PVT-DSSE: A Polyp Segmentation Model based on Channel Weight Adjustment

In recent years, automatic segmentation of colorectal polyps based on deep learning has become a popular research area, but in practical applications due to the diversity of polyp morphology and size, as well as the blurring of the boundaries between the polyps and the surrounding mucosa, such that the detection and segmentation of polyps has been a challenging task. The PVT-DSSE model is proposed in this paper to solve these problems. We design a weight adjustment module (DSSE) to enhance the model's segmentation ability for targets of different scales. In addition, a boundary-aware module (CBAM) is introduced, which enhances the network's focus on polyp edges, suppresses irrelevant features, and improves the overall segmentation effect. The experimental results show that the PVT-DSSE model reaches the current state-of-the-art level on five datasets, and is 2.7 percentage points ahead of the current cutting-edge CASCADE on the ETIS dataset containing a large number of small polyps, which proves its potential for clinical application in polyp segmentation.