PDA

View Full Version : SUMMARY: Active Electrode Advice Needed



Gordon Chalmers
06-03-1998, 01:29 AM
Thanks very much to all of those who took the time to address my question.
I received a lot of good advice, which has solved my problem, without the
need for active electrodes. Below I list (1) my original question, (2) the
result I obtained when I tried the different suggested solutions & (3) the
individual responses I received.

To answer a question asked to me by a few of the people who responded: YES,
I am using differential amplifiers. Also, I should have mentioned that the
data was hardware high pass filtered at 10Hz (& low pass filtered at 1kHz)
as it was collected.
*********************************************
(1) MY ORIGINAL QUESTION

I have been collecting EMG from the Soleus muscle while subjects walk on a
treadmill. On some subjects there is an oscillation in the EMG baseline
immediately around the time of heel strike, which is large enough to make
the EMG level useless. The electrodes are firmly attached and sway of the
wires is minimized as much as I see is possible. I have not used active
electrodes before, but I assume that one of their purposes is to boost the
EMG signal well beyond this noise level.
I use standard passive disc electrodes, Grass Amplifiers, and Cambridge
Electronic Designs A/D system.

So here are three questions:
1) Will the use of active electrodes increase the signal to noise
(movement artifact) ratio, and so eliminate my problem of the baseline
oscillation?
2) What active electrodes are commercially available to be added to an
existing system such as mine (i.e., feed a signal into my Grass amps).
3) Any other comments or suggestions people have?


*********************************************
(2) THE RESULT I OBTAINED WHEN I TRIED THE DIFFERENT SUGGESTED SOLUTIONS.

My conclusions are:

(A) While I thought I was firmly attaching the electrodes, they were not
firm enough. Increasing the tension on the tape & increasing tape length
alone was sufficient to largely eliminate my problem (while using the 10 hz
high pass filter I had in place).

(B) The use of a small plate (very firmly attached) to keep the
interelectrode distance constant, as suggested in some responses, was able
to further reduce the low frequency noise component (as indicated by power
spectrum analysis) compared to the result with firm tape alone.

(C) Future data collection will include on line power spectrum analysis to
check for lower freq noise so problems can be identified and corrected
immediately.


(3) THE RESPONSES I RECEIVED

*********************************************
Date: Wed, 27 May 1998 17:05:37 +0200
From: Armin Kibele
Organization: Institut fr Sport und Sportwissenschaft
MIME-Version: 1.0
To: Gordon Chalmers
Subject: Re: Active Electrode Advice Needed
Status:

Before buying new expensive electrodes try the following:

1) Tap the body of your electrodes and check the emg signal. Sometimes
electrode paste does create funny artifacts. If you get a response
without any muscular activity, something wrong with the attachment of
the electrode via electrode paste or with the attachment of the cables
to the skin.
2) Attach all loose electrode cable firmly to the skin.
3) Make sure, your system is properly ground connected.


*********************************************
Date: Wed, 27 May 1998 14:03:05 -0300 (ADT)
From: Dave Grimshire
To: Gordon Chalmers
Subject: Re: Active Electrode Advice Needed
MIME-Version: 1.0
Status:

Gordon,
I'm no expert but this sounds like artifact noise. It occurs when wires
move through the earths magnetic field. The solution is to make sure your
wires move as little as possible. Also keep them away from magnetic fields
produced by motors etc.
Hopefully you are collecting using a differential amplifier. This means
that signals generated on the two input wires will be ignored while
signals only on one of the input leads will be amplified. Hope this helps.

Dave Grimshire
Technician
School of Health and Human Performance
Dalhousie University


*********************************************

I saw your message on EMG artifact and have a couple of comments. However,
I'm not completely disinterested since my company makes and sells EMG
preamplifiers. I've been installing gait analysis systems with EMG and
force plates for quite a while so I've seen most of these problems from the
users end...

My guess it that there are a couple of possible causes for the artifact that
you're seeing - both, as you guess, deriving from the use of passive
electrodes:
1. The relatively high impedance of the passive electrodes may change
substantially as the subjects heel hits the treadmill and this change of
impedance appears as artifact.
2. The leads connecting the electrodes to the Grass Amplifiers are probably
moving suddenly as the subject strike the treadmill. The leads are moving
in an electro-magnetic field and generate EMF which appears as artifact.

