Deep Neural Network Model based on One-Stage Detector for Identifying Fundus Lesions

Diabetic Retinopathy is a major cause of vision loss caused by retina lesions, including hard and soft exudates, microaneurysms, and hemorrhages. The development of a computational tool capable of detecting these lesions can assist in the early diagnosis of the most severe forms of the lesions and assist in the screening process and definition of the best treatment form. However, the detection of tiny objects of very different sizes and shapes makes the detection process more complicated. This paper proposes a computational model based on pre-trained convolutional neural networks capable of detecting fundus lesions to promote medical diagnosis support. We trained, adjusted, and evaluated the model using the DDR diabetic retinopathy dataset and implemented it based on a YOLOv4 architecture and Darknet framework, achieving an mAP of 7.26% and a mloU of 11.64%. The experimental results show that the proposed model presented results superior to those obtained in related works found in the literature.