Temporal spectral approach to surface electromyography based fatigue classification of biceps brachii during dynamic contraction
Muscle fatigue is defined as a reduction in muscle’s ability to contract and produce force due to prolonged submaximal exercise. Since fatigue is not a physical variable, fatigue indices are commonly used to detect and monitor muscle fatigue development. One suggested approach to quantitative measur...
Saved in:
Main Author: | |
---|---|
Format: | Thesis |
Language: | English |
Published: |
2016
|
Subjects: | |
Online Access: | http://eprints.utm.my/id/eprint/77841/1/NurulAsyikinKamaruddinMFKE2016.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Muscle fatigue is defined as a reduction in muscle’s ability to contract and produce force due to prolonged submaximal exercise. Since fatigue is not a physical variable, fatigue indices are commonly used to detect and monitor muscle fatigue development. One suggested approach to quantitative measurement of muscle fatigue is based on surface electromyography (sEMG) signal. Short-Time Fourier Transform (STFT) and Continuous Wavelet Transform (CWT) are commonly used techniques to obtain time-frequency representation of sEMG signals. However, S Transform (ST) technique has not been applied much to physiological signals. No found literature has used ST technique to extract muscle fatigue indices. Thus, this study intends to determine the feasibility of using ST technique to extract muscle fatigue indices from sEMG signal. Thirty college students with no illness history were randomly selected to perform bicep curl activities for 130 seconds while holding a 2 kg dumbbell. Using the three time-frequency techniques (STFT, CWT, and ST), four commonly extracted muscle fatigue indices (Instantaneous Energy Distribution (IED), Instantaneous Mean Frequency (IMNF), Instantaneous Frequency Variance (IFV) and Instantaneous Normalize Spectral Moment (INSM)) were extracted from the acquired biceps sEMG signals. Indices from fatigue signals were found to be significantly different (p-value < 0.05) from the non-fatigue signals. Based on the Normalization of Root Mean Square Error (NRMSE) and Relative Error, ST technique was found to produce less error than STFT and CWT techniques in extracting muscle fatigue indices. Through the use of 3-fold cross validation procedure and with the help of Support Vector Machine (SVM) classifier, IMNF-IED-IFV was selected as the best feature combination for classifying the two phases of muscle fatigue with consistent classification performance (accuracy, sensitivity and specificity) of 80%. Therefore, this study concludes that ST processing technique is feasible to be applied to sEMG signals for extracting screening or monitoring measures of muscle fatigue with a good degree of certainty. |
---|