Related Experiment Video
Updated: Jun 2, 2025
07:32
Author Spotlight: 3D Movement Assessment of Maxillary Posterior Teeth in Clear Aligner Treatment
Published on: February 23, 2024
1.0K
Assessment of using transfer learning with different classifiers in hypodontia diagnosis
Tansel Uyar1, Didem Sakaryalı Uyar2
1Biomedical Engineering Department, Başkent University, 06810, Ankara, Turkey. tuyar@baskent.edu.tr.
BMC Oral Health
|January 14, 2025
Summary
Artificial intelligence accurately classifies tooth agenesis in children using panoramic radiographs. The VGG-19 model achieved high accuracy in identifying single premolar agenesis, multiple premolar agenesis, and without tooth agenesis.
Area of Science:
- Dentistry
- Medical Imaging
- Artificial Intelligence
Background:
- Hypodontia, the congenital absence of teeth, is typically diagnosed using radiographic imaging.
- Artificial intelligence (AI) decision support systems are increasingly used for accurate diagnoses.
Purpose of the Study:
- To classify single premolar agenesis, multiple premolar agenesis, and absence of tooth agenesis.
- To evaluate various AI approaches for diagnosing hypodontia in pediatric patients.
Main Methods:
- Utilized 1,068 panoramic radiographs from pediatric patients (6-12 years old).
- Trained pretrained convolutional neural network (CNN) models using fine-tuning.
- Classified data using machine learning algorithms and evaluated performance metrics (accuracy, precision, recall, F1-score, specificity, AUC).
Main Results:
- The VGG-19 model with a neural network classifier achieved the highest performance.
- Achieved 95.63% accuracy, 93.34% recall, and 95.03% AUC.
- High accuracy was observed across all classes: without tooth agenesis (96.72%), multiple premolar agenesis (95.79%), and single premolar agenesis (94.39%).
Conclusions:
- Pretrained AI models demonstrate significant success in the radiographic diagnosis of hypodontia.
- AI approaches are expected to enhance the diagnostic process for hypodontia in pediatric dental care.
