Tooth Segmentation with An Edge-Guided SAM-Based

Segmentation of teeth in dental panoramic X-ray images is a task of significant clinical importance. However, previous studies often overlooked the importance of edge information, leading to inaccurate segmentation of teeth with blurred boundaries and low contrast. In this paper, we propose an edge-guided transformer method aimed at utilizing edge information to assist in the segmentation of dental panoramic X ray images through salient features. A dual-stream structure is employed, with the image encoding layer of SAM used for the salient branch, and another part adopting a CNN structure similar to U-Net as the edge branch. The salient branch is then fused with the edge branch, and the fused edge feature map is combined with the results of the salient branch and fed into the subsequent structure. Extensive experiments on three public benchmark datasets demonstrate the effectiveness and superiority of our proposed method compared to other state-of-the-art methods. This method demonstrates the ability to accurately identify and analyze the structure of teeth, thereby providing important information for dental diagnosis, treatment planning, and research.