Tweet
Many thanks to all that replied to my sEMG and the Rectus Femoris question, the responses were invaluable. I have posted a summary of all the responses to date below, as well as my original question. Since these are quite long I’ve also summarized the main points from the answers received.
The main points were:
Skin preparation: a general consensus that we needed to use more extensive skin preparation (i.e. at least abrading before cleaning, to remove dead skin cells), although people’s views on whether shaving is necessary varied. People also tape/bandage down the electrodes to minimise motion artefact.
Adjustable gains: increasing the gain of the signals from all thigh muscles was suggested, to increase the SNR (unfortunately we are not able to do this with the fixed gain system we have).
Checking the electrode function: it was suggested to tap/fiddle with the electrodes once in position, whilst checking the EMG signal to check the amount of interference generated.
Placement checks: both passive movement and active movements to elicit isolated muscle activity were mentioned. Pushing down on the subjects shoulders in standing (knees flexed and resisting) was one of the precise suggestions.
The Rectus should show a unique burst of activity just after toe off that is not seen from the vasti, it was suggested to record from both the RF and vasti enabling any crosstalk to be better identified. Alternatively, use fine-wire electrodes and then compare output to sEMG to check placement and crosstalk.
Thanks again
Emma
---------------------------------------------------------------------------------------------------------------------------
Original Question:
At our clinical gait lab we have an 8 channel Aurion Zerowire surface EMG system, and have recently started to collect sEMG data during gait to form our normal adult database. So far we have struggled to collect ‘clean’ sEMG data from the Rectus Femoris (RF), and were wondering if anyone with experience of using sEMG for measuring timing of the quadriceps could offer us some advice/tips etc.
A brief summary of our protocol - We used the SENIAM guidelines for electrode placement, but simply prepped the skin using alcohol wipes. We checked our electrode position (extend knee from sitting position with resistance) and got strong signals, however data during gait had a lot of noise associated with it and no distinct pattern. The sampling rate was 2000Hz, data was bandwidth filtered 20-500Hz , and we displayed both raw and linear enveloped data (25ms bins, GCV woltring filter). As an aside, we are currently completing work looking at the affect of different filtering and cut-off parameters but are using the above as a starting point.
The signals seen could of course be because the ‘normals’ tested don’t use their RF much, however this raises a very important point for us. Our main clinical aim is to use sEMG is to know when (and indeed if) the RF and hamstrings are firing in the gait cycle. How can we be confident enough in our data to conclude this?
I have done a literature search on the use of sEMG to measure RF. So far I have read Zipp et al. 1982, Rainoldi et al. 2004, Ounpuu et al. 1997 and Byrne et al. 2005. There are a few contradictions, and the latter two papers have different conclusions whether sEMG measurement of RF activity is possible reliably. There does seem evidence to change our placement check for RF to being voluntary hip flexion in the sitting position (against resistance) if this is possible for the patient. We could also consider shaving patients legs (although we would prefer not to) and measuring the vasti (lateralis or medialis) instead if these produce more reliable signals practically (which is suggested by Rainoldis work).
Any help/thoughts would be much appreciated.
-------------------------------------------------------------------------------------------------------
Stephanie Di Stasi (udel.edu)
You may be having difficulty because of the gains you are using to amplify your signal. When using MVICs to collect our RF data (patient is seated, knee flexed to 60 degrees), we tune our gains so that the digital signal we see is approximately two-thirds of the max signal that we can get without clipping. Increasing gains on the thigh muscles, including hamstrings, has allowed us good signal to noise ratios during collection and post-processing. In general, as you've mentioned, the %MVC of the quads and hamstrings that our patients use is lower than that of the shank musculature, but the thigh muscles are certainly used during level walking gait.
-----------------------------------------------------------------------------------------------
Luke R Garceau (Dept of Physical Therapy, Marquette University, WI)
We use a more extensive skin preparation protocol. We shave the skin (if necessary), abrade the skin (with a coarse skin abrasion pad), clean with an alcohol pad, allow skin to dry, place the electrode, and then apply elastic tape to minimize motion artifact and provide strain relief for the electrode cables.
