Approved Research

Machine learning for detection of retinal focal abnormalities and for cross-modality prediction.

Lay summary

Modern imaging technology called optical coherence tomgraphy, or OCT, is able to provide three dimensional images of the back of the eye, the retina.  Interpretation of these images can be time-consuming for ophthalmologists and optometrists. Given the widespread use of these devices these eye healthcare professionals could make better use of their time if assisted by computerised interpretation.  Most commercial OCT devices provide some level of computer assistance, mainly as a measurements of the thickness of the retina. However, there is much more information that can be obtained from detailed examination of the images.  For example, it is important to be able to evaluate the build up of unwanted material in the retina that is associated with old-age and systemic diseases such as diabetes.  In most OCT devices, computer assisted methods are not yet available to quantify these pathologies and so these could be missed or imprecisely evaluated in many patient examinations. 

Another strand of this research is to look at glaucoma, one of the major causes of blindness which can be detected using an OCT image.  However, not all patients are receiving an OCT image when they visit an optician.  The research will look at ways in which computer assessment can determine whether it is likely that a patient would benefit from receiving an OCT image.  This would be determined by computer processing a photograph of the retina, an image which is quicker and cheaper to obtain than OCT and is offered as standard by most opticians. The UK Biobank images will be used to help develop computer assisted methods that can be applied to imaging systems from manufacturers other than those used for imaging by UK Biobank, including UK manufacturer, Optos plc.

This project will assist the development of technology that aims to develop faster and better quantified tracking of some of the major diseases that affect the back of the eye.  As this technology becomes available in medical practices, including optometric practices, it will enable better and more timely detection of diseases. This will make contributions to public health particularly in an aging population.  The estimated duration of this project is 24 months.