AMERICAN SOCIETY OF BIOMECHANICS
Presented at the Twenty-First Annual Meeting
of the American Society of Biomechanics
Clemson University, South Carolina
September 24-27, 1997
|
| AMERICAN SOCIETY OF BIOMECHANICS
Presented at the Twenty-First Annual Meeting |
The correlation of spectral changes in electromyographic (EMG) recordings with corresponding force production has shown promise in recent years (2,3). Previous studies have used calculations based on Fast Fourier Transforms (FFT) to accomplish this work. This research alternatively applies wavelet transforms to surface EMG's recorded during isometric ramped force production. This mathematical technique is shown to have promise in unraveling the physiologic phenomena underlying the EMG spectral changes.
Basmajian and De Luca (1) have shown how the affects of superposition and tissue filtering join to produce a single Motor Unit Action Potential (MUAP) detected by electrodes. In additional work both De Luca (3) and Solomonow (2) have reported overall spectral shifts in the surface EMG. These shifts are attributed to the type of muscle fibers activated and may therefore be used for characterization of motor unit recruitment and muscle composition. Investigation of these shifts have been limited to changes in the median frequency of the power spectrum derived using windowed FFT. These methods, however, try to capture time varying spectral shifts, that are due to changes in the underlying irregular discrete waveforms, using continuous regular sine waves. Wavelet analysis allows investigation of these changes using irregular discrete "little waves". These are functions whose shape and duration are much more similar to an actual MUAP. By scaling and translating these "little waves" the resulting decomposition may produce information about the recruitment of the motor units of different type.
Subjects were asked to produce ramped isometric elbow flexion. A maximal force value was obtained by averaging three attempts. Using the max force value several ramps were constructed of one, three, and five second duration that ended in 25%, 50%, 75%, and 100% of maximal force. The subject was then asked to match his force output with that of the prescribed ramp. Force and EMG signals were collected at 1000 Hz. Surface EMG was collected using Delsys Inc. Differential Electrode DE-02.3H.
Typical results are shown in figure 1. Wavelet Analysis was performed using the 'db3' Daubechies wavelet shown in figure 2.
Figure 1: Ramp Force and EMG vs Time
Figure 2: The Daubechies 'db3' Wavelet
The analysis is based on a decomposition of the EMG signal using the following scaled and translated function :
[Webeditor: no graphic file was included for this equation.]
The scales a were chosen in conjunction with the sampling rate to give wavelets with a period in the 3-20 ms range. This range is reported for single human muscle action potentials. The 'db3' wavelet was chosen because of it's similarity to the MUAP, described by Basmajian and De Luca (1). The magnitude of C(a,b) is a measure of the matching of the original with the 'db3' scaled and translated wavelet. Results of the decomposition are shown in figure 3. Analysis was performed using the Matlab 5 (The Math Works, Inc.) Wavelet Toolbox. This is a gray scale plot with greater blackness reflecting increases in the coefficients absolute value. A force tracing is also provided as a reference.
Figure 3: Wavelet decomposition of EMG.
As the force increases nearly linearly with time, scales containing higher bands of frequency become activated. Near 50% of the maximum force all bands that will be activated have been. This is consistent with the findings of both Solomonow(2) and De Luca(3) who found that between 50 to 75% of MVC the spectral shifts they observed also reached an upper limit.
The spectral shifts that take place during changes in force production have generally been explained by the changes in types of motor units recruited. Fast motor units have larger propagation velocities and also appear to posses different membrane qualities that lead to the changes in spectral characteristics. While similar changes are noted using a windowed FFT approach they are more clearly and realistically documented using the wavelet approach. This is due to the wavelets ability to mimic a MUAP, the underlying building block of EMG, as opposed to the continuous sine waves utilized by FFT's. The wavelet analysis promises to be a useful tool in explaining the underlying muscle recruitment strategies that make up the EMG signal.
1. Basmajian, J.V. and De Luca, C.V. Muscles Alive. Baltimore, MD: Williams & Wilkins, 1985.
2. Solomonow, M., Baten, C., Smit, J., Baratta, R., Hermens, H., D'Ambrosia, and Shoji, H. Electromyogram power spectra frequencies associated with motor unit recruitment strategies. J. Appl. Physiol. 68(3): 1177-1185, 1990.
3. Kupa, E.J., Roy, S.H., Kandarian, S.C., and De Luca, C.V. Effects of muscle fiber type and size on EMG median frequency and conduction velocity. J. Appl. Physiol. 79(1): 23-32, 1995.