Uveitis describes a group of conditions characterised by eye inflammation. It is a major cause of disease, accounting for 10-15% of total blindness in the developed world.
One modern technology used for assessing inflammatory disease in the eye is called optical coherence tomography (OCT), which provides a simple and non-invasive way to visualise the deep tissue of the retina.
Researchers aim to analyse the diseased eyes of animals at known stages of uveitis, and use modern methods of computer aided image processing to develop our understanding of how to interpret changes in tissue texture and appearance. This information can be applied to human patients to make more reliable predictions about the stage of the disease and its likely outcome.
Researchers will investigate two ways of using images in the assessment of disease. The first of these is FUSION, which describes the computer aided superposition of different pictures of the same object. By combining images, subtle changes in tissues that are difficult to distinguish with the naked eye could be revealed. To achieve image fusion, pictures from OCT will be combined with those obtained from microscopy to identify features that correspond to underlying disease associated changes.
The second way is CLASSIFICATION. This process exploits the power of machine learning to uncover subtle reproducible patterns that indicate the underlying state of the tissues. Hundreds of images of healthy eyes will be used to allow the computer system to learn about normal tissue. Material obtained from animals with uveitis will then be introduced. Some of these will be generated specifically for this project, but access to a large repository of images that have already been collected in other experiments can also be used to validate the trained computer system to measure how well it can detect the changes caused by disease.
Improved quantitative methods for identifying the different stages of uveitis will aid clinicians in determining prognosis and treatment will be developed. The project will produce processes and algorithms that can be tested to identify the most effective methods.
One modern technology used for assessing inflammatory disease in the eye is called optical coherence tomography (OCT), which provides a simple and non-invasive way to visualise the deep tissue of the retina.
Researchers aim to analyse the diseased eyes of animals at known stages of uveitis, and use modern methods of computer aided image processing to develop our understanding of how to interpret changes in tissue texture and appearance. This information can be applied to human patients to make more reliable predictions about the stage of the disease and its likely outcome.
Researchers will investigate two ways of using images in the assessment of disease. The first of these is FUSION, which describes the computer aided superposition of different pictures of the same object. By combining images, subtle changes in tissues that are difficult to distinguish with the naked eye could be revealed. To achieve image fusion, pictures from OCT will be combined with those obtained from microscopy to identify features that correspond to underlying disease associated changes.
The second way is CLASSIFICATION. This process exploits the power of machine learning to uncover subtle reproducible patterns that indicate the underlying state of the tissues. Hundreds of images of healthy eyes will be used to allow the computer system to learn about normal tissue. Material obtained from animals with uveitis will then be introduced. Some of these will be generated specifically for this project, but access to a large repository of images that have already been collected in other experiments can also be used to validate the trained computer system to measure how well it can detect the changes caused by disease.
Improved quantitative methods for identifying the different stages of uveitis will aid clinicians in determining prognosis and treatment will be developed. The project will produce processes and algorithms that can be tested to identify the most effective methods.