3D reconstruction for maxillary anterior tooth crown based on shape and pose estimation networks

Yuan Feng; BaoXin Tao; JiaCheng Fan; ShiGang Wang; JinQiu Mo; YiQun Wu; QingHua Liang

doi:10.1007/s11548-023-02841-1

3D reconstruction for maxillary anterior tooth crown based on shape and pose estimation networks

Int J Comput Assist Radiol Surg. 2023 Aug;18(8):1405-1416. doi: 10.1007/s11548-023-02841-1. Epub 2023 Feb 8.

Authors

Yuan Feng^#¹, BaoXin Tao^#², JiaCheng Fan¹, ShiGang Wang¹, JinQiu Mo¹, YiQun Wu³, QingHua Liang⁴

Affiliations

¹ School of Mechanical Engineering, Shanghai Jiao Tong University, Room 901, Dongchuan Road 800, Minhang District, Shanghai, 200240, China.
² Department of Second Dental Center, Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine; College of Stomatology, Shanghai Jiao Tong University; National Center for Stomatology; National Clinical Research Center for Oral Diseases; Shanghai Key Laboratory of Stomatology; Research Unit of Oral and Maxillofacial Regenerative Medicine, Chinese Academy of Medical Sciences, Shanghai, China.
³ Department of Second Dental Center, Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine; College of Stomatology, Shanghai Jiao Tong University; National Center for Stomatology; National Clinical Research Center for Oral Diseases; Shanghai Key Laboratory of Stomatology; Research Unit of Oral and Maxillofacial Regenerative Medicine, Chinese Academy of Medical Sciences, Shanghai, China. yiqunwu@hotmail.com.
⁴ School of Mechanical Engineering, Shanghai Jiao Tong University, Room 901, Dongchuan Road 800, Minhang District, Shanghai, 200240, China. qhliang@sjtu.edu.cn.

^# Contributed equally.

PMID: 36754949
DOI: 10.1007/s11548-023-02841-1

Abstract

Purpose: The design of a maxillary anterior tooth crown is crucial to post-treatment aesthetic appearance. Currently, the design is performed manually or by semi-automatic methods, both of which are time-consuming. As such, automatic methods could improve efficiency, but existing automatic methods ignore the relationships among crowns and are primarily used for occlusal surface reconstruction. In this study, the authors propose a novel method for automatically reconstructing a three-dimensional model of the maxillary anterior tooth crown.

Method: A pose estimation network (PEN) and a shape estimation network (SEN) are developed for jointly estimating the crown point cloud. PEN is a regression network used for estimating the crown pose, and SEN is based on an encoder-decoder architecture and used for estimating the initial crown point cloud. First, SEN adopts a transformer encoder to calculate the shape relationship among crowns to ensure that the shape of the reconstructed point cloud is precise. Second, the initial point cloud is subjected to pose transformation according to the estimated pose. Finally, the iterative method is used to form the crown mesh model based on the point cloud.

Result: The proposed method is evaluated on a dataset with 600 cases. Both SEN and PEN are converged within 1000 epochs. The average deviation between the reconstructed point cloud and the ground truth of the point cloud is 0.22 mm. The average deviation between the reconstructed crown mesh model and the ground truth of the crown model is 0.13 mm.

Conclusion: The results show that the proposed method can automatically and accurately reconstruct the three-dimensional model of the missing maxillary anterior tooth crown, which indicates the method has promising application prospects. Furthermore, the reconstruction time takes less than 11 s for one case, demonstrating improved work efficiency.

Keywords: Maxillary anterior tooth crown reconstruction; Pose and shape decoupling; Pose estimation network; Shape estimation network.

MeSH terms

Crowns
Humans
Imaging, Three-Dimensional* / methods
Maxilla / diagnostic imaging
Tooth Crown / diagnostic imaging
Tooth*

Grants and funding

2017YFB1302901/National Key R&D Program of China