Deep Learning for Automatic Screening of Diabetic Retinopathy
Research type
Research Study
Full title
Deep Learning for Automatic Screening of Diabetic Retinopathy
IRAS ID
198668
Contact name
Yalin Zheng
Contact email
Clinicaltrials.gov Identifier
5126, Trust RD&I Number
Duration of Study in the UK
2 years, 11 months, 31 days
Research summary
Diabetic retinopathy (DR), a diabetes related eye problem, is the leading cause of blindness in the working age population in the western world. To reduce the risk of blindness all people with diabetes in the UK are invited for digital photographic screening of the retina every year. Trained graders then diagnose and monitor DR by manually grading the colour retinal photographs. According to the NHS Diabetic Eye Screening Programme, there are currently over 2.5 million people with diabetes in England alone and over 1.9 million people were screened during 2011-12 at a cost of around £60 million. The number of patients is rising rapidly in the UK (approx. 5% per year) and around the world. Manual grading requires extensive training and tight quality assurance to reduce the risk of error. Patients can wait up to six weeks before receiving a result.
New technology can help healthcare provision. The development of computer programs that can grade images like human graders will help reduce cost and improve patient care in the UK and worldwide, in particular in developing countries where resources are limited. With continuous development these computer programs will continue to accrue benefits long into the future. There are some previous programs developed for the purpose of DR screening, but their functions are limited and performance is not as good as needed for screening DR.
The aim of this project is to build new “deep learning” computer models that can provide at point of care full disease retinopathy grading results for people with diabetes. We will develop the programs on publically available datasets, and validate them on a large real image dataset collected from patients participating in the Liverpool Diabetic Eye Screening Programme.
REC name
West Midlands - Edgbaston Research Ethics Committee
REC reference
16/WM/0054
Date of REC Opinion
22 Jan 2016
REC opinion
Favourable Opinion