Semantic image segmentation, which becomes one of the key applications in image processing and computer vision domain, has been used in multiple domains such as medical area and intelligent transportation. Lots of benchmark datasets are released for researchers to verify their algorithms. Semantic segmentation has been studied for many years. Since the emergence of Deep Neural Network (DNN), segmentation has made a tremendous progress. In this paper, we divide semantic image segmentation methods into two categories: traditional and recent DNN method. Firstly, we briefly summarize the traditional method as well as datasets released for segmentation, then we comprehensively investigate recent methods based on DNN which are described in the eight aspects: fully convolutional network, upsample ways, FCN joint with CRF methods, dilated convolution approaches, progresses in backbone network, pyramid methods, Multi-level feature and multi-stage method, supervised, weakly-supervised and unsupervised methods. Finally, a conclusion in this area is drawn.

The hypoxic microenvironment, an important feature of solid tumors, promotes tumor cells to release exosomes and enhances tumor angiogenesis. However, the detailed functions of hypoxic exosomes and the mechanisms underlying their effects in pancreatic cancer (PC) remain mysterious. Here, we observed that hypoxic exosomes derived from PC cells promoted cell migration and tube formation of human umbilical vein endothelial cells (HUVECs). The long noncoding RNA (lncRNA) UCA1 , a key factor, was highly expressed in exosomes derived from hypoxic PC cells and could be transferred to HUVECs through the exosomes. In addition, the expression levels of UCA1 in exosomes derived from PC patients’ serum were higher than in healthy controls and were associated with poor survival of PC patients. Moreover, hypoxic exosomal UCA1 could promote angiogenesis and tumor growth both in vitro and in vivo . With respect to the functional mechanism, UCA1 acted as a sponge of microRNA (miR)-96-5p, relieving the repressive effects of miR-96-5p on the expression of its target gene AMOTL2. Collectively, these results indicate that hypoxic exosomal UCA1 could promote angiogenesis and tumor growth through the miR-96-5p/AMOTL2/ERK1/2 axis and therefore, serve as a novel target for PC treatment.

Pancreatic ductal adenocarcinoma (PDAC), one of the most aggressive tumors all over the world, has a generally poor prognosis, and its progression is positively correlated with the density of blood vessels. Recently, tumor-associated macrophages (TAMs) were proven to be beneficial for angiogenesis, but their mechanism of action remains unclear. Our study indicated that M2 macrophages were positively correlated with the microvessel density (MVD) of PDAC tissues, and M2 macrophage-derived exosomes (MDEs) could promote the angiogenesis of mouse aortic endothelial cells (MAECs) in vitro. At the same time, the M2 MDEs could also promote the growth of subcutaneous tumors and increase the vascular density of mice. Moreover, we also found that miR-155-5p and miR-221-5p levels in the M2 MDEs were higher than those in M0 MDEs, and they could be transferred into MAECs, as demonstrated by RNA sequencing (RNA-seq) and qPCR analysis. Our data confirmed the interaction between TAMs and the angiogenesis of PDAC by exosomes. Additionally, targeting the exosomal miRNAs derived from TAMs might provide diagnostic and therapeutic strategies for PDAC.