Enhanced Semantic Segmentation of Retinal Microlesions through R2U-Net Architecture

Abstract

Diabetic Retinopathy (DR) is a microvascular complication related to diabetes that affects approximately 33% of individuals with this condition and, if not detected and treated early, can lead to irreversible vision loss. Fundus lesions such as Hard and Soft Exudates, Hemorrhages, and Microaneurysms typically identify DR. The development of computational methods to segment these lesions plays a fundamental role in the early diagnosis of the disease. This article proposes a new approach that uses an R2U-Net combined with data augmentation techniques for segmenting fundus lesions. We trained, adjusted, and evaluated the proposed approach in the DDR dataset, achieving an accuracy of 99.87% and an mIoU equal to 59.69%. Furthermore, we assessed it in the IDRiD dataset, achieving an mIoU of 49.92%. The results obtained in the experiments highlight the potential contribution of the approach in generating lesion annotations in creating new DR datasets, which is essential given the scarcity of annotations in publicly available datasets.