Improving Diabetic Retinopathy Classification Using Class Imbalance Correction Techniques

Diabetic retinopathy (DR) is a severe complication of diabetes mellitus (DM) that produces damage to the retinal blood vessels. It can thus be detected from retinal images. In this work, we address the problem of DR machine learning-based prediction from clinical records and retinal images that suffer from an underrepresentation of the DR cases in the clinical dataset. Several classification models are compared in the task of predicting DR in DM type 1 patients, applying different class imbalance correction techniques. The results show the ability of class weighting and a combined undersampling and oversampling approach to significantly improve the classification of DR cases, which is the class of interest in the medical application.