Classification of diabetic retinopathy lesions from stereoscopic fundus images

Classification of the severity of diabetic retinopathy (DR) is vital for determining appropriate therapies for this frequent complication of diabetes. The purpose of this study is develop an automated system to help with the classification of some diabetic retinopathy lesions in stereoscopic fundus images. The system is used in particular to distinguish drusen from hard exudates (DR lesions), and to evaluate the severity of lesions such as retinal oedema.