View Full Version : Comments from EMG & MVC

04-15-2003, 06:53 PM
Hi All,

Many thanks for everyone’s comments and time in responding to my question.
All comments have been of great interest. Thanks again.

Here are the responses:
You have asked an important question that has not been resolved in the EMG
normalization literature. What is certian is that normalization will not give
you a standard 0 to 100% relative scale of activation, because of differences
in static an dynamic actions and normal variation.

Some favor the isometric MVC since it does factor out variations in activation
due to dynamics (changes in muscle length, velocity, position under the
electrodes), while others favor multiple position or dynamic normalization
procedures but these include unknown between trial variation into the data. It
is not clear if variation due to muscle action or maximal effort is larger and
most in need of elimination in the normaliztion process.

There is considerable reserach on this topic, but no clear consensus as to
what method is best. It may even depend on the muscle/muscle group involved.
Here are a few sources for you to look at:

Kasprisin & Grabiner 2000 Clin Biomech 15:743-749.
Vint & Hinrichs 1999 J Appl Biomech 15:210-220.
Miaki et al. 1999 Eur J Appl Physiol 80:185-191.
Burden & Bartlett 1999 Med Eng Phys 21:247-257.
Kasprisin & Grabiner 1998 J EMG Kines 8:45-50.
Bamman et al. 1997 J Strength Cond Res 11:68-72.
Zabik & Dawson 1996 Percept Mot Skills 83:976-978.
Knutson et al. 1994 J. EMG Kines 4:47-59.
Allison et al. 1993 J. EMG Kines 3:236-244.
Knudson & Johnston 1993 J Hum Mov Stud 25:39-50.
Veiersted 1991 Eur J Appl Physiol 62:91-98.

Duane Knudson, Ph.D.
Associate Chair
Department of Physical Education and Exercise Science
California State University, Chico
Chico, CA 95929-0330 USA
530-898-4932 Fax


For adult healthy subjects I believe the maximum isometric contraction
is the best reference contraction. You may have seen an article I
co-authored in 1994 in JEK. Also see Soderberg and Knutson in Physical
Therapy in 2000.

Loretta M. Knutson, PhD, PT, PCS
Professor, Physical Therapy
Southwest Missouri State University
email: LorettaKnutson@smsu.edu
phone: 417/836-8728 fax: 6229

The following reference may be of some help:
Yang JF, Winter DA.
Electromyographic amplitude normalization methods: improving their sensitivity
as diagnostic tools in gait analysis
Arch Phys Med Rehabil. 1984 Sep;65(9):517-21.

************************************************** *********************
Gordon Chalmers, Ph.D.
Dept. of Physical Education, Health and Recreation
Western Washington University
516 High St.
Bellingham, WA, U.S.A.

Phone: 360-650-3113
FAX: 360-650-7447

Email: Gordon.Chalmers at wwu dot edu
Web: www.wwu.edu/~chalmers


I've attached two PowerPoint files that I've used in recent EMG classes
at Arizona State University. One speaks directly to normalization
schemes, the other to reliability (which is strongly related). I would
suggest staying away from MVC normalization schemes because they tend to
take a lot of time and do not really give you much back in terms of
reliability. The only real benefit of MVC normalization is that it is
one of the schemes that actually allows you to assess the "level" of EMG
activity relative to some physiological capacity. Other normalization
schemes can improve reliability, but lose the ability to describe "how
active" a muscle is during a given task relative to "how active" it
could be when maximally active. Yang and Winter's work probably shows
this most conclusively.

Yang, J.F., & Winter, D.A. (1984). Electromyographic amplitude
normalization methods: Improving their sensitivity as diagnostic tools
in gait analysis. Archives of Physical Medicine and Rehabilitation, 65
(9), 517-521.

Yang, J.F., & Winter, D.A. (1983). Electromyography reliability in
maximal and submaximal isometric contractions. Archives of Physical
Medicine and Rehabilitation, 64, 417-420.

