Related Experiment Video
Updated: Jun 29, 2025
04:32
Murine Oropharyngeal Aspiration Model of Ventilator-associated and Hospital-acquired Bacterial Pneumonia
Published on: June 28, 2018
11.8K
An automatic diagnosis model of otitis media with high accuracy rate using transfer learning
Fangyu Qi1,2,3, Zhiyu You2,3, Jiayang Guo2,3
1Department of Anesthesiology, The Seventh Affiliated Hospital, Sun Yat-sen University, Shenzhen, China.
Frontiers in Molecular Biosciences
|April 5, 2024
Summary
This study uses deep learning on CT scans to accurately diagnose Chronic Suppurative Otitis Media (CSOM) and middle ear cholesteatoma, achieving over 92% accuracy. The method improves upon existing models for these common ear conditions.
Area of Science:
- Medical Imaging
- Artificial Intelligence in Medicine
- Otolaryngology
Background:
- Chronic Suppurative Otitis Media (CSOM) and middle ear cholesteatoma present diagnostic challenges due to similar appearances on CT scans.
- Accurate differentiation is crucial for effective patient management and treatment planning.
Purpose of the Study:
- To develop an automated diagnostic system for CSOM and middle ear cholesteatoma using CT imaging.
- To leverage transfer learning and deep learning for improved lesion segmentation and classification accuracy.
Main Methods:
- Utilized a dataset of 1019 CT scans of the internal auditory canal.
- Employed the nnUnet skeleton model with coarse-grained focal segmentation labeling for pre-training.
- Fine-tuned the model for a three-classification diagnosis task (CSOM, cholesteatoma, normal).
Main Results:
- Achieved a classification accuracy of 92.33% for CSOM and middle ear cholesteatoma, outperforming the benchmark model by approximately 5%.
- The upstream segmentation task yielded a mean Intersection of Union (mIoU) of 0.569.
- Demonstrated that coarse-grained contour boundary labeling significantly enhances downstream classification performance.
Conclusions:
- Deep learning combined with automatic diagnosis shows high sensitivity and specificity for CSOM and middle ear cholesteatoma detection via CT.
- The proposed method offers a promising tool for accurate and efficient diagnosis of these common otitis media conditions.
