Segmentation of Laser Marks of Diabetic Retinopathy in the Fundus Photographs Using Lightweight U-Net

Figure 6

Dataset and prediction result presentation. Each row represented a sample in the testing sets and its corresponding processed results. (a) Laser marks showed in the fundus photographs of patients with DR. (b) The ground truth labels of laser marks in the testing images. (c) The raw predictions for laser marks made by the lightweight U-Net. (d) The predictions for laser marks made by the lightweight U-Net after optic disc areas were covered. (e) The postprocessed lightweight U-Net showed more accurate outcomes in the segmentation of laser makers. (f) The merge of postprocessed prediction maps and the original fundus photographs.