Related Experiment Video
Updated: Jan 9, 2026
09:10
Digital Hybrid Model Preparation for Virtual Planning of Reconstructive Dentoalveolar Surgical Procedures
Published on: August 5, 2021
2.2K
Deep learning-based tooth segmentation for enhanced visualization of dental anomalies and pathologies
Fulin Jiang1, Shihao Li2, Jialing Liu3
1Chongqing University Three Gorges Hospital, Chongqing University, Chongqing 400044, China.
Summary
A new deep learning method accurately segments individual teeth in Cone Beam CT scans, improving dental anomaly detection and reducing diagnosis time for dentists.
Area of Science:
- Artificial Intelligence
- Medical Imaging
- Dentistry
Background:
- Dental anomaly detection is crucial for oral health.
- Current visualization methods can be time-consuming.
- Accurate tooth segmentation aids in identifying pathologies.
Purpose of the Study:
- To develop and validate a deep learning method for instance-level tooth segmentation in Cone Beam CT (CBCT) scans.
- To enhance visualization and streamline the detection of dental anomalies.
- To assess the accuracy and efficiency of the automated segmentation.
Main Methods:
- A deep learning model was trained on 470 CBCT scans with diverse dental anomalies and histories.
- An accelerated annotation procedure utilized expert input to train the model.
- Experienced dentists validated anomaly detection using model-generated segmentations on 60 scans.
Main Results:
- The model achieved high accuracy (0.934 ± 0.045 Jaccard index) in segmenting teeth.
- Segmentation of a single scan took an average of 7.025 ± 2.885 seconds.
- Dentists using segmentation overlays reduced anomaly detection time by 20%.
Conclusions:
- The deep learning framework provides automated, accurate tooth segmentation in CBCT volumes.
- The system demonstrates high geometric fidelity and clinically acceptable processing times.
- This accurate segmentation tool shows significant potential for general dentistry applications.
