3D Tooth Segmentation and Labeling Using Deep Convolutional Neural Networks.

Area of Science:

• Computer Vision
• Medical Imaging
• Dental Technology

Background:

• Traditional 3D dental segmentation methods struggle with complex tooth anatomy and labeling.
• Existing techniques often fail to accurately segment features like missing or crowded teeth.
• Individual tooth labeling is a significant challenge in current segmentation approaches.

Purpose of the Study:

• To develop a robust and accurate 3D dental model segmentation method using deep learning.
• To overcome limitations of geometry-based methods in handling dental anomalies.
• To enable precise labeling of individual teeth and dental structures.

Main Methods:

• Utilized deep Convolutional Neural Networks (CNNs) for segmentation, labeling each mesh face.
• Extracted geometry features for face representation and trained a 2-level hierarchical CNN structure.
• Implemented a boundary-aware tooth simplification method for efficient feature extraction.
• Applied graph-based label optimization and fuzzy clustering for boundary refinement.

Main Results:

• Achieved a segmentation accuracy of 99.06% measured by area, surpassing state-of-the-art methods.
• The proposed CNN approach demonstrated robustness against foreign matters like air bubbles and dental accessories.
• Successfully enabled accurate labeling of individual teeth and gingiva.

Conclusions:

• The novel deep learning approach offers a significant advancement in 3D dental model segmentation.
• This method provides a highly accurate and reliable tool for applications in orthodontic CAD systems.
• The developed technique addresses key challenges in dental segmentation, paving the way for improved digital dentistry.