3D Teeth Scan Segmentation and Labelling Challenge

Computer-aided design (CAD) tools have become increasingly popular in modern dentistry for highly accurate treatment planning. In particular, in orthodontic CAD systems, advanced intraoral scanners (IOSs) are now widely used as they provide precise digital surface models of the dentition. Such models can dramatically help dentists simulate teeth extraction, move, deletion, and rearrangement and ease therefore the prediction of treatment outcomes. Hence, digital teeth models have the potential to release dentists from otherwise tedious and time consuming tasks.

Although IOSs are becoming widespread in clinical dental practice, there are only few contributions on teeth segmentation/labelling available in the literature [1,2,3] and no publicly available database. A fundamental issue that appears with IOS data is the ability to reliably segment and identify teeth in scanned observations. Teeth segmentation and labelling is difficult as a result of the inherent similarities between teeth shapes as well as their ambiguous positions on jaws.

In addition, it faces several challenges:
1- The teeth position and shape variation across subjects.
2- The presence of abnormalities in dentition. For example, teeth crowding which results in teeth misalignment
and thus non-explicit boundaries between neighboring teeth. Moreover, lacking teeth and holes are commonly
seen among people.
3- Damaged teeth.
4- The presence of braces, and other dental equipment.

The challenge we propose will particularly focus on point 1, i.e. the teeth position and shape variation across subjects. With the extension of available data in the mid and long term, the other points will also be addressed in further editions of the challenge.

[1] Lian, Chunfeng, et al. "MeshSNet: Deep multi-scale mesh feature learning for end-to-end tooth labeling on 3D dental surfaces." International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, Cham, 2019.
[2] Xu, Xiaojie, Chang Liu, and Youyi Zheng. "3D tooth segmentation and labeling using deep convolutional neural networks." IEEE transactions on visualization and computer graphics 25.7 (2018): 2336-2348.
[3] Sun, Diya, et al. "Automatic Tooth Segmentation and Dense Correspondence of 3D Dental Model." International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, Cham, 2020.