Related Experiment Video
Updated: Jun 22, 2026
05:57
Long-term Video Tracking of Cohoused Aquatic Animals: A Case Study of the Daily Locomotor Activity of the Norway Lobster Nephrops norvegicus
Published on: April 8, 2019
6.9K
Automatic Recognition of Concealed Fish Bones under Laryngoscopy: A Practical AI Model Based on YOLO-V5
Xiaoyao Tao1, Xu Zhao2, Hairui Liu3
1Otorhinolaryngology Head and Neck Surgery Department, The First Affiliated Hospital of Anhui Medical University, Hefei, China.
The Laryngoscope
|November 20, 2023
Summary
An AI model using YOLO-V5 can efficiently detect fish bones in the throat using laryngoscopy images. This artificial intelligence tool assists doctors in identifying these foreign bodies faster than human experts.
Area of Science:
- Otolaryngology
- Artificial Intelligence
- Medical Imaging
Background:
- Fish bone impaction is a frequent otolaryngology emergency.
- Identifying fish bones during laryngoscopy is challenging due to their appearance and pharyngeal anatomy complexity.
- Expert clinical experience is crucial for efficient fish bone detection.
Purpose of the Study:
- To develop an AI model to aid clinicians in detecting pharyngeal fish bones during laryngoscopy.
- To improve the efficiency and accuracy of fish bone identification in emergency settings.
Main Methods:
- Trained a YOLO-V5 algorithm model using 3133 laryngoscopic images of fish bones.
- Validated the model's performance on a separate test dataset, comparing its predictions to human experts.
- Assessed real-time detection capabilities using seven laryngoscopic videos.
Main Results:
- The YOLO-V5 model achieved an average precision of 0.857 (IOU threshold 0.5), with precision, recall, and F1 scores of 0.909, 0.818, and 0.87, respectively.
- Overall accuracy in the validation set was 0.821, comparable to ENT specialists.
- The model processed images significantly faster (0.012s per image) than humans and showed strong video recognition performance.
Conclusions:
- The AI model effectively identifies and localizes fish bone foreign bodies in both static and dynamic laryngoscopic views.
- This AI tool demonstrates significant potential for clinical application in otolaryngology emergencies.
