Related Experiment Video
Updated: Aug 22, 2025
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
944
Combined Deep Learning Techniques for Mandibular Fracture Diagnosis Assistance
Dong-Min Son1, Yeong-Ah Yoon2, Hyuk-Ju Kwon1
1School of Electronic and Electrical Engineering, Kyungpook National University, 80 Daehakro, Buk-gu, Daegu 41566, Korea.
Life (Basel, Switzerland)
|November 11, 2022
Summary
This study introduces a deep learning approach combining YOLO and U-Net to improve mandibular fracture diagnosis from panoramic X-rays. The enhanced method increases the detection of fractures, aiding dentists in diagnosis without cone beam computed tomography (CBCT).
Area of Science:
- Dentistry
- Medical Imaging
- Artificial Intelligence
Background:
- Mandibular fractures are common but challenging to diagnose accurately using only panoramic radiography.
- Cone beam computed tomography (CBCT) is often required, increasing costs and radiation exposure.
- Deep learning offers potential for improved diagnostic accuracy from standard radiographic images.
Purpose of the Study:
- To develop and evaluate a combined deep learning model (YOLO and U-Net) for detecting mandibular fractures on panoramic images.
- To improve the diagnostic performance compared to using YOLO alone, particularly for difficult-to-detect fractures.
- To provide an auxiliary diagnostic tool for dentists, reducing the need for CBCT.
Main Methods:
- Utilized YOLOv4 for initial fracture detection and U-Net for semantic segmentation of fracture areas.
- Trained and tested the combined deep learning model on panoramic radiographic images.
- Evaluated model performance using precision and recall scores.
Main Results:
- The YOLOv4 module achieved a precision of approximately 97% but a recall of 79% for mandibular fractures.
- The combined YOLOv4 and U-Net model demonstrated a precision of 95% and an improved recall of 87%.
- The U-Net component effectively addressed limitations of YOLO in segmenting complex and overlapping fracture regions.
Conclusions:
- The combined deep learning approach significantly enhances the detection rate of mandibular fractures from panoramic images.
- This AI-powered tool can assist dentists by improving diagnostic accuracy and potentially reducing reliance on CBCT.
- Further development of AI in dental diagnostics holds promise for more efficient and accessible patient care.
