Simi Motion module: EMG wavelet analysis
 This module contains software developed by BIOMECHANIGG RESEARCH INC., Calgary, Canada, which reserves all rights.
Frequency analysis of EMG signals can be carried out using wavelets. This opens up completely new possibilities for interpreting and analyzing EMG signals.
For further information, please see the scientific publications on this subject.
Theoretical aspects
There are a few theoretical aspects that are essential to understand when using the present software.
Heisenberg’s Uncertainty Relationship
The uncertainty relationship states that one cannot measure a frequency, and the time at which the frequency occurs, simultaneously with infinite accuracy. The product of time-resolution and bandwidth is a constant close to one. Thus a compromise must always be made. The compromise adopted for this software was based on physiologically relevant time resolutions.
Comparison With Classical Wavelet Analysis
Wavelet analysis is still a new subject. Classical analysis uses linearly scaled wavelets. This leads to un-physiologically short time resolutions at higher frequencies. The theory of wavelet transforms has therefore been modified and adapted for the purpose of analyzing EMG’s using non-linearly scaled wavelets. The wavelets at higher frequencies have more wiggles, thus resolving areas where the EMG has an oscillatory pattern over a time period similar to time resolution. Shorter spike like signals will be detected with a much lower intensity.
The Concept of Intensity
A wavelet transformed EMG yields a convolved signal reflecting the oscillation of the EMG and of the wavelets. e.g. a spike like EMG signal would be seen as a short oscillation after a classical transformation. However, for the analysis of EMG’s it is important to obtain an intensity that reflects the power extracted from the EMG by the wavelets. The present software uses the convolved signal and its derivative to compute an intensity, which is proportional to the power in the original EMG. The proportionality was set such that if the EMG was a pure sine wave of amplitude “A”, the total intensity (intensity summed over all wavelets) would approximate “A2”. The computation of the intensity represents a valid approximation of the power in the EMG signal. The square of root mean square (RMS) values used in classical EMG analysis are nearly equal to the total intensity. However, compared to the RMS taken over a similar time period, the intensity reveals more detail. For special cases, e.g. spikes, the RMS and the total intensity show very different results. The intensity is proportional to the power and is therefore additive, which allows easy processing of intensity patterns.
von Tscharner V.,
Intensity analysis in time-frequency space of surface myoelectric signals by wavelets of specified resolution,
Journal of Electromyography & Kinesiology, 10(6): 433-445, 2000.
Wakeling JM, Pascual SA, Nigg BM, von Tscharner V.,
Surface EMG shows distinct populations of muscle activity when measured during sustained sub-maximal exercise.
European Journal of Applied Physiology, 2001;86:40-47.
Von Tscharner V.,
Time-frequency and principalcomponent methods for the analysis of EMG’s recorded during a mildly fatiguing exercise on a cycle ergometer.
Journal Electromyography & Kinesiology, 2002, in press.
Univ. of Calgary: [Annual Report 2001]
|
