Related Experiment Video
Updated: May 30, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
1.7K
Automatic Identification of Adenoid Hypertrophy via Ensemble Deep Learning Models Employing X-ray Adenoid Images
Sedat Örenç1, Emrullah Acar2, Mehmet Siraç Özerdem3
1Electrical-Electronics Engineering Department, Batman University, Batman, Turkey. sedat.orenc@batman.edu.tr.
Journal of Imaging Informatics in Medicine
|January 30, 2025
Summary
Deep learning models accurately classify adenoid hypertrophy using X-ray images. Masking images significantly improved classification accuracy, with ResNet50 achieving 100% for diagnosing this common childhood condition.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Pediatric Otolaryngology
Background:
- Adenoid hypertrophy, or enlarged adenoids, causes breathing and sleep issues in children.
- Traditional diagnostic methods for adenoid hypertrophy can be inaccurate.
- Accurate diagnosis is crucial for effective treatment and management.
Purpose of the Study:
- To evaluate an ensemble deep learning approach for classifying adenoid hypertrophy.
- To compare the performance of various convolutional neural network (CNN) models.
- To assess the impact of image masking on classification accuracy.
Main Methods:
- Utilized a dataset of masked and non-masked X-ray images from Batman Training and Research Hospital.
- Trained and compared six deep learning models: EfficientNet, MobileNet, ResNet50, ResNet152, VGG16, and Xception.
- Evaluated model performance based on classification accuracy and F1-score.
Main Results:
- ResNet50 achieved 100% accuracy on masked images, demonstrating superior performance.
- Xception showed the lowest performance with a 65% F1-score.
- Image masking significantly enhanced the accuracy and reliability of adenoid classification across models.
Conclusions:
- Deep learning, particularly with image masking, offers a reliable method for adenoid hypertrophy classification.
- ResNet50 and EfficientNet exhibit strong generalization capabilities for this task.
- The study highlights the importance of preprocessing techniques in medical AI applications.
