Novel Deep Learning Approach for Tooth Segmentation in Orthodontic Treatment Using STU-RestNet

In this paper we introduce the novel approach in dentistry deep learning method for image segmentation which is helpful for orthodontics treatment. In oral diagnosis treatment the image segmentation plays an important role to determine caries for accurate diagnosis in dental treatment. The previous traditional method failed in the clinical segment in labeling the process of tooth i.e. both label and unlabel data. We address these types of issues using the Swin Transformer UNet and Rest Net method. It is a self-learning-based algorithm named as STU-RestNet–Swin Transformer U-Net RestNet method demonstrated the performance on the dataset which proving its performance and effectiveness, robustness, accuracy in tooth segmentation. Our architecture is based on the segmentation approach with novel mechanisms to decode the various regions and deep supervision technique using STU-RestNet.