Quantifying Change in Retinal Images
Allocated in academic year 2014-15
The clinical significance of the relationship between signs detected on retinal imaging and our understanding of other diseases, such kidney or cardiovascular disease is an evolving area. Since microcirculation can be imaged directly in the retina, changes in microvasculature that convey important information about, for example, arterial hypertension or coronary heart disease can be quantified. We have built a collaborating network of 14 ophthalmic units who have pooled data from electronic medical records and associated retinal colour images and optical coherence tomography (OCT) scan to create a unique data set (80 to 100,000 patients) on eye disease. Discussions are underway to link this with two key datasets: NICOR and the UK Renal Registry Dataset. This will allow us to explore a variety of hypotheses about the link about, for example, kidney function and the progression of diabetic eye disease. In order to extend the potential of this work we want to create a library of validated algorithms for the accurate quantification of key signs of changes to the microvasculature, and macular oedema as identified on retinal and OCT images. This will involve the segmentation of retinal vessels, accurate measurement of vessel diameter and the measurement of vessel tortuosity and branching patterns as well as the segmentation of OCT scans. The development of efficient and accurate algorithms for the quantification of these features, coupled with a large database of eye disease linked to other key national resources will be a unique asset to answer important clinical questions.