Multiloss Function Based Deep Convolutional Neural Network for Segmentation of Retinal Vasculature into Arterioles and Venules

Figure 12

Example of how the segment-wise loss judgment enhanced the segmentation results. The images are from (a) original image, (b) label, (c) pixel-wise deep learning loss, and (d) multiloss function deep learning results. The two circles in Figure 12(c) show the wrong judgment that is corrected by multiloss function improvement in our proposed method.