But, before you go to the trouble of choosing and implementing a
normalization scheme, I would suggest that you to define your problem or
needs more explicitly. For example, what measures are you actually
comparing? If you are comparing within subject EMG changes to different
plyometric exercises (or even have a repeated measures, within subject
experimental design), then normalization is not really necessary because
each subject's response on one condition will be compared with that same
subject's response on another condition. If, however, you want to
establish a between-day reliability of your data, then normalization may

I would be happy to help if you have additional problems.

Peter F. Vint, Ph.D.
Research Scientist
Research Integrations, Inc.
9280 S. Kyrene Rd. Suite 101
Tempe, AZ 85284
Phone: 480-893-1600 x214
Fax: 480-893-0602
e-mail: peter.vint@riimail.com


Check out the following reference for one approach to normalising EMG signals.

Authors Dolan P. Adams MA.
Institution Department of Anatomy, University of Bristol, Park Row, U.K.
Title The relationship between EMG activity and extensor moment generation
in the erector spinae muscles during bending and lifting activities.
Source Journal of Biomechanics. 26(4-5):513-22, 1993 Apr-May.
Abstract The relationship between EMG activity and extensor moment
generation in the erector spinae muscles was investigated under isometric
and concentric conditions. The full-wave rectified and averaged EMG signal
was recorded from skin-surface electrodes located over the belly of the
erector spinae at the levels of T10 and L3, and compared with measurements
of extensor moment. The effects of muscle length and contraction velocity
were studied by measuring the overall curvature (theta) and rate of change
of curvature (d theta/dt) of the lumbar spine in the sagittal plane, using
the '3-Space Isotrak' system. Isometric contractions were investigated with
the subjects pulling up on a load cell attached to the floor. Hand height
was varied to produce different amounts of lumbar flexion, as indicated by
changes in lumbar curvature. The extensor moment was found to be linearly
related to EMG activity, and the 'gradient' and 'intercept' of the
relationship were themselves dependent upon the lumbar curvature at the
time of testing. Concentric contractions were investigated with the
subjects extending from a seated toe-touching position, at various speeds,
while the torque exerted on the arm of a Cybex dynamometer was continuously
measured. Under these conditions the EMG signal (E) was higher than the
isometric signal (E0) associated with the same torque. E and E0 were
related as follows: E0 = E/(1 + A d theta/dt), where A = 0.0014 exp
(0.045P) and P = percentage lumbar flexion. This equation was used to
correct the EMG data for the effect of contraction velocity. The corrected
data were then used, in conjunction with the results of the isometric
calibrations, to calculate the extensor moment generated by the erector
spinae muscles during bending and lifting activities. The extensor moment
can itself be used to calculate the compressive force acting on the lumbar

Dear Robert: You will find that EMG activity levels achieved during
dynamic events such as jumping will exceed those during isometric
contractions. Thus, the exact meaning of a 100% isometric MVC is not
clear. If you are doing near maximal dynamic activities I would use the
highest EMG activity levels observed during all the activities that you do
as that will probably be most representative of 100% maximum EMG. We used
this to normalize finger muscle EMGs (Darling et al., Journal of
Biomechanics 27:479-491) during dynamic movements which included "as fast
as possible" movements. One problem with your EMG analysis for plyometric
activities such as drop jumps is that the impacts will cause large movement
artifacts in the EMG that you must deal with (I suggest using a high pass
filter on the raw EMG at about 30-50 Hz - see Fagenbaum and Darling,
American Journal of Sports Medicine 31:233-240).

Good luck,


Warren G. Darling, Ph.D.
Associate Professor,
Department of Exercise Science,
The University of Iowa
Iowa City, IA 52242

phone: 319-335-9514
fax: 319-335-6966

Hi, Robert. We have encountered the issue of EMG normalization during
our isometric and isokinetic studies. We have used both techniques you
described in your initial post (i.e., EMG expressed as a percentage of
the highest recorded value within the data set and expressed as a
percentage of isometric MVC). Below are four references. The quotes
from these references simply support the notion of EMG normalization,
but additional reading from these sources will provide more detailed
accounts of different normalization techniques (i.e., also expressed as
a percentage of force production or ratios thereof). I'm not sure that
it really matters what normalization procedure you use, but when
comparing between individuals, muscles, and muscle actions, it is
important that you do normalize.

