Related Experiment Video
Updated: Jan 8, 2026
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
1.3K
Development of a deep learning classification model using a codeless platform for orthodontic extraction
1Professor, Department of Orthodontics, Chonnam National University School of Dentistry, Gwangju, South Korea.
Journal of Dentistry
|December 13, 2025
Summary
Deep learning models can aid orthodontic treatment planning. Digital scans slightly outperform intraoral photos for AI-driven extraction decisions, improving accuracy in AI-assisted orthodontic planning.
Area of Science:
- Artificial Intelligence in Dentistry
- Orthodontic Treatment Planning
- Deep Learning Classification Models
Background:
- Orthodontic treatment planning relies on accurate diagnosis and case assessment.
- Artificial intelligence (AI) offers potential for automating and enhancing diagnostic processes.
- Evaluating different imaging modalities is crucial for AI model development in orthodontics.
Purpose of the Study:
- To assess the impact of image type (intraoral photographs vs. digital model scans) on AI model performance for orthodontic treatment planning.
- To investigate the clinical applicability of AI in determining extraction versus non-extraction treatment plans.
- To develop and optimize a deep learning classification model using a codeless platform.
Main Methods:
- Retrospective collection of 1,200 intraoral photographs and 1,200 digital scans from completed orthodontic cases (600 extraction, 600 non-extraction).
- Development of a deep learning classification model using a codeless platform with automated hyperparameter tuning.
- Evaluation of model performance using accuracy, precision, recall, and F1 scores for both image types.
Main Results:
- The deep learning model trained with digital scans achieved higher performance metrics (accuracy: 74.2%, precision: 74.6%, recall: 73.3%, F1 score: 73.9%).
- The model trained with intraoral photographs showed slightly lower performance (accuracy: 71.5%, precision: 72.2%, recall: 71.1%, F1 score: 71.6%).
- Digital scans demonstrated superior results compared to intraoral photographs for the AI classification task.
Conclusions:
- Deep learning models can be effectively developed for orthodontic treatment planning using either intraoral photographs or digital model scans.
- Digital model scans provide a more robust data source for AI-driven orthodontic extraction decision-making.
- AI, utilizing various imaging inputs, shows promise in assisting clinical decisions for orthodontic extraction planning.
