Vnet-Transform Architecture For CBCT Image Segmentation With Ai Empowerment

This study developed and evaluated a V-Net-Transformer-based segmentation network to enhance the accuracy and efficiency of fully automatic tooth and alveolar bone segmentation in cone-beam computed tomography (CBCT) imaging. A dataset of 939 CBCT scans was used for training and validation using V-net-Transformer network. The segmentation pipeline and multitask learning for teeth, teeth roots and alveolar bone. Performance was evaluated using the Dice similarity coefficient (DSC) on internal and external testing sets. The performance of Dice of 96.21% and 95.21%, mPrecision of 0.87, and mRecall of 0.85 on internal and external test sets. AI system reduced processing time with maintaining high accuracy compared to manual methods. The V-net-Transformer-based network provides an accurate and precise segmentation of CBCT images in digital dentistry.