Developing tools to better diagnose and monitor birdshot uveitis

Research details

  • Type of funding: Project Grant
  • Grant Holder: Professor Alastair Denniston
  • Institute: University Hospitals Birmingham NHS Foundation Trust
  • Region: West Midlands
  • Start date: August 2019
  • End Date: January 2023
  • Priority: Early diagnosis
  • Eye Category: Ocular inflammatory
Brief lay background

Birdshot chorioretinopathy (birdshot uveitis) is a rare type of uveitis (eye inflammation) affecting 400-500 people in the UK. People with birdshot have inflammation in the retina and the middle layer of the eye, the choroid. If left undiagnosed or under-treated, this progressive disease can lead to sight loss.

What problem/knowledge gap does it help address

Current treatment for birdshot uses drugs to suppress the immune system. But a major challenge is accurately measuring inflammation in the eye to guide appropriate use of these medicines. There is a lack of widely available, sensitive, objective disease markers which hinders monitoring of disease and treatment.

Current tests for diagnosing birdshot are also time-consuming and uncomfortable for patients. In addition, those with early-stage birdshot often report only mild symptoms and are frequently diagnosed late.

Some patients are therefore under-treated, allowing the disease to progress, whereas others are overtreated and at higher risk of side effects such as cataracts, increased pressure in the eye and infections.

Aim of the research project

This project will determine if there is ‘hidden’ information within patient eye scans that is unrecognizable to the human eye but can be detected by a computer. This information could then be used to diagnose birdshot at an early stage and help guide treatment.

Key procedures/objectives
  1. Establish a Birdshot Image Bank – a database of patient eye scans that can be used for research.
  2. Use computer algorithms to measure features on eye scan images from the UK Birdshot Registry and Biobank.
  3. Test whether these identified features can be used to distinguish between people with and without birdshot, or between those with early or late-stage disease.
  4. Determine whether a computer using artificial intelligence is as good as a pathologist at spotting the hallmarks of birdshot from patient eye scans.
  5. Develop an algorithm specifically for birdshot that can diagnose patients earlier and help predict the course of their disease.
Potential impact on people with sight loss

If successful, this project could: 

  1. Lead to a new way to diagnose birdshot earlier using standard scans in hospitals and high-street optometrists. 
  2. Improve monitoring of people who already have birdshot uveitis, avoiding uncomfortable procedures and potentially preventing them from losing their sight.
  3. Make it easier to detect and monitor levels of inflammation within the eye, which is crucial to determine appropriate immunosuppressant treatment.