Related Experiment Video
Updated: Jun 12, 2026
12:43
A Protocol for Comprehensive Assessment of Bulbar Dysfunction in Amyotrophic Lateral Sclerosis (ALS)
Published on: February 21, 2011
Vocalization removal for improved automatic segmentation of dual-axis swallowing accelerometry signals
Ervin Sejdić1, Tiago H Falk, Catriona M Steele
1Bloorview Research Institute, Bloorview Kids Rehab and the Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, Ontario, Canada. esejdic@ieee.org
Medical Engineering & Physics
|May 21, 2010
Summary
This study introduces a novel method to remove vocalizations from swallowing accelerometry signals, improving automatic segmentation accuracy by 55%. This advancement aids in developing better medical devices for detecting swallowing difficulties.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Speech and Swallowing Analysis
Background:
- Automatic segmentation of swallowing accelerometry signals is crucial for diagnosing dysphagia.
- Strong vocalizations frequently interfere with the accuracy of swallowing signal segmentation.
- Existing methods struggle to effectively isolate and remove vocalization artifacts.
Purpose of the Study:
- To propose and evaluate a novel method for detecting and removing vocalizations from dual-axis swallowing accelerometry data.
- To enhance the accuracy of automatic segmentation algorithms for swallowing signals.
- To improve the reliability of accelerometry-based swallowing assessments.
Main Methods:
- A periodicity detection method was developed to identify vocalization components in swallowing signals.
- Conventional speech processing techniques were employed for signal analysis.
- Information from both accelerometer axes was fused to increase vocalization detection precision.
Main Results:
- The proposed method achieved an average sensitivity of 95.3% and specificity of 96.3% in experiments with 408 healthy subjects.
- Vocalization detection was tested across dry, wet, and wet chin tuck swallows.
- Integrating the method with an automatic segmentation algorithm resulted in a ~55% improvement in segmentation accuracy.
Conclusions:
- The developed periodicity detection method effectively removes vocalizations from swallowing accelerometry signals.
- Improved signal segmentation accuracy facilitates more reliable swallowing disorder detection.
- This research supports the advancement of medical devices for assessing swallowing function.
