Tweet
Hi
Thankyou to all who contributed responses to my question about minimum
periods of muscle inactivity. It's nice to know that there are a few
people tackling the same hurdle.
There are some useful references contained within the summary below.
Again thankyou for your contribution.
Cheers.
Mick Dillon
ORIGINAL MESSAGE
_______________________________________________
0000,0000,ffffGood afternoon.
I am currently completing some software to analysed EMG data including
muscle on/off times. This topic has received much attention both within
the literature and Biomech-l postings but despite our best efforts
remains a dubious practice.
Periods of muscle activity have been determined in numerous ways. After
much consideration I have chosen the following 2 criteria for my
software.
1. Mean resting level + 3 SD above that mean to define a resting
threshold for each muscle.
2. The muscle must remain 'on' for a minimum of ~50 m/s. This
t-value has been specifically selected for each muscle
analysed based on previous literature.
My concern is that multiple bursts of activity which meet the above 2
criteria are interspersed with very small periods of inactivity (about
9-22 ms). The muscles most affected are Tibialis Anterior and Biceps
femoris during late swing.
I wish to set another test condition to examine the muscle 'off' times to
alleviate this problem, but have been unable to find any literature to
provide a physiological basis for a test condition.
I hope someone may be able to point me in the direction of some further
literature which defines the minimum period of muscle inactivity
between contractions.
Thank you in anticipation
Kindest regards,
Mick Dillon.
RESPONSES
_______________________________________________
Michael:
Probably the most comprehensive review of the subject is in:
1. Hodges, P.W., & Bui, B.H. (1996). A comparison of computer-based
methods for determination of onset of muscle contraction using
electromyography. Electroencephalography and Clinical Neurophysiology,
101, 511-519.
You are correct that the end of an EMG burst is more difficult to
determine. However, if you have clean EMG signals and a knowledge of
how your EMG activity relates to kinematic events, you should be able
to
develop an efficient algorithm based on the combination of the two.
Finally, it is important to develop a program that will allow you to
manually correct the points selected by the computer algorithm should
they be wrong.
Best Wishes,
David A. Gabriel
__________________________________________________ ____
Dear Mick,
I had similar dilemma with determining muscle off times. I used the
criterion for "off" that the muscle had to fall below the threshold for
at
least 100 msec in order to be considered off. The 100 msec duration
was
chosen subjectively based on the assumption that the activity I was
investigating (muscle activity around initial contact of a stride of
walking) was rather slow and that 100 msec would be between 10-15% of
the
cycle. I felt that if the muscle was inactive for that long then when
it
turned on again it was for another functional reason. I would have
ignored the little blips in between since they would likely have little
functional contribution to movement.
Also, you might be using a threshold that is too low. I had to use a
threshold of 2.5 times the resting value, not the standard deviation of
the resting value. (we were using a telemetered system with rather high
noise).
I'll be interested in the summary of responses. Will you post them?
Thanks.
************************************************** **********************
Katy Rudolph, PhD, PT Ph: 617-353-5463
Center for BioDynamics fax: 617-353-5462
Department of Biomedical Engineering krudolph@bu.edu
Boston University
44 Cummington Street
Boston, MA 01887
____________________________________________
I think that you may go to find out a paper in literature:
"Computer algorithms to characterize individual subject EMG profiles
during
gait," by Ross A. Bogey, Lee A. Barnes and Jacquelin Perry in
Arch Phys Med Rehabil Vol.73, pp.835-841, Sept 1992.
The paper would provide some interesting criteria which might fit to
your
interests.
-----
Szi-Wen Chen, Ph.D.
Post-Doctoral Researcher
1027 Dodd Hall
Gait Analysis Laboratory, Department of Surgery
The Ohio State University
480 W. 9th Ave., Columbus OH 43210
Tel: 614-293-4832
Fax: 614-293-4834
http://o2.gait.ohio-state.edu/chens
__________________________________________________
Mike-
I read your post on the biomechanics list serve. Although I can't help
you with the question you posted, I have an interest in the software
you
are developing to measure muscle onset. I am currently working on a
project looking at reflex latencies to imposed joint perturbation at
the
ankle (an inversion perturbation device) as part of a larger
investigation
considering ankle sensorimotor characheristics. Several authors have
reported with similar methodologies, however the exact methods utilized
to determine muscle onset have varied. Several have reported extremely
high reliability using several filters (high pass, low pass, followed by
a
smoothing filter) with either a visual inspection or 5SD above
background
noise to determine onset. In fear of introducing a bias through
filtering
and processing the EMG data too much I have tried using 5SD for a
minimum
of 50ms burst as the criteria for onset, however I have gotton
incredible
amounts of variablity between trials. I suspect that this may be
related
to several factors concerning the onset definitions, and other
methodological considerations (number of trials being averaged for
example). I am interested in any insight you may be able to share
based
on your experiences with the software development. Is your software
appropriate for an application in reflex latencies? Thank you in
advance
for any insight. I will be traveling over the course of the next week
so
I may be away from email temporarily. Bryan
Bryan L. Riemann, MA, ATC
Doctoral Student
Neuromuscular Research Laboratory
University of Pittsburgh
blrst15+@pitt.edu
_________________________________________________-
Michael,
I am grappling with the same problem with EMG data from cervical muscles.
I would appreciate receiving any information you gather on the topic.
Regards,
John
--
John Brault
Biomechanics Research & Consulting, Inc.
840 Apollo St., Suite 218
El Segundo, CA 90245
Phone: 310-615-3112
Fax: 310-615-3038
Email: john@brcinc.com
http://www.brcinc.com/
-------------------------------------------------------------------
To unsubscribe send UNSUBSCRIBE BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://www.bme.ccf.org/isb/biomch-l
-------------------------------------------------------------------
Thankyou to all who contributed responses to my question about minimum
periods of muscle inactivity. It's nice to know that there are a few
people tackling the same hurdle.
