Related Experiment Video
Updated: Aug 1, 2025
05:56
Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
Published on: April 14, 2023
2.6K
MIB-ANet: A novel multi-scale deep network for nasal endoscopy-based adenoid hypertrophy grading
Mingmin Bi1, Siting Zheng2,3, Xuechen Li2,3
1Department of Otolaryngology, The Seventh Affiliated Hospital of Sun Yat-sen University, Shenzhen, China.
Frontiers in Medicine
|May 1, 2023
Summary
A new deep learning model, MIB-ANet, accurately grades adenoid hypertrophy from nasal endoscopy images. It outperforms junior clinicians in accuracy and speed, offering a promising tool for diagnosis.
Area of Science:
- Otorhinolaryngology
- Artificial Intelligence
- Medical Imaging
Background:
- Adenoid hypertrophy diagnosis relies on subjective grading.
- Objective and automated methods are needed for accurate assessment.
Purpose of the Study:
- To develop a novel deep learning model for automated adenoid hypertrophy grading.
- To compare the model's performance against Ear, Nose, and Throat (ENT) clinicians.
Main Methods:
- A multi-scale grading network (MIB-ANet) was developed using 3,179 nasoendoscopic images.
- The Parikh grading standard (4 grades) was used for adenoid hypertrophy classification.
- Performance was evaluated against junior ENT clinicians.
Main Results:
- MIB-ANet achieved an F1 score of 0.76251 and accuracy of 0.76807 on the SYSU-SZU-EA Dataset.
- The model demonstrated superior classification performance compared to other networks.
- MIB-ANet was 80x faster than junior clinicians, with higher F1 scores and accuracy.
Conclusions:
- The MIB-ANet model shows significant potential for automated adenoid hypertrophy grading.
- It offers improved accuracy and efficiency over manual grading by clinicians.
- Further research is needed to enhance the model's accuracy.