--------------------------------------------------------------------------------------------------------
CK Mummidisetty (Rehab, Inst of Chicago)
We do dry shave, clean it with alcohol wipes. Placing the electrode (Motion lab systems) and sticking it to skin surface millipore medical grade tape. Upon placing the electrode, we tap them to see the signals (checking electrode function). To confirm the right placement, we try to elict the isolated muscle activity (by making passive movements). Once confirmed, we wrap the electrode with a co-band (limiting electrode movement and reliable signal from muscle while walking).
--------------------------------------------------------------------------------------------------------------
Derek Rutherford (dal.ca)
We do find that the amplitudes of RF are much lower than than of the vastii but also evident is a unique burst just after toe off indicating we are capturing an aspect of hip flexor function that isn't apparent in the vastii electromyogram.
See doi:10.1016/j.joca.2010.01.005
Just a couple of clarifying questions. I'm not sure what you mean by "clean". Have you looked at the frequency spectrum?
While placement guidelines will get you to the right place, skin preparation is essential for artifact free signals. This usually involves shaving the skin to ensure adequate coupling of the electrode gel (Ag/AgCl) with the skin (removal of hair and dead skin cells). Also cleaning with an an alcohol swab or distilled water prior to application of the electrode. You may find that this will help improve your signal
to noise tremendously.
----------------------------------------------------------------------------------------------------
At Hof (Centre for Human Movement Sciences, University of Groningen)
An important paper on RF function is:
Nene, A., Byrne, C., Hermens, H., 2004. Is rectus femoris really a part of quadriceps? Assessment of rectus femoris function during gait in able-bodied adults. Gait & Posture 20, 1-13.
It is shown there that you can certainly see a distinct activity pattern in RF. For knee extension, vastus medialis is more relevant. Shaving does not improve your signal, only gives better adhesion of the electrodes.
-------------------------------------------------------------------------------------------------------
Jeff Bingham (Neuroengineering lab, Georgia Institute of Technology)
Perhaps I misread your post. Are you shaving the area under the EMG electrode? Hair between the electrode and the skin can cause a myriad of problems.
Is the noise that you are observing consistent with what you would expect if the electrode had an increased impedance? Is there motion artifact? Looking at the raw EMG before it is filtered may give you some insight into whether a measurement problem exists.
We have been successful recording surface EMG from rectus femoris when the skin is clean (no hair) and using disposable gel electrodes. We also wrap the electrode and leg with ace bandage to reduce relative motion of the electrode and the skin.
-------------------------------------------------------------------------------------------------------
Jacqueline Romkes (Laboratory for Movement Analysis, Basel)
We also measure surface emg of rectus femoris in our gaitlab with every patient. Since our majority of patients have CP, this muscle is important for us to see if activity in mid-swing is present. I suggest measuring one of the vasti together with rectus. Since rectus usually has crosstalk with vasti. Rectus is active around toe-off whereas vasti are active around foot strike and not around toe-off. Also see the work of Nene et al. (Gait and Posture 20:1-13, 2004).
We clean the skin with alcohol and shave the skin where the electrodes are place. We also use SENIAM as a guideline for electrode placement.
We have good signals of the rectus but in healthy controls this might also be very low. See the work of Nene where he looked at the influence of speed on the muscle activity as well.
-------------------------------------------------------------------------------------------------------
Edmund Cramp (Motion Lab Systems, Inc)
It's very hard to make any detailed comments without actually seeing the EMG signals that you're getting.
But it's a common problem - generally RF doesn't do much in "normal" gait so the signals are going to be very small when normal subjects walk. EMG QA needs to be able to see the "live, real-time" EMG signal in detail - I don't know what facilities you have to do this but generally I find that motion capture systems provide only a cursory "live" analog monitoring system.
The first step would be to verify that they electrodes are placed correctly - pushing down on the subjects shoulders while they are standing (knees slightly bent and resisting) should generate a decent RF signal so that you know that your set up is good. Then fiddle with the electrodes and associated amp and check that motion doesn't generate excessive interference and you should be OK - and thus have a high degree of confidence that the signals that you record, as the subject walks, are accurate.
I suspect the basic problem is that the Aurion - like so many of these "fixed gain" EMG systems - simply doesn't have enough gain to amplify low level EMG signals.
DISCLAIMER: We make and sell EMG systems with variable gain controls precisely to avoid this type of problem - you get exactly the same issues when working with CP, amputee subjects, etc when monitoring the activity in atrophied muscles.
