Related Experiment Video
Updated: Jun 30, 2025
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
1.8K
DENTALMODELSEG: FULLY AUTOMATED SEGMENTATION OF UPPER AND LOWER 3D INTRA-ORAL SURFACES
Mathieu Leclercq1, Antonio Ruellas2, Marcela Gurgel2
1University of North Carolina, Chapel Hill, United States.
Summary
We developed a deep learning method for surface segmentation using 2D views and convolutional neural networks. This technique accurately segments dental crowns, achieving high precision and sensitivity.
Area of Science:
- Computer Vision
- Medical Imaging
- Machine Learning
Background:
- Accurate surface segmentation is crucial for medical image analysis.
- Traditional methods often struggle with complex surface geometries and variations.
Purpose of the Study:
- To present a novel deep learning-based method for surface segmentation.
- To apply and evaluate this method for dental crown segmentation.
Main Methods:
- Acquiring 2D views and extracting surface features like normal vectors.
- Utilizing a 2D convolutional neural network (UNET) for image analysis.
- Training the network for multi-class segmentation using image labels as ground truth.
Main Results:
- Achieved an average Dice score of 0.97 for dental crown segmentation.
- Demonstrated high performance with an average sensitivity of 0.98 and precision of 0.98.
- Validated the method using 5-fold cross-validation.
Conclusions:
- The proposed deep learning method offers a robust and accurate approach for surface segmentation.
- The technique is effective for dental crown segmentation and is available as a 3DSlicer extension.
