The extraction method of tooth preparation margin line based on S-Octree CNN

Area of Science:

• Computer-aided dentistry
• Digital imaging and machine learning

Background:

• Accurate tooth preparation margin line identification is crucial for dental restoration marginal fitness.
• Existing manual methods for margin line extraction are complex and time-consuming.

Purpose of the Study:

• To develop an automated method for extracting tooth preparation margin lines.
• To improve the efficiency and accuracy of margin line identification in dental restorations.

Main Methods:

• Utilized a convolutional neural network (CNN) integrated with a sparse octree (S-Octree) structure for margin line extraction.
• Employed data augmentation through dental preparation rotation.
• Applied spatial partitioning with S-Octree to create labeled sparse point clouds.
• Implemented dense conditional random field (dense CRF), point cloud reconstruction, and back projection for precise margin line identification.

Main Results:

• The automated method achieved an average accuracy of 97.43% in predicting margin line labels.
• The CNN model successfully divided dental preparation point clouds into distinct regions based on feature lines.
• The developed technique effectively automates the extraction of tooth preparation margin lines.

Conclusions:

• The proposed CNN-based method with S-Octree structure offers an efficient and accurate solution for automated tooth preparation margin line extraction.
• This approach significantly overcomes the limitations of manual methods, enhancing digital dentistry workflows.