Our users generally just turn the gain up on these channels. As a general rule, the largest EMG signals in normal gait are from the muscle closest to the floor - the EMG signal decreases in amplitude as you get further away from the floor.
----------------------------------------------------------------------------------------------------------------------------
Marilynn Wyatt (Biomechanics Laboratory, Naval Medical Center San Diego)
I do not believe you can get isolated RF data with a surface electrode. At the least, you get crosstalk with the Vastus Intermedius and most likely with the V. medius and V. Lateralis. I recommend that if you want pure Rectus Femoris data, you use a fine-wire electrode. With the surface electrode you are getting "group" data.
--------------------------------------------------------------------------------------------------------------------------
Graham Webb (University of Surrey)
Is it anatomical or a characteristic of using wireless DSP? The bandwidth should be ok, even with bluetooth, but you could look at the spectra from one wireless sensor whilst adding more sensors, to see if the DSP compresses the signal. It is only 16-bit (presumably before transmission). Also compare the spectra from wireless vs. wired for the same reason.
---------------------------------------------------------------------------------------------------------------------------
Ross Bogey (Rehabilitation Institute of Chicago) and Lee Barnes (B&L Engineering)
"The EMG signal from the RF can be difficult to separate from the other quadriceps muscles, particularly when acquiring data with surface electrodes We obtained RF data with indwelling (wire) electrodes from more than a dozen normal adults during self-selected speed walking.
There was little variability across normal subjects. The relative intensity of the signal during its activity period was typically between 15-20 percent of maximum voluntary contraction. Thus the force generated is not trivial. The onset of Rectus Femoris EMG was consistently just before ipsilateral toe off (occurred at 59% of the gait cycle (GC), toe off at 62%GC), and cessation occurred about 70 milliseconds later (66% GC). This is in contrast with the vasti muscles, where activity was limited to onset just before the end of swing phase, and continuing through the first few percent of the next ipsilateral stance phase.
We have not tried to establish norms with surface electrodes. One approach might be to use the RF wire EMG signal as a reference, and assess various electrode configurations with the goal of having the surface EMG signal for this muscle be a close match to the wire EMG signal..."
__________________________________________________ _______________
http://clk.atdmt.com/UKM/go/197222280/direct/01/
Do you have a story that started on Hotmail? Tell us now
The main points were:
Skin preparation: a general consensus that we needed to use more extensive skin preparation (i.e. at least abrading before cleaning, to remove dead skin cells), although people’s views on whether shaving is necessary varied. People also tape/bandage down the electrodes to minimise motion artefact.
Adjustable gains: increasing the gain of the signals from all thigh muscles was suggested, to increase the SNR (unfortunately we are not able to do this with the fixed gain system we have).
Checking the electrode function: it was suggested to tap/fiddle with the electrodes once in position, whilst checking the EMG signal to check the amount of interference generated.
Placement checks: both passive movement and active movements to elicit isolated muscle activity were mentioned. Pushing down on the subjects shoulders in standing (knees flexed and resisting) was one of the precise suggestions.
The Rectus should show a unique burst of activity just after toe off that is not seen from the vasti, it was suggested to record from both the RF and vasti enabling any crosstalk to be better identified. Alternatively, use fine-wire electrodes and then compare output to sEMG to check placement and crosstalk.
Thanks again
Emma
---------------------------------------------------------------------------------------------------------------------------
Original Question:
At our clinical gait lab we have an 8 channel Aurion Zerowire surface EMG system, and have recently started to collect sEMG data during gait to form our normal adult database. So far we have struggled to collect ‘clean’ sEMG data from the Rectus Femoris (RF), and were wondering if anyone with experience of using sEMG for measuring timing of the quadriceps could offer us some advice/tips etc.
A brief summary of our protocol - We used the SENIAM guidelines for electrode placement, but simply prepped the skin using alcohol wipes. We checked our electrode position (extend knee from sitting position with resistance) and got strong signals, however data during gait had a lot of noise associated with it and no distinct pattern. The sampling rate was 2000Hz, data was bandwidth filtered 20-500Hz , and we displayed both raw and linear enveloped data (25ms bins, GCV woltring filter). As an aside, we are currently completing work looking at the affect of different filtering and cut-off parameters but are using the above as a starting point.
