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 Power Density Spectrum (PDS) of the EMG was shown to provide important information on various muscle properties such as fatigue, force production process (motor unit recruitment), fiber composition and changes with skill acquisition. While the PDS is based on statistical analysis which has its inherent variability, other factors such as power line, environmental and system noises were responsible for introduction of additional variability which prevents scientists from developing logical conclusions as to the muscles mechanical properties. For example, during low level contractions, typically in the 0 - 30% maximal voluntary contraction (MVC) force, the signal to noise ratio is very low, and the noises PDS is dominant, introducing significant artifacts into the conclusion.
Furthermore, the production of MVC can not be taken at its face value since most subjects generate only 70 - 80% MVC without training. Calculations of PDS when considering the "untrained" MVC as true MVC, further creates an environment for additional artifactual conclusions which are not correct or logical.
The objectives of this report is to describe three methods that can significantly increase the reliability of PDS data, and restore the confidence in this analytical tool.
The first methods consists of recording a short segment of the EMG at rest (e.g. 0% MVC), calculating the PDS of this segment which mostly includes noises of various sources. The PDS of the rest period is then subtracted from the PDS of each EMG epoch during active contraction, such that the resulting PDS excludes any components due to noise of any sources.
The second method consists of estimating the amplitude and phase of a noisy periodic waveform that may contaminate the EMG baseline during rest (e.g. 0% MVC). The periodic waveform, e.g. a sinusoid, could be than subtracted from the active EMG to obtain a clean signal from analysis.
Training subjects to obtain their true MVC consists of first asking a subject to produce his maximal force contraction while displaying it on the monitor as a moving line on a scale. The experimenter than sets a new line, 10% higher than the last obtained by the subject, than asking him to exceed it. The procedure is repeated until the subject can not exceed the new goal, and his last (highest obtained) force level is set as MVC.
Figure 1 shows the Median Frequency (MF) of the PDS of a subject performing linearly increasing isometric force contraction from 0 to 100% MVC over 3 seconds. The curve consisting of the circle data points is from calculations of the MF of the PDS without eliminating the noise, whereas the curve consisting of the square data points is from calculations in which the noises PDS was subtracted from each epoch of the EMG's PDS. Clearly, one can see that the "clean" MF trace has a near linear increase up to about 80% MVC whereas the unclean MF curve is erratic, showing decrease in the low force range of 0 -25% MVC, and then an increase.
Figure 2 shows the EMG, PDS and MF of the EMG obtained in a mild isometric elbow flexion immediately after a rest (0% MVC). The left column shows a low level cyclic noise prior to the contraction, and the associated MF of the PDS at 57 Hz. The right column shows the EMG after subtracting the power line noise according to Method 2, and the associated MF of the PDS at 52 Hz. Clearly, the noise resulted in an erroneous increase of 5 Hz in the estimation of the MF.
The MF of the PDS as a function of force before and after training a subject to generate his true MVC shows that nearly 37% additional force was obtained with a completely different curve of MF progression during the force increase.
Three Methods were described, with an example for each, that could significantly reduce the variability and components of the PDS of the EMG during force production of a muscle. The Methods allow scientist to use the PDS as an analytical tool to provide precise insights into the force production mechanism of a skeletal muscle free of artifactual inputs from environmental noise and subject/experimenter error. It is conceived that the described techniques will restore lost confidence in the PDS as a tool to assess muscle properties.
