Identifying common genetic variants associated with diabetic retinopathy in the CLSA

Project Summary

Diabetes currently affects more than 1 in 10 adults in Canada, and is the most common cause of adult-onset blindness in this country. Genetic, epidemiologic and clinical research has identified many risk factors for eye complications of diabetes including the degree of hyperglycemia, blood pressure elevation and genetic predisposition. To date, this research has been limited by a paucity of evidence pertaining to the genetic determinants of diabetic retinopathy due, in part, to the paucity of large databases of retinal photographs for which genetic and phenotypic data are available. When such data are available Mendelian randomization techniques can be very creatively used to confirm and discover novel causal factors.

Project Findings

With this study, we tested feasibility of using deep learning algorithms to automatedly read retinal images. Preliminary analyses were conducted to assess the image quality. Eighty per cent (80%) of images were found suitable to be automatedly read by deep learning algorithms.