You ask "Will the use of active electrodes increase the signal to noise
(movement artifact) ratio, and so eliminate my problem of the baseline
oscillation?"

In general you should find that an active electrode will perform better than
the passive electrodes since the signal is amplified close to the source and
the majority of the cable to the recording system is carrying a high level
signal (~20-100mV) at a low impedance (400-1000 ohms). Currently your EMG
cables to the Grass amplifiers are probably carrying a signal in the 1-2mV
range with an impedance of 10-60,000 ohms.

Note however that EMG pre-amplifiers are subject to a couple of constraints:
Most EMG preamplifiers (a.k.a. active electrodes) come pre-packaged in a
hard plastic shell with stainless steel sensor surfaces - it's very
important that the pre-amplifier is well taped to the body so that the
sensor do not come off the skin surface during movement. If they do come
off the surface, or change the contact impedance, then you're no better off
then before.
You can attach passive electrodes (disposable EKG electrodes work well) to
the sensor surfaces of many preamplifiers which gives you the best of both
worlds (electrodes that stay put and preamplifiers) at the cost of a little
more mess and complexity. This is really the only way to use pre-amplified
electrodes on upper-body muscles since they're difficult to attach to the
skin if you can't wrap tape around a limb.

Since the preamplified electrodes boost the EMG signal considerably you're
likely to have problems connecting them directly to the Grass amplifiers.

There's a list of various EMG preamp and other gait related manufacturers on
our links page at www.emgsrus.com

Good Luck,
Edmund Cramp,
Motion Lab Systems, Inc.
4326 Pine Park Drive,
Baton Rouge, LA 70809 USA
+1 504 928-4248 (voice, 2 lines)
+1 504 928-0261 (fax)
My email address is eac@emgsrus.com
For information about Motion Lab Systems please visit our web site at
http://www.emgsrus.com
*********************************************
From: Brian Davis
To: "'Gordon Chalmers'"
Subject: RE: Active Electrode Advice Needed
Date: Wed, 27 May 1998 13:28:57 -0400
Mime-Version: 1.0
Status:

Hi Gordon

When Jeannie McCrory was doing her Ph.D. at Penn State university (in
the Center for Locomotion Studies) she had problems with static
electricity building up on the treadmill belt. This interferred with her
EMG data. I think they solved the problem why lightly wetting the
belt.

Regards, Brian Davis

*********************************************
Date: Wed, 27 May 1998 13:29:46 -0400 (EDT)
From: James Dowling
To: Gordon Chalmers
Subject: Re: Active Electrode Advice Needed
MIME-Version: 1.0
Status:

Hello Gordon,
Active electrodes will increase the signal to movement artifact ratio
but it may not increase it enough for your application. I would suggest a
highpass filter of at least fifth order with a cutoff frequency of 10 Hz
as the best initial solution to your problem. I have very good active
electrodes (DELSYS) and I sometimes have to filter the data anyway. If
you have trouble implementing a suitable highpass filter for EMG, let me
know and I can send you a copy of mine.

*********************************************
Mime-Version: 1.0
X-Sender: glg@acs-mail.bu.edu
Date: Wed, 27 May 1998 13:39:17 -0400
To: Gordon Chalmers
From: "Gerald L. Gottlieb"
Subject: Re: Active Electrode Advice Needed
Status:


>1) Will the use of active electrodes increase the signal to noise
>(movement artifact) ratio, and so eliminate my problem of the baseline
>oscillation?
>
In principal they should be increasing the input inpedance and improving
common mode rejection.

>2) What active electrodes are commercially available to be added to an
>existing system such as mine (i.e., feed a signal into my Grass amps)?
>
I have been useing Delsys (formerly liberty mutual) and am pleased with
their performance.

