Weakly Supervised Tooth Instance Segmentation on 3D Dental Models with Multi-label Learning

Automatic tooth segmentation on 3D dental models is a fundamental task for computer-aided orthodontic treatment. Many deep learning methods aimed at precise tooth segmentation currently require meticulous point-wise annotations, which are extremely timeconsuming and labor-intensive. To address this issue, we propose a weakly supervised tooth instance segmentation network (WS-TIS) with multi-label learning, which only requires subject-level class labels along with approximately 50% of point-wise tooth annotations. Our WS-TIS consists of two stages, including fine-grained multi-label classification and tooth instance segmentation. Precise tooth localization is frequently pivotal in instance segmentation. However, annotation of tooth centroids or bounding boxes is often challenging when we have limited point-wise tooth annotations. Therefore, we design a proxy task to weakly supervise tooth localization. Specifically, we utilize a fine-grained multi-label classification task, equipping with the disentangled re-sampling strategy and a gated-attention mechanism, which can assist the network in learning discriminative tooth features. Based on discriminative features, discriminative regions can be easily obtained, thereby accurately cropping each tooth. In the second stage, a segmentation module is trained on limited annotated data (approximately 50% of all teeth) to accurately segment each tooth within the cropped regions. Experiments on Teeth3DS demonstrate that our WS-TIS achieves superior performance compared to state-of-the-art approaches under full annotations.