Related Experiment Video
Updated: May 5, 2026
08:51
Data Acquisition and Analysis In Brainstem Evoked Response Audiometry In Mice
Published on: May 10, 2019
11.6K
Machine-learning based classification of middle-ear fixation and separation using sweep frequency impedance
Teruki Toya1,2, Hisashi Sugimoto3, Di Zhou2
1Faculty of Engineering, Graduate School Department, University of Yamanashi, Kofu, 400-8511, Japan.
The Journal of the Acoustical Society of America
|May 13, 2025
Summary
A new sweep frequency impedance (SFI) meter enhances middle ear analysis. Machine learning applied to SFI data accurately predicts ossicular chain fixation and separation, outperforming traditional tympanometry.
Area of Science:
- Otoacoustic Emissions and Middle Ear Analysis
- Biomedical Engineering
- Diagnostic Technology
Background:
- The sweep frequency impedance (SFI) meter analyzes middle ear dynamics by measuring sound pressure changes in response to frequency-sweeping sounds.
- Previous research indicated SFI's potential for diagnosing ossicular fixation and separation via 2D mobility maps, but lacked concrete classification criteria.
- Clinical application was hindered by overlapping data in the 2D mobility maps, necessitating improved diagnostic methods.
Purpose of the Study:
- To develop and validate a machine-learning-based method for predicting middle ear dysfunctions using 2D characteristics derived from SFI measurements.
- To improve the accuracy and clinical applicability of SFI measurements for diagnosing ossicular chain abnormalities.
- To compare the diagnostic performance of the proposed machine learning method with conventional tympanometry.
Main Methods:
- Renewal of the SFI meter to achieve a higher signal-to-noise ratio for detecting sound pressure variations.
- Conducting SFI measurements on both normal and impaired ears using the renewed apparatus.
- Applying a machine-learning algorithm to analyze the 2D mobility map features for predicting middle ear dysfunctions.
Main Results:
- Ossicular chain fixation was predicted with 0.8 accuracy and an AUC of 0.86.
- Ossicular chain separation was predicted with perfect accuracy (1.0) and an AUC of 1.0.
- The machine learning-based SFI method demonstrated higher predictive potential compared to conventional tympanometry.
Conclusions:
- The proposed machine learning approach effectively utilizes 2D SFI characteristics for accurate prediction of middle ear dysfunctions.
- This novel method shows significant promise for improving the diagnosis of ossicular chain fixation and separation.
- The enhanced SFI meter and machine learning integration offer a more accurate diagnostic tool for middle ear pathologies.
