Related Experiment Video
Updated: Jun 9, 2026
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
Deep Learning for Endoscopic Classification of Adenoid Hypertrophy
Xuan-Sheng Wang1, De-Sheng Jia2, Zhi-Xin Mo2
1Shenzhen Institute of Information Technology Shenzhen China.
Summary
A new deep learning algorithm accurately classifies adenoid hypertrophy from endoscopic images. This artificial intelligence approach enhances diagnostic efficiency and objectivity for adenoid size assessment.
Area of Science:
- Otolaryngology
- Medical Imaging
- Artificial Intelligence
Background:
- Endoscopy is standard for evaluating adenoid size, but subjective interpretation leads to inaccuracies.
- Adenoid hypertrophy assessment requires objective and reliable methods.
Purpose of the Study:
- To develop an automated classification strategy for adenoid hypertrophy using deep learning.
- To improve the accuracy and objectivity of adenoid size diagnosis from nasal endoscopic images.
Main Methods:
- A convolutional neural network (CNN) algorithm, Xception, was trained on 26,060 labeled nasal endoscopic images.
- Images were classified into four grades based on choanal obstruction.
- The model's performance was validated using a separate test set and receiver operating characteristic (ROC) curves.
Main Results:
- The Xception model achieved an overall classification accuracy of 95.53%.
- Area under the curve (AUC) values for Grades I-IV were 0.93, 0.94, 0.97, and 0.91, respectively.
- The deep learning approach demonstrated high reliability in grading adenoid hypertrophy.
Conclusions:
- Artificial intelligence, specifically deep learning, is effective for classifying endoscopic adenoidal hypertrophy.
- This AI-driven method offers potential improvements in diagnostic efficiency, objectivity, and stability.
