Processing EMG
David DeLion
UNLV Biomechanics Lab
Why do we process EMG?
• Raw EMG offers us valuable
information in a practically useless form
• Raw EMG signals cannot be
quantitatively compared between
subjects
• If electrodes are moved raw EMG
signals cannot be quantitatively
compared for the same subject
Types of Signal
Processing
•
•
•
•
•
•
•
Raw
Half-wave rectified
Full-wave rectified
Filtering
Averaging
Smoothing
Integration
•
•
•
•
Root-mean Square
Frequency spectrum
Fatigue analysis
Number of Zerocrossings
• Amplitude
Probability
Distribution Function
• Wavelet
Removing Bias
• Low amplitude voltage offset present in
hardware
• Can be AC or DC
• Calculate the mean of all the data
• Subtract mean from each data point
Raw EMG
• Unprocessed signal
-Amplitude of 0-6 mV
-Frequency of 10-500 Hz
• Peak-to-Peak
-Measured in mV
-Represents the amount of muscle
energy measured
• Onset times can be determined
• Analysis is mostly qualitative
Rectification
• Only positive values are analyzed
-Mean would be zero
• Half-wave rectification - all negative
data is discarded, positive data is kept.
• Full-wave rectification- the absolute
value of each data point is used
• Full-wave is preferred
Filtering
• Notch filter
-Band reject filter; usually very
narrow
-For EMG normally set from 59-61
Hz
-Used to remove 60 Hz electrical
noise
-Also removes real data!
-Too much noise will overwhelm the
filter
Filtering
• Band Pass filter
-allows specified frequencies to
pass
-low end cutoff removes electrical
noise associated with wire sway and
biological artifacts
-high end cutoff eliminates tissue
noise at the electrode site
-often set between 20-300 Hz
Filtering
• There are no perfect filters!
• Face muscles can emit frequencies up
to 500 Hz
• Heart rate artifact can be eliminated
with low end cutoffs of 100 Hz
• Filters which include 60 Hz include the
noise from equipment
Averaging
• Average EMG can be used to quantify
muscle activity over time
• Measured in mV
• Values are averaged over a specified time
window
• Window can be moved or static
• Moving windows are a digital smoothing
technique
• For moving windows the smaller the time
window the less smooth the data will be
Averaging
• For EMG window is typically between 100200 ms
• Window is moved over the length of the
sample
• Moving averages introduce a phase shift
• Moving averages create biased values
-values are calculated from data
which are common to the data used to
calculate the previous value
• Very commonly used technique
Integration
• Calculation of area under the rectified signal
• Measured in Vs
• Values are summed over the specified time
then divided by the total number of values
• Values will increase continuously over time
• The integrated average will represent 0.637
of one-half of the peak to peak value
• Quantifies muscle activity
• Can be reset over a specified time or voltage
Root Mean Square
• Recommended quantification method
by Basmajian and DeLuca
• Calculated by squaring each data point,
summing the squares, dividing the sum
by the number of observations, and
taking the square root
• Represents 0.707 of one half of the
peak-to-peak value
Number of Zero Crossings
• Counting the number of times the
amplitude of the signal crosses the zero
line
• Based on the idea that a more active
muscle will generate more action
potentials, which will cause more zero
crossings in the signal
• Primarily used before the FFT algorithm
was widely available
Frequency Analysis
• Fast Fourrie Transformation is used to
break the EMG signal into its frequency
components.
• Frequency components are graphed as
function of the probability of their
occurrence
• Useful in determining cutoff frequencies
and muscle fatigue
Fatigue Analysis
• Isometric contraction
• The two most important parameters for
fatigue analysis are the median and
mean frequency.
• Median frequency decreases with the
onset of fatigue
• If fatigue is being measured it is
important to have a large band pass
filter
Amplitude Probability
Distribution Function
• Illustrate variance in the signal
• X-axis shows range of amplitudes
• Y-axis shows the percentage of time
spent at any given amplitude
• Distribution during work should be
bimodal -peak associated with effort
-peak associated with rest
Wavelet analysis
• Used for the processing of signals that are
non-stationary and time varying
• Wavelets are parts of functions or any
function consists of an infinite number of
wavelets
• The goal is to express the signal as a linear
combination of a set of functions
• Obtained by running a wavelet of a given
frequency through the original signal
Wavelet Analysis
• This process creates wavelet coefficients
• When an adequate number of coefficients
have been calculated the signal can be
accurately reconstructed
• The signal is reconstructed as a linear
combination of the basis functions which are
weighted by the wavelet coefficients
Wavelet Analysis
• Time-frequency localization
• Most of the energy of the wavelet is restricted
to a finite time interval
• Fourier transform is band limited
• Produces good frequency localization at low
frequencies, and good time localization at
high frequencies
• Segments, or tiles the time-frequency plane
Wavelet Analysis
• Wavelets remove noise from the signal
• Signal energy becomes concentrated into
fewer coefficients while noise energy does
not
Normalizing
• There is no absolute scale so direct
comparisons between subjects or
conditions cannot be made
• Maximum voluntary contraction levels
are often used to compare EMG
readings between subjects (i.e..50%
MVC)
• Relies on subject to give max effort
Normalizing
• Record contractions over a dynamic
movement cycle
• At least 4 repetitions are required
• Peak values are averaged which
creates an anchor point
• Subsequent values are represented as
a percentage of the anchor point
Conclusion
• EMG offers a great deal of useful
information
• The information is only useful if it can
be quantified
• Quantifying EMG data can be a
qualitative process
Thank You
Any Questions?
Bibliography
• Kleissen, R.F.M, Buurke, J.H., Harlaar, J., Zilvold, G. (1998)
Electromyography in the Biomechanical analysis of human
movement and its clinical application. Gait and Posture. Vol.
8,143-158
• Aminoff, M.J. (1978) Electromyography in Clinical Practice.
Addison-Wesley Publishing Company, Menlo Park, CA
• Dainty, D.A., Norman, R.W. (1987) Standardized Biomechanical
Testing in Sport. Human Kinetics Publishers, Champaign, IL
• Cram, J.R., Kasman, G.S. (1998) Introduction to Surface
Electromyography. Aspen Publishers, Gaithersburg, MD
• Medved, V. (2001) Measurement of Human Locomotion. CRC
Press, New York, NY
Bibliography
• Loeb, G.E., Gans, C. (1986) Electromyography for
Experimentalists. The University of Chicago Press, Chicago, IL
• Basmajian, J.V., DeLuca, C.J. (1985) Muscles Alive. Williams
& Wilkins, Baltimore, MD
• Moshou, D., Hostens, I., Ramon, H. (2000) Wavelets and SelfOrganizing Maps in Electromyogram Analysis. Katholieke
Universiteit Leuven
• DeLuca, C.J. (1993) The Use of Surface Electromyography in
Biomechanics. NeuroMuscular Research Center, Boston
University
• DeLuca, C.J. (2002) Surface Electromyography: Detection and
Recording. Delsys Incorporated.