Related Experiment Video
Updated: Sep 4, 2025
High-Speed Human Temporal Bone Sectioning for the Assessment of COVID-19-Associated Middle Ear Pathology
Published on: May 18, 2022
Machine Learning in Diagnosing Middle Ear Disorders Using Tympanic Membrane Images: A Meta-Analysis.
Zuwei Cao1, Feifan Chen2, Emad M Grais2
1Center for Rehabilitative Auditory Research, Guizhou Provincial People's Hospital, Guiyang City, China.
Machine learning models show strong diagnostic accuracy for middle ear disorders using tympanic membrane images. Developing standardized image protocols is recommended for future advancements in this field.
Area of Science:
- Otolaryngology
- Medical Imaging
- Artificial Intelligence
Background:
- Middle Ear Disorders (MED) diagnosis relies on tympanic membrane (TM) imaging.
- Machine Learning (ML) offers potential for automated diagnostic accuracy.
Approach:
- Systematic review and meta-analysis of 16 studies (up to Nov 2021) evaluating ML models for MED classification.
- Included 20,254 TM images; assessed diagnostic accuracy using sensitivity, specificity, and AUC.
- Evaluated risk of bias using QUADAS-2 and PROBA tools.
Key Points:
- ML model accuracy for MED classification ranged from 76.00% to 98.26%.
- Overall sensitivity was 93% and specificity was 85%, with an AUC of 94%.
- Otoendoscopic images yielded higher AUC than otoscopic images.
Conclusions:
- ML models demonstrate robust performance in differentiating normal TM from MED.
- Standardized TM image acquisition and annotation protocols are crucial for further development.
More Related Videos
10:27In Vivo Morphometric Analysis of Human Cranial Nerves Using Magnetic Resonance Imaging in Menière's Disease Ears and Normal Hearing Ears
Published on: February 21, 2018
10:40Discovering Middle Ear Anatomy by Transcanal Endoscopic Ear Surgery: A Dissection Manual
Published on: January 11, 2018