>3) Any other comments or suggestions people have?
>
The key question is whatis the bandwidth of your amplified signal? It
sounds like you have the high-pass cut-off too low. It should be above 20
hz and 50 hz is better. There is no EMG signal below there, just
mechanical artifact of various sources. getting the right band pass might
solve the problem without changing electrodes.
__________________________________________________ ___
| Gerald Gottlieb (617) 353-8984 beauty |
| NeuroMuscular Research Center 353-9757 is |
| Boston University fax 353-5737 truth |
| 44 Cummington St. etc, etc, etc. |
| Boston MA 02215 /\ http://nmrc.bu.edu/MCL/glg.html |
|_______________________/\ / \ /\_______________
|

*********************************************
X-Sender: wdarling@blue.weeg.uiowa.edu
Date: Wed, 27 May 1998 13:17:10 -0500
To: Gordon Chalmers
From: Warren Darling
Subject: Re: Active Electrode Advice Needed
Mime-Version: 1.0
Status:

I have some suggestions after your questions below. Good Luck.

Warren Darling

At 09:14 AM 5/27/98 -0700, you wrote:
> I have been collecting EMG from the Soleus muscle while subjects walk on
>a treadmill. On some subjects there is an oscillation in the EMG baseline
>immediately around the time of heel strike, which is large enough to make
>the EMG level useless. The electrodes are firmly attached and sway of the
>wires is minimized as much as I see is possible. I have not used active
>electrodes before, but I assume that one of their purposes is to boost the
>EMG signal well beyond this noise level.
> I use standard passive disc electrodes, Grass Amplifiers, and Cambridge
>Electronic Designs A/D system.
>
>So here are three questions:
>1) Will the use of active electrodes increase the signal to noise
>(movement artifact) ratio, and so eliminate my problem of the baseline
>oscillation?

It should definitely help reduce the problem, if not eliminate it.

>
>2) What active electrodes are commercially available to be added to an
>existing system such as mine (i.e., feed a signal into my Grass amps)?
>
There are a number of active electrodes available, some do not even need
your Grass amps because they provide an onsite amplification of 1,000
(e.g., Delsys electrodes).

>3) Any other comments or suggestions people have?

You can try twisting together the wires of the two electrodes over soleus.
This sometimes helps reduce such movement artifact.


*********************************************
Date: Wed, 27 May 1998 14:23:42 -0500 (CDT)
X-Sender: runtech@popd.ix.netcom.com (Unverified)
Mime-Version: 1.0
To: Gordon Chalmers
From: RUN Technologies
Subject: Re: Active Electrode Advice Needed
X-Attachments: D:\RUNCUST\CHALMER1.GIF;
Status:

Dear Dr. Chalmers,

RUN Technologies produces a software package named Datapac III that may be
able to help you. Datapac III is a collection of software modules designed
for many different applications. One of them is a signal processing module
that, among other functions, offers a high pass filter that can eliminate
low frequency artifact from your signal. The filter has adjustable cutoff
and rolloff characteristics, so you can shape it to exactly suit your needs.
To illustrate its use, I have attached a gif graphics file that shows an EMG
signal containing a significant amount of low frequency movement artifact,
both before and after the application of a 20 Hz high pass filter with a
steep rolloff (approx 50 dB/oct).

In addition to its own acquisition capabilities, Datapac III can accept
almost any form of data file, whether it is in ASCII, binary, or floating
point format.
Thus, it may offer a very cost-effective solution for you. If you like, I
would be happy to demonstrate what it can do with your data files. Please
let me know if you would like to set something up. Additional information
about Datapac III is also available on our web site (http://www.runtech.com)
Thank you.

Sincerely,

Richard W. Lambert




Attachment converted: Macintosh HD:CHALMER1.GIF (GIFf/JVWR) (000203CD)
RUN Technologies
25622 Rolling Hills Road
Laguna Hills, CA 92653 USA
Phone/Fax: (949) 348-1234
email: Main@runtech.com
web site: http://www.runtech.com

Please visit our web site for product information and prices, as well as the
latest program updates and demonstration software.

*********************************************
Date: Thu, 28 May 1998 09:12:36 +0800 (WST)
Mime-Version: 1.0
To: Gordon Chalmers ,
From: iallison@info.curtin.edu.au (Garry Allison)
Subject: Re: Active Electrode Advice Needed
Status:

>Dr Gordon Chalmers - re low frequency artefact around heel strike

It still seems that there is a low frequency artefact in the signal. In the
simplest sense it could be movement artefact yet you have noted is unlikely
from observation of the leads etc - it could be skin movement (or the
muscle under the skin).
I'm not clear on how you are collecting & processing the data.- here are
some things to consider.
If you have access to the raw data then you may wish to consider the power
spectrum to see if it is truely a low frequency problem - then it may be
advantageous to increase your high pass cut off filter frequency - Lower
your low pass to 1/3rd Sampling Freq.

