Automatic 3-dimensional quantification of orthodontically induced root resorption in cone-beam computed tomography...

Area of Science:

• Biomedical Imaging
• Artificial Intelligence in Dentistry
• Orthodontics

Background:

• Orthodontically induced root resorption (OIRR) is a common complication of orthodontic treatment.
• Manual 3D quantitative analysis of OIRR using CBCT is subjective and time-consuming.
• Deep learning offers potential for automated medical image analysis.

Purpose of the Study:

• To develop and validate a deep learning-based model for automatic OIRR quantification.
• To extract root volume information and localize root resorption from CBCT images automatically.

Main Methods:

• A retrospective study utilized 4534 teeth from 105 patients.
• CBCT images underwent preprocessing, including automatic tooth segmentation and conversion to point clouds.
• A Dynamic Graph Convolutional Neural Network segmented tooth crowns and roots for volume calculation and OIRR localization.

Main Results:

• The model demonstrated strong correlation with manual measurements for root volume and OIRR severity.
• Intraclass correlation coefficients for average volume measurements exceeded 0.95 (P < 0.001).
• Accuracy for classifying OIRR severity surpassed 0.8.

Conclusions:

• The proposed methodology provides automatic and reliable tools for OIRR assessment.
• This AI-driven approach can potentially enhance orthodontic treatment planning and monitoring.