Boundary and Reverse Attention with Channel-wise Feature Pyramid Network for Polyp Segmentation

After years of development, deep convolutional neural networks have been widely used in the field of polyp segmentation. In clinical medicine, the segmentation of polyps in medical images plays an important role in the diagnosis and treatment of diseases. However, due to the different shapes and sizes of polyps, and the boundary between polyps and surrounding tissues and mucosa is not obvious, it is very difficult to accurately segment polyps. Therefore, this paper introduces boundary attention module and reverse attention module in the segmentation network to improve the accuracy of segmentation. At the same time, this paper introduces a Channel-wise Feature Pyramid (CFP) module, which saves computing resources and improves segmentation accuracy. Experiments on five popular polyp segmentation datasets Kvasir-SEG, CVC-ColonDB, CVC-ClinicDB, CVC-300 and ETIS-LaribPolypDB show that The model of this articlel has a good effect in the field of medical image segmentation.