There are some useful references contained within the summary below.
Again thankyou for your contribution.
Cheers.
Mick Dillon
ORIGINAL MESSAGE
_______________________________________________
0000,0000,ffffGood afternoon.
I am currently completing some software to analysed EMG data including
muscle on/off times. This topic has received much attention both within
the literature and Biomech-l postings but despite our best efforts
remains a dubious practice.
Periods of muscle activity have been determined in numerous ways. After
much consideration I have chosen the following 2 criteria for my
software.
1. Mean resting level + 3 SD above that mean to define a resting
threshold for each muscle.
2. The muscle must remain 'on' for a minimum of ~50 m/s. This
t-value has been specifically selected for each muscle
analysed based on previous literature.
My concern is that multiple bursts of activity which meet the above 2
criteria are interspersed with very small periods of inactivity (about
9-22 ms). The muscles most affected are Tibialis Anterior and Biceps
femoris during late swing.
I wish to set another test condition to examine the muscle 'off' times to
alleviate this problem, but have been unable to find any literature to
provide a physiological basis for a test condition.
I hope someone may be able to point me in the direction of some further
literature which defines the minimum period of muscle inactivity
between contractions.
Thank you in anticipation
Kindest regards,
Mick Dillon.
RESPONSES
_______________________________________________
Michael:
Probably the most comprehensive review of the subject is in:
1. Hodges, P.W., & Bui, B.H. (1996). A comparison of computer-based
methods for determination of onset of muscle contraction using
electromyography. Electroencephalography and Clinical Neurophysiology,
101, 511-519.
You are correct that the end of an EMG burst is more difficult to
determine. However, if you have clean EMG signals and a knowledge of
how your EMG activity relates to kinematic events, you should be able
to
develop an efficient algorithm based on the combination of the two.
Finally, it is important to develop a program that will allow you to
manually correct the points selected by the computer algorithm should
they be wrong.
Best Wishes,
David A. Gabriel
__________________________________________________ ____
Dear Mick,
I had similar dilemma with determining muscle off times. I used the
criterion for "off" that the muscle had to fall below the threshold for
at
least 100 msec in order to be considered off. The 100 msec duration
was
chosen subjectively based on the assumption that the activity I was
investigating (muscle activity around initial contact of a stride of
walking) was rather slow and that 100 msec would be between 10-15% of
the
cycle. I felt that if the muscle was inactive for that long then when
it
turned on again it was for another functional reason. I would have
ignored the little blips in between since they would likely have little
functional contribution to movement.
Also, you might be using a threshold that is too low. I had to use a
threshold of 2.5 times the resting value, not the standard deviation of
the resting value. (we were using a telemetered system with rather high
noise).
I'll be interested in the summary of responses. Will you post them?
Thanks.
************************************************** **********************
Katy Rudolph, PhD, PT Ph: 617-353-5463
Center for BioDynamics fax: 617-353-5462
Department of Biomedical Engineering krudolph@bu.edu
Boston University
44 Cummington Street
Boston, MA 01887
____________________________________________
I think that you may go to find out a paper in literature:
"Computer algorithms to characterize individual subject EMG profiles
during
gait," by Ross A. Bogey, Lee A. Barnes and Jacquelin Perry in
Arch Phys Med Rehabil Vol.73, pp.835-841, Sept 1992.
The paper would provide some interesting criteria which might fit to
your
interests.
-----
Szi-Wen Chen, Ph.D.
Post-Doctoral Researcher
1027 Dodd Hall
Gait Analysis Laboratory, Department of Surgery
The Ohio State University
480 W. 9th Ave., Columbus OH 43210
Tel: 614-293-4832
Fax: 614-293-4834
http://o2.gait.ohio-state.edu/chens
__________________________________________________
Mike-
I read your post on the biomechanics list serve. Although I can't help
you with the question you posted, I have an interest in the software
you
are developing to measure muscle onset. I am currently working on a
project looking at reflex latencies to imposed joint perturbation at
the
ankle (an inversion perturbation device) as part of a larger
investigation
considering ankle sensorimotor characheristics. Several authors have
reported with similar methodologies, however the exact methods utilized
to determine muscle onset have varied. Several have reported extremely
high reliability using several filters (high pass, low pass, followed by
a
smoothing filter) with either a visual inspection or 5SD above
background
noise to determine onset. In fear of introducing a bias through
filtering
and processing the EMG data too much I have tried using 5SD for a
minimum
of 50ms burst as the criteria for onset, however I have gotton
incredible
amounts of variablity between trials. I suspect that this may be
related
to several factors concerning the onset definitions, and other
methodological considerations (number of trials being averaged for
example). I am interested in any insight you may be able to share
based
on your experiences with the software development. Is your software
appropriate for an application in reflex latencies? Thank you in
advance
for any insight. I will be traveling over the course of the next week
so
I may be away from email temporarily. Bryan
Bryan L. Riemann, MA, ATC
Doctoral Student
Neuromuscular Research Laboratory
University of Pittsburgh
blrst15+@pitt.edu
_________________________________________________-
Michael,
I am grappling with the same problem with EMG data from cervical muscles.
I would appreciate receiving any information you gather on the topic.
Regards,
John
--
John Brault
Biomechanics Research & Consulting, Inc.
840 Apollo St., Suite 218
El Segundo, CA 90245
Phone: 310-615-3112
Fax: 310-615-3038
Email: john@brcinc.com
http://www.brcinc.com/
-------------------------------------------------------------------
To unsubscribe send UNSUBSCRIBE BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://www.bme.ccf.org/isb/biomch-l
-------------------------------------------------------------------