The specific procedure you use is somewhat dependent upon the number of
data points (EMG amplitude values within subjects) you wish to
normalize. If you have many data points, then expressing your EMG
amplitude values as a percentage of the highest recorded value would be
fine. If you are analyzing a limited number of data points for each
subject, then you may want to consider normalizing to an external value,
such as a percentage of an isometric MVC. In addition, there have been
studies that have expressed EMG amplitude values as a ratio of the force
produced during the EMG signal epoch.

Hopefully the following information will be helpful:

The absolute amplitude of the EMG signal is influenced by many factors
including the subcutaneous tissue between the electrode and muscle,
muscle architecture, fiber type, and fiber diameter. Thus, EMG data
used for comparisons between individuals, muscles or activities should
be normalized. Basmajian and DeLuca (Muscles Alive, 5th ed., 1985, p.
77) have stated:

"A generalized representation of the EMG signal must contain a
formulation which allows a comparison of the signal between different
muscles and individuals. This is not a problem in some contractions,
such as those involving ballistic movements. However, it is a
requirement in isometric and anisometric contractions. The formulation
for comparison may be obtained by normalizing the variables of the EMG
signal with respect to their maximal measurable value in the particular
experimental procedure."

LeVeau and Andersson (Output forms Data analysis and applications,
Interpretation of the electromyographic signal. In: Soderberg, G.L.
(ed.) Selected Topic in Surface Electromyography for Use in Occupational
Settings: Expert Perspectives. U.S. Department of Health and Human
Services, Publication No. 91-100, 1992, p. 70) have stated:

"Quantification of the myoelectric signal, although not the goal, is
done so that comparisons may be made among muscles, individuals, and
activities. The myoelectric signal amplitude is used as an indirect
measure of contraction-force. Because there is not a one-to-one
relationship between the two, a standard of reference must be
established for any comparison among subjects, muscles, or activities.
Such a process is referred to as normalization."

Cram and Kasman (Introduction to Surface Electromyography. Aspen
Publishers Inc. 1998, p. 62) have stated:

"Comparison of sEMG values both within and between individuals is
potentially fraught with problems. Anthropomorphic differences between
different recording sites and between individuals suggest cautious use
of such comparisons. Some of the factors that might affect these
comparisons include: thickness of subcutaneous adipose tissue, muscle
mass/cross-sectional area, fiber type, age, sex, subtle changes in
posture, interelectrode distance, and impedance of the skin. The
effects of the various anthropomorphic moderating variables on
comparison between individuals are reviewed in some depth in Chapter 5.
The bottom line is that it is possible to compare RMS values across
muscle sties for the resting baseline conditions only. Population
statistics for the sites of comparison are extremely valuable. However,
during dynamic movements, comparison of amplitude measurements alone
(ie, peak RMS) across muscle groups can be very misleading without first
normalizing the sEMG data.
Researchers and practitioners have attempted to deal with these
issues. One technique used to control for these variables is called

Recently, Soderberg and Knutson (A Guide for Use and Interpretation of
Kinesiologic Electromyographic Data. Physical Therapy. 80:5 p. 492)
summarized the need for normalization by stating:

"Either raw or processed versions of data can be entered into the
normalization module, which allows for the process of referencing the
EMG data to some standard value, usually by dividing the derived EMG
data by a reference value. The decision to normalize or not normalize
is based on the type of descriptions or comparisons to be made. For
example, if comparisons are made between subjects, days, muscles, or
studies, the process is required."

Good luck!