The signals seen could of course be because the ‘normals’ tested don’t use their RF much, however this raises a very important point for us. Our main clinical aim is to use sEMG is to know when (and indeed if) the RF and hamstrings are firing in the gait cycle. How can we be confident enough in our data to conclude this?
I have done a literature search on the use of sEMG to measure RF. So far I have read Zipp et al. 1982, Rainoldi et al. 2004, Ounpuu et al. 1997 and Byrne et al. 2005. There are a few contradictions, and the latter two papers have different conclusions whether sEMG measurement of RF activity is possible reliably. There does seem evidence to change our placement check for RF to being voluntary hip flexion in the sitting position (against resistance) if this is possible for the patient. We could also consider shaving patients legs (although we would prefer not to) and measuring the vasti (lateralis or medialis) instead if these produce more reliable signals practically (which is suggested by Rainoldis work).
Any help/thoughts would be much appreciated.
-------------------------------------------------------------------------------------------------------
Stephanie Di Stasi (udel.edu)
You may be having difficulty because of the gains you are using to amplify your signal. When using MVICs to collect our RF data (patient is seated, knee flexed to 60 degrees), we tune our gains so that the digital signal we see is approximately two-thirds of the max signal that we can get without clipping. Increasing gains on the thigh muscles, including hamstrings, has allowed us good signal to noise ratios during collection and post-processing. In general, as you've mentioned, the %MVC of the quads and hamstrings that our patients use is lower than that of the shank musculature, but the thigh muscles are certainly used during level walking gait.
-----------------------------------------------------------------------------------------------
Luke R Garceau (Dept of Physical Therapy, Marquette University, WI)
We use a more extensive skin preparation protocol. We shave the skin (if necessary), abrade the skin (with a coarse skin abrasion pad), clean with an alcohol pad, allow skin to dry, place the electrode, and then apply elastic tape to minimize motion artifact and provide strain relief for the electrode cables.
--------------------------------------------------------------------------------------------------------
CK Mummidisetty (Rehab, Inst of Chicago)
We do dry shave, clean it with alcohol wipes. Placing the electrode (Motion lab systems) and sticking it to skin surface millipore medical grade tape. Upon placing the electrode, we tap them to see the signals (checking electrode function). To confirm the right placement, we try to elict the isolated muscle activity (by making passive movements). Once confirmed, we wrap the electrode with a co-band (limiting electrode movement and reliable signal from muscle while walking).
--------------------------------------------------------------------------------------------------------------
Derek Rutherford (dal.ca)
We do find that the amplitudes of RF are much lower than than of the vastii but also evident is a unique burst just after toe off indicating we are capturing an aspect of hip flexor function that isn't apparent in the vastii electromyogram.
See doi:10.1016/j.joca.2010.01.005
Just a couple of clarifying questions. I'm not sure what you mean by "clean". Have you looked at the frequency spectrum?
While placement guidelines will get you to the right place, skin preparation is essential for artifact free signals. This usually involves shaving the skin to ensure adequate coupling of the electrode gel (Ag/AgCl) with the skin (removal of hair and dead skin cells). Also cleaning with an an alcohol swab or distilled water prior to application of the electrode. You may find that this will help improve your signal
to noise tremendously.
----------------------------------------------------------------------------------------------------
At Hof (Centre for Human Movement Sciences, University of Groningen)
An important paper on RF function is:
Nene, A., Byrne, C., Hermens, H., 2004. Is rectus femoris really a part of quadriceps? Assessment of rectus femoris function during gait in able-bodied adults. Gait & Posture 20, 1-13.
It is shown there that you can certainly see a distinct activity pattern in RF. For knee extension, vastus medialis is more relevant. Shaving does not improve your signal, only gives better adhesion of the electrodes.
-------------------------------------------------------------------------------------------------------
Jeff Bingham (Neuroengineering lab, Georgia Institute of Technology)
Perhaps I misread your post. Are you shaving the area under the EMG electrode? Hair between the electrode and the skin can cause a myriad of problems.
Is the noise that you are observing consistent with what you would expect if the electrode had an increased impedance? Is there motion artifact? Looking at the raw EMG before it is filtered may give you some insight into whether a measurement problem exists.
We have been successful recording surface EMG from rectus femoris when the skin is clean (no hair) and using disposable gel electrodes. We also wrap the electrode and leg with ace bandage to reduce relative motion of the electrode and the skin.
