Related Experiment Video
Updated: Jan 8, 2026
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
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.
Objectives:
To evaluate the effect of image type on the performance of a deep learning classification model for decision-making in orthodontic treatment planning, and to investigate the clinical applicability of artificial intelligence in determining extraction and non-extraction treatment plans using intraoral photographs and digital model scans.
Methods:
Pre-treatment patient data and orthodontic treatment plans were retrospectively collected from patients who had completed their orthodontic treatment. A total of 1200 intraoral photographs and 1200 digital scans of pre-treatment dental models were obtained, comprising 600 extraction cases and 600 non-extraction cases for each image type. All images were randomly divided into a training set (80 %), validation set (10 %), and a test set (10 %). A deep learning classification model was automatically developed using a codeless platform, which optimized the model through automated hyperparameter tuning without the need for manual coding. Model performance was evaluated by comparing accuracy, precision, recall, and F1 scores between the different image types.
Results:
The performance of the deep learning model trained with intraoral photographs achieved an accuracy of 71.5 %, a precision of 72.2 %, a recall of 71.1 %, and an F1 score of 71.6 %. The performance of the deep learning model trained with digital scans showed an accuracy of 74.2 %, a precision of 74.6 %, a recall of 73.3 %, and an F1 score of 73.9 %. Overall, the model trained with digital scans demonstrated superior performance compared to the model trained with intraoral photographs.
Conclusions:
This study elucidated the feasibility of developing an optimal deep learning classification model for determining extraction and non-extraction orthodontic treatment plans using intraoral photographs or digital model scans.
Clinical Significance Statement:
Intraoral photographs or digital dental model images alone may assist clinical decision-making for orthodontic extraction planning through the application of artificial intelligence.