Joel T. Cramer
University of Nebraska-Lincoln
Department of Health and Human Performance
Center for Youth Fitness and Sports Research
145 Mabel Lee Hall
Lincoln, NE 68588-0229
Phone: (402) 472-3846
Fax: (402) 472-4305
E-mail: jcramer@unlserve.unl.edu
Web: http://tc.unl.edu/jtcramer/


Hi Robert,

I am not exactly sure of your research design, I would think that using a
condition as the reference (100%) and making all other condition a percentage
of that condition would probably be best. We have had a number of cycling EMG
studies published that have used this method. Also, I have a paper in review
at the moment that has been rejected a number of times due to the fact that I
have compared EMG from concentric and eccentric contractions together.
Apparently, due to signal cancellation issues diferent muscle action types
should not be directly compared against each other (Day & Hullioger 2001, J
Neurophysiol 86:2144-2158). If you want to get your study published once
finished I suggest that you don't normalise against an isometric max,
normalise against a reference or standard depth jump even if you need to
include a new jump in your protocol solely for this purpose, it will only add
a few minites to your total testing time.

Hope this helps,

Jack Cannon
Postgraduate Student
Human Movement Studies Unit
Faculty of Education
Charles Sturt University
Bathurst, 2795
Ph: +61 2 6338 4334
Fax: +61 2 6338 4065

Hi Robert,
I have looked at the issue of amplitude normalisation for the SSC and
abdominal muscles.
Esentially there are three positions adopted by researchers.
A. Express it as a % of MVIC (generally independent of motor strategy
hence the maximal obtained in different actions. [Ref])
Most common form (by researchers who utilise the EMG signal as a
mapping of the force) hence you can compare data with other research
raw data results.
Easiest conceptually - avoids the conceptual problem of having an
amplitude of 125% - easier to describe
Easiest to perform prior to activity - most people understand MVIC.
Mid range isometric not generally pattern specific
Inappropriate (my opinion) in situations where there is pain or
potential inhibition i.e LBP groups.
Usually highly variable (within subject performance between sessions)
- adds to variability in the data.
Not good for low level activities.
Mid range isometric - difficult correlates with different lengths and actions.

B. Submaximal normalisation
Same as MVIC but a standard load is used that reflect the similar
amplitude of the task being studied.
For example a set load held against gravity.
Easy to perform - most people understand.
Better reliability within and between sessions [ref]
Appropriate where there may be pain inhibition of maximal performance
Good for low amplitude activities
Mid range isometric not generally pattern specific
Mid range isometric - difficult correlates with different lengths and actions.
Some people have problems interpreting the amplitude scale.

C. Peak or mean amplitude derived from the activity.
easy to determine.
May not be independent of the derived variable of investigation.
Pattern specific
Difficult to compare between muscles

How to choose which is best amplitude normalisation technique?

There is probably no correct answer but at least you can justify what
you pick by using the literature. ...

My suggestions.
1. Understand the derived variable you are drawing from the EMG
signal is it changed by different amp norm techniques? [ Does mm,
length, action difference etc in the Amp norm impact on the derived
2. The Amp Norm technique has to be reliable
3. The Amp Norm technique (unless using MVIC) should be similar to
the amp you are observing.
4. A reduction in coefficient of variation (CV) by itself is not the
single best indicator of the the value of amplitude normalisation.
5. a Chnge in the statistical power of the derived variable of a
known clinical difference is the key to the quantification of Amp
Norm assessments.

I mention these in some of my publications.

How you process the EMG signal is also important to understand the
variability in the data - for example the reliability of the peak
signals is related to how hard the data are filtered.

Good luck.

Garry T Allison Associate Professor of Physiotherapy
The Centre for Musculoskeletal Studies http://www.cms.uwa.edu.au/
School of Surgery and Pathology, The University of Western Australia.
Level 2 Medical Research Foundation Building
Rear 50 Murray Street
Perth Western Australia 6000.
ph: (618) 9224 0219
Fax (618) 9224 0204

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l