Analysis of retinal vasculature for MACE risk stratification in patients with diabetes

A cost efficient, non-invasive means of predicting the onset of a Major Adverse Cardiac Event (MACE), or identifying high-risk individuals, would be of tremendous value to the NHS and healthcare systems in general. The potential of retinal imaging as a component of such assessment is under investigation. There already exists strong evidence in recent literature that retinal features can build predictive models of cardiovascular risk factors. Here, we report a specific preliminary study from the Tayside region in Scotland using an extract from the GoDARTS bio-resource. Retinal vascular measurements were computed using VAMPIRE software from images obtained from 600 individuals with type 2 diabetes. Additionally, routine clinical measurements were available. We introduced a risk score derived from retinal vascular information only, via a supervised classification framework, and demonstrated its utility in stratifying risk of developing MACE. We also studied the value of the retinal score in the context of the other clinical measurements. The findings support scaling the work to include a larger number of subjects and additional outcomes of interest such as vascular dementia.