Related Experiment Videos
Surface myoelectric signal classification for prostheses control
1School of Engineering, University of Sharjah, Sharjah, UAE. yassaf@ausharjah.edu
Journal of Medical Engineering & Technology
|August 30, 2005
Summary
This study explores surface myoelectric signal analysis for movement classification. Combining principal components analysis, wavelet analysis, and artificial neural networks achieved a 5.1% error rate for elbow and wrist movements.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Machine Learning
Background:
- Surface myoelectric signals are crucial for understanding muscle activity.
- Accurate segmentation and classification of these signals are challenging but essential for prosthetics and diagnostics.
Purpose of the Study:
- To investigate and improve methods for segmenting and classifying surface myoelectric signals.
- To evaluate the effectiveness of a combined approach using classical and advanced signal processing techniques.
Main Methods:
- Signal segmentation using moving average, principal components analysis (PCA), and time-frequency analysis.
- Feature extraction via multiresolution wavelet analysis.
- Classification using artificial neural networks (ANNs).
Main Results:
- The developed methodology successfully classified four types of elbow and wrist movements.
- A classification error rate of 5.1% was achieved using two signal channels from biceps and triceps muscles.
Conclusions:
- The combination of PCA, wavelet analysis, and ANNs provides an effective framework for myoelectric signal classification.
- The approach demonstrates potential for accurate real-time movement recognition in biomedical applications.