Intelligent Surgery Planning of CBCT-based Autogenous Tooth Transplantation

Autogenous tooth transplantation (ATT) offers significant clinical potential as it presents the highest biocompatibility for restoration. In conventional procedures, the procedure results in a low success rate due to the increased extra-oral drying time associated with the alveolar socket preparation process. To improve this, we propose an intelligent surgery planning method for ATT using a 3D UX-Net based on centroid compensation. This approach enables precise region detection, target localization, and instance segmentation of each tooth in CBCT images. Incorporating centroid compensation allows the tri-phase network to accurately identify each tooth. We use SegResNet to extract the mandible and maxilla. Principal Component Analysis (PCA) extracts critical information from the segmented teeth. We apply morphological computations for pinpointing locations, determining long-axis orientation, and positioning the dental arch, which assists in defining the post-transplant position of the donor tooth. Additionally, we perform angle optimization based on multi-category constraints. The experiments demonstrate that our method effectively segments the mandible, the maxilla, and individual teeth, while automatically generating an optimal surgical plan that minimizes bone grinding and maximizes overlap between the recipient and donor tooth.