Related Experiment Video
Updated: Aug 15, 2025
05:49
Author Spotlight: Advancing CBCT and Digital Dental Image Integration with AI-Assisted Digitization
Published on: February 23, 2024
935
An efficient annotation method for image recognition of dental instruments.
Shintaro Oka1, Kazunori Nozaki2, Mikako Hayashi3
1Joint Research Department for Oral Data Science, Osaka University Dental Hospital, Suita, Japan. oka.shintaro.dent@osaka-u.ac.jp.
Scientific Reports
|January 4, 2023
Summary
This study developed an image recognition system to detect dental instruments, improving treatment efficiency and safety. Targeting characteristic instrument parts enhanced detection accuracy, crucial for preventing injuries and optimizing dental procedures.
Area of Science:
- Biomedical Engineering
- Computer Vision
- Dental Surgery
Background:
- Efficient dental treatment requires accurate identification of instruments.
- Preventing needlestick injuries and instrument mismanagement is critical in dentistry.
- Automated instrument recognition can enhance surgical workflow and patient safety.
Purpose of the Study:
- To develop and evaluate an image recognition system for detecting dental instruments during treatment.
- To create a dataset of dental instruments for training artificial intelligence models.
- To establish an efficient method for real-time instrument detection in dental procedures.
Main Methods:
- A dataset of 23 common dental instruments was created, with annotations focusing on characteristic parts or entire instruments.
- Object detection models YOLOv4 and YOLOv7 were employed for instrument recognition.
- System performance was assessed using detection accuracy (DA) and average precision (AP).
Main Results:
- YOLOv7 achieved higher performance, with mean DA of 89.7% and AP of 80.8% when annotating characteristic instrument parts.
- Annotating characteristic parts generally yielded better detection accuracy (DA) and average precision (AP) compared to annotating entire instruments.
- Both YOLOv4 and YOLOv7 demonstrated effective detection capabilities, with YOLOv7 showing superior performance.
Conclusions:
- Image recognition systems, particularly YOLOv7, can effectively detect dental instruments.
- Focusing on characteristic instrument parts improves the efficiency and accuracy of detection.
- This technology holds promise for enhancing safety and efficiency in dental treatments.
