step which will likely be conducted by experienced graders in the NHS. The candidate will use these labelled images to train the retinal imaging model for analysis of all 9000 LENS retinal images. The candidate will then conduct statistical analyses to determine the effects of fenofibrate on these retinal lesions. Further analyses will be conducted in important subgroups (e.g. type 1 vs. type 2 diabetes). The candidate will also analyse effects on non-diabetic lesions (e.g. drusen) to explore the potential for fenofibrate to reduce the progression of macular degeneration. Given the renewed interest in repurposing this cheap medication to treat diabetic retinopathy, we anticipate that the results will be published within leading diabetes and AI/ML and medical imaging journals.

FIELD WORK, SECONDMENTS, INDUSTRY PLACEMENTS AND TRAINING

The candidate will work with supervisors based in both CTSU and the BDI. The supervisors already hold all the relevant images and data for the project. Training required to conduct the work, including image analysis and statistical analysis, will be provided within CTSU and the BDI as needed. Opportunities for external placements can be explored.

prospective student

We welcome applications from applicants with training and/or experience in any of ophthalmology, machine learning, randomised trials, statistics.

Supervisors