-------------------------------------------------------------------------------------------------------
Jacqueline Romkes (Laboratory for Movement Analysis, Basel)
We also measure surface emg of rectus femoris in our gaitlab with every patient. Since our majority of patients have CP, this muscle is important for us to see if activity in mid-swing is present. I suggest measuring one of the vasti together with rectus. Since rectus usually has crosstalk with vasti. Rectus is active around toe-off whereas vasti are active around foot strike and not around toe-off. Also see the work of Nene et al. (Gait and Posture 20:1-13, 2004).
We clean the skin with alcohol and shave the skin where the electrodes are place. We also use SENIAM as a guideline for electrode placement.
We have good signals of the rectus but in healthy controls this might also be very low. See the work of Nene where he looked at the influence of speed on the muscle activity as well.
-------------------------------------------------------------------------------------------------------
Edmund Cramp (Motion Lab Systems, Inc)
It's very hard to make any detailed comments without actually seeing the EMG signals that you're getting.
But it's a common problem - generally RF doesn't do much in "normal" gait so the signals are going to be very small when normal subjects walk. EMG QA needs to be able to see the "live, real-time" EMG signal in detail - I don't know what facilities you have to do this but generally I find that motion capture systems provide only a cursory "live" analog monitoring system.
The first step would be to verify that they electrodes are placed correctly - pushing down on the subjects shoulders while they are standing (knees slightly bent and resisting) should generate a decent RF signal so that you know that your set up is good. Then fiddle with the electrodes and associated amp and check that motion doesn't generate excessive interference and you should be OK - and thus have a high degree of confidence that the signals that you record, as the subject walks, are accurate.
I suspect the basic problem is that the Aurion - like so many of these "fixed gain" EMG systems - simply doesn't have enough gain to amplify low level EMG signals.
DISCLAIMER: We make and sell EMG systems with variable gain controls precisely to avoid this type of problem - you get exactly the same issues when working with CP, amputee subjects, etc when monitoring the activity in atrophied muscles.
Our users generally just turn the gain up on these channels. As a general rule, the largest EMG signals in normal gait are from the muscle closest to the floor - the EMG signal decreases in amplitude as you get further away from the floor.
----------------------------------------------------------------------------------------------------------------------------
Marilynn Wyatt (Biomechanics Laboratory, Naval Medical Center San Diego)
I do not believe you can get isolated RF data with a surface electrode. At the least, you get crosstalk with the Vastus Intermedius and most likely with the V. medius and V. Lateralis. I recommend that if you want pure Rectus Femoris data, you use a fine-wire electrode. With the surface electrode you are getting "group" data.
--------------------------------------------------------------------------------------------------------------------------
Graham Webb (University of Surrey)
Is it anatomical or a characteristic of using wireless DSP? The bandwidth should be ok, even with bluetooth, but you could look at the spectra from one wireless sensor whilst adding more sensors, to see if the DSP compresses the signal. It is only 16-bit (presumably before transmission). Also compare the spectra from wireless vs. wired for the same reason.
---------------------------------------------------------------------------------------------------------------------------
Ross Bogey (Rehabilitation Institute of Chicago) and Lee Barnes (B&L Engineering)
"The EMG signal from the RF can be difficult to separate from the other quadriceps muscles, particularly when acquiring data with surface electrodes We obtained RF data with indwelling (wire) electrodes from more than a dozen normal adults during self-selected speed walking.
There was little variability across normal subjects. The relative intensity of the signal during its activity period was typically between 15-20 percent of maximum voluntary contraction. Thus the force generated is not trivial. The onset of Rectus Femoris EMG was consistently just before ipsilateral toe off (occurred at 59% of the gait cycle (GC), toe off at 62%GC), and cessation occurred about 70 milliseconds later (66% GC). This is in contrast with the vasti muscles, where activity was limited to onset just before the end of swing phase, and continuing through the first few percent of the next ipsilateral stance phase.
We have not tried to establish norms with surface electrodes. One approach might be to use the RF wire EMG signal as a reference, and assess various electrode configurations with the goal of having the surface EMG signal for this muscle be a close match to the wire EMG signal..."
__________________________________________________ _______________
http://clk.atdmt.com/UKM/go/197222280/direct/01/
Do you have a story that started on Hotmail? Tell us now