Another point that may not be raised by others and could be worth an
outside guess...
Some digital processing packages offer the ability to increase the order of
the filter - this has the advantage of making the cut-off frequency profile
much sharper. The disadvantage is that they become unstable and develop a
ripple effect in the processed data. This is easily seen if you run a
square pulse through the processing program and is exaggerated with "high
impulse" data.
The fact that the ripple is just before heal strike is a little more
difficult to explain - however the digital processes pass the data in both
directions so it is possible to have it before and after the actual
impulse.

Finally, if you are creating a linear envelop (leaky integrators) before
the AD conversion the raw data may be clipping (too high a gain) which is
then mis-interpreted as (intermittent - corresponding with expected high
amplitudes) low frequency noise of large amplitude wiping out a signal with
was previously seen to be normal.

Hope this helps
cheers
Garry.

------------------------------------------------------------------------
Dr Garry T Allison Lecturer in Functional Rehabilitation,
School of Physiotherapy, email:iallison@info.curtin.edu.au
Curtin University of Technology, Tel. +61 8 9266 3648
Selby Street, Shenton Park, Fax. +61 8 9266 3636
Western Australia 6008
http://www.curtin.edu.au/curtin/dept/physio/pt/staff/allison/
================================================== ======================

*********************************************
Date: Sat, 04 Jan 1997 09:34:25 +0800
From: JerJunn Luh
Reply-To: d81068@me.ee.ntu.edu.tw
MIME-Version: 1.0
To: Gordon Chalmers
Subject: Re: Active Electrode Advice Needed
Status:

>

Dear Sir:

I used the active electrodes which made by the Motion Control Inc. before.
Now I am developing a new type of active electrodes in our Lab.
The followings are my opions about active electrodes:


> So here are three questions:
> 1) Will the use of active electrodes increase the signal to noise
> (movement artifact) ratio, and so eliminate my problem of the baseline
> oscillation?

Yes, the active electrodes can improve the S/N ratio. The EMG signals canbe
amplified by the pre-amplifier which attached on the electrodes. That means
the noise from cable movement could be reduced. The passive electrodes can
apply on the skin more firmly. The motion artefact between the skin and the
passive electrode disc
is less than the artefact between the skin and the active electrode. This
effect could
be reduced when you apply the conductive jelly between the electrode and skin,
and apply a constant
pressure on the electrodes.


> 2) What active electrodes are commercially available to be added to an
> existing system such as mine (i.e., feed a signal into my Grass amps)?
>

You may contact with the Motion Control Inc.http://www.utaharm.com

> 3) Any other comments or suggestions people have?

The noise from the cable moving is hard to eliminate. Generally we have to
used a high pass filterto reduce the effect of this problem. Some frequency
band (DC-10 or 20) shoud not be used in your
study. I am sorry that I do not have the useful reference on my hand now.


Sincerely yours
Jer-Junn Luh
Ph.D. candidate,
Dept. Electrical Eng.
National Taiwan Univ.
Taipei, Taiwan.


*********************************************
X-Sender: tiburon@pop.flash.net
Mime-Version: 1.0
Date: Wed, 27 May 1998 23:19:55 -0500
To: chalmers@cc.wwu.edu
From: James Carollo
Subject: EMG Active Electrodes
Status:

The low frequency base-line shift you are experiencing is motion artifact
induced by a mechanical displacement of one recording electrode relative to
the other in the bipolar pair, as a direct result of the impact wave
propagating up the posterior calf. The secret to eliminating this is to
link your two electrodes together (like a bar electrode), and securely
fastening the entire low mass bar assembly to the calf. The electrodes
should be close together (2 to 4 cm, center-to-center), have small active
area (1cm or less, to reduce crosstalk), and be fixed rigidly to the bar.

Linking the electrodes together increases the probability that when the
impact wave arrives, both electrodes will be displaced together and by the
same amount. If you use a differential recording amplifier with a high
Common mode rejection (CMRR), the unavoidable displacement of the bar will
produce the same voltage change in both electrodes simultaneously, and the
amplifier will ignore it.

Active electrodes generally show lower motion artifact, but it is usually
because they include the 2 EMG electrodes in the assembly, not necessarily
because they are preamplified. The amplifiers do increase the signal/noise
ratio, but the net effect is to reduce high frequency or 60 cycle noise
induced from the environment. This type of noise can generally be removed
just as well by twisting the 2 lead wires together soon after they leave
the electrode, continuing all the way back to the main amplifier.

I'd try these tips before investing in active EMG electrodes that may not
be necessary.

Best Regards,
-JJC

+----------------------------------------------------------------------+
| James Carollo, Ph.D., P.E. james.carollo@email.swmed.edu
| Senior Research Scientist
| Mobility Research and Assessment Laboratory
| The University of Texas
| Southwestern Medical Center at Dallas
| 9705 Harry Hines Blvd., Suite 105
| Dallas, TX USA 75208 (214) 351-2041
+----------------------------------------------------------------------+

*********************************************
Subject: Re: Active Electrode Advice Needed
Date: Thu, 28 May 98 13:41:58 -0000
From: Ippei Akiya
To: "Gordon Chalmers"
Mime-Version: 1.0
Status:

Hello Dr. Gordon Chalmers
I have a experience that record EMG from soleus during jumping. In that
time, I did not seem the problem of baseline oscillation. Explaining my
method, I used a passive disc electrodes (5mm diameter) and 10mm inter
electrodes distance. I think that EMG in narrow inter electrodes distance
condition include smaller artifacts, since a changing of relative
electrode positions is little.
And I was setting a filter of 0.03sec time constants and 500Hz high cut
filter. It is standard setting, I think you know it.

Thanks
Ippei Akiya
Nagoya University

*********************************************
X-Sender: gdeluca@90.0.0.1
Date: Thu, 28 May 1998 10:26:31 -0400
To: chalmers@cc.wwu.edu
From: Gianluca De Luca
Subject: Active Electrodes Needed
Mime-Version: 1.0
Status:

In response to your recent posting on the Biomch newsgroup:

You are quite correct in hypothesizing that active electrodes increase the
signal to noise ratio. It is important to used effective electrode-skin
attachment methods. There are several commercial interfaces which are
specifically designed to remove motion artifact. Another point to consider
is the bandwidth of the system you are using. Motion artifact tends to
appear in the lower frequencies. If you limit the lower corner of your
bandpass to something like 20 Hz, much (not all) of this noise will be
eliminated. Naturally, this would be a problem if you are interested in
viewing the spectrum below 20 Hz.

I suggest that you look at the Delsys Inc. home page (www.delsys.com). You
may find what you are looking for. The electrodes advertised at this site
can be used with the amplifiers you described. Adhesive interfaces
designed specifically for these electrodes are also available.

Hope this helps.

Gianluca De Luca

*********************************************
From: GSRASH01@ULKYVM.LOUISVILLE.EDU
Date: 28 May 1998 13:23:19 EDT
To:
Subject: EMG
Status:


The active electrode should help your signal to noise & decrease your movement
artifact, but it doesn't eliminate it. Use a high pass filter at 15Hz & the
movement artifact do to movement of the electrode will go away. If you
don't have software to do it, we have some shareware my lab programer wrote
which will do this & much more with the EMG.

*********************************************
> So here are three questions:
> 1) Will the use of active electrodes increase the signal to noise
> (movement artifact) ratio, and so eliminate my problem of the baseline
> oscillation?

No. Active electrodes can greatly reduce noise and mains (60 Hz)
interference, not electrode artifacts. But see 3)

> 2) What active electrodes are commercially available to be added to an
> existing system such as mine (i.e., feed a signal into my Grass amps)?
?>
> 3) Any other comments or suggestions people have?
Electrode artifacts are low frequency. The main solution is to
raise the high-pass cut-off frequency of the unprocessed EMG, try 10
Hz. Myself I use 20 or 30 Hz, which gives really artifact free EMGs,
but the reviewers of some papers don't like it.

At Hof
Vakgroep Medische Fysiologie
University of Groningen
Bloemsingel 10 |----/\/\/\/----| |\
NL-9712 KZ GRONINGEN