Page 1 of 1

Strange activity on a single electrode

Posted: Fri Feb 21, 2014 8:23 pm
by Clint
We are recording EEG using a Biosemi ActiveTwo box (64 channel cap).
All of our equipment is either new or recently repaired, so I don't think we have a technical problem. We have had several successful sessions. Every once in a while, though, we get a strange signal from one or two electrodes. I've attached a jpeg - the electrode in question is P2.
No matter what we have done, we cannot get this to go away. We have made sure that all connections are clean, that all offsets are low, that everything we can think of is controlled. Sometimes, it just goes away on its own, but it seems to take quite a while.
Any ideas? What is this? Is it a problem? If so, can it be fixed?

Thanks!!

Re: Strange activity on a single electrode

Posted: Sat Feb 22, 2014 10:34 am
by Coen
Gel bridge between P2 and (adjacent) DRL causes 50/60 Hz interference. Use less gel to prevent gel bridges between the electrodes.

Best regards, Coen (BioSemi)

Re: Strange activity on a single electrode

Posted: Tue Nov 28, 2017 7:57 pm
by pcp15srl
Hi,

I'm responding to this old post because we have had the exact same issue during data collection in our lab (high frequency noise in P2, almost certainly due to bridging between P2 and DRL).

I'm just wondering whether this bridging introduces noise back into the entire electrode set (due to noise from P2 being fed into DRL), or whether this problem is just confined to P2 (whereby the only solution would be to remove P2 during analysis)?

Just having some trouble conceptualizing what effect this type of bridging has on the dataset as a whole.

Best wishes,

Sam

Re: Strange activity on a single electrode

Posted: Wed Nov 29, 2017 1:27 pm
by Coen
Only the electrode(s) shorted to the DRL are affected.

Refer to the schematic in https://www.biosemi.com/pics/zero_ref1_big.gif . A gel bridge between an electrode and the DRL can be represented in the schematic as a short between the electrode and the DRL opamp output. The DRL will still drive the subject as close as possible to the AMP/DAC reference voltage. However, the shorted electrode will now measure the Vcm (Common Mode voltage) instead of the voltage at the body surface.

From the schematic also follows:

- a gel bridge between an electrode and CMS causes a flat line (just amp noise) for that electrode in the unreferenced signal display
- a gel bridge between CMS and DRL reduces the CMRR with 40 dB at 50 Hz (equal hum on all channels)

Best regards, Coen (BioSemi)

Re: Strange activity on a single electrode

Posted: Mon Dec 11, 2017 2:13 pm
by pcp15srl
Hi Coen,

This is really useful, thanks a lot for your help!

Best wishes,

Sam

Re: Strange activity on a single electrode

Posted: Thu Sep 06, 2018 4:01 pm
by Douwehorsthuis
Hi,
I was wondering how problematic this issue is.
It happens to us for a decent amount of participants, yet when I look at the preproccesed data, I see that both P2 and P1 are almost never excluded. Does this mean that when we look at the data during the data collection and see these noisy/flat channels, it doesn't imply that these channels should be considered "bad'?


Thank you for all this information!
Douwe

Re: Strange activity on a single electrode

Posted: Sat Sep 08, 2018 10:10 am
by Coen
No, they are bad.

Best regards, Coen (BioSemi)

Re: Strange activity on a single electrode

Posted: Sat Sep 08, 2018 2:46 pm
by Douwehorsthuis
That is very interesting. Would you have an idea why matlab (or EEG lab) would not exclude these channels all the time then?
Thanks again,
Douwe

Re: Strange activity on a single electrode

Posted: Tue Sep 18, 2018 7:19 pm
by Douwehorsthuis
Is it possible that when we reference (lets say to the mastoids) in MATLAB we add this information to those flat channels? after this maybe the channels do not look flat anymore, but contain mastoid information that shouldn't be there?

Re: Strange activity on a single electrode

Posted: Wed Sep 19, 2018 5:07 pm
by Coen
Certainly, if you subtract a reference from a flat channel than the result is the (inverted) reference with respect to CMS.

Best regards, Coen (BioSemi)

Re: Strange activity on a single electrode

Posted: Wed Jun 09, 2021 2:35 pm
by Douwehorsthuis
I know BioSemi suggests re-referencing the data (https://www.biosemi.com/faq/cms&drl.htm). Would it matter if this re-referencing happens at the beginning or later in the pre-processing pipeline? (EEGlab allows you to directly apply references to your data when you start working with a raw BDF file).

As an example, I was working on an old dataset where I wasn't sure if we had mastoid references, so I couldn't do it at the start. This made it clear how many more channels were flat that I would normally never notice. Which then made me think that it might be better to first delete bad and flat channels and only re-reference after that.

Re: Strange activity on a single electrode

Posted: Thu Jun 10, 2021 3:27 pm
by Coen
You should indeed first remove the flat/bad channels, and then reference to one or more of the remaining valid channels.

Best regards, Coen (BioSemi)

Re: Strange activity on a single electrode

Posted: Wed Dec 22, 2021 8:53 pm
by Douwehorsthuis
Would an average reference at this point do the same? Or does it need to be a reference to a electrode/external

Re: Strange activity on a single electrode

Posted: Thu Dec 23, 2021 1:59 pm
by Coen
You can use an average reference (from 2 electrodes to all electrodes), as long as you make sure that the average only contains good/valid channels.

Best regards, Coen (BioSemi)

Re: Strange activity on a single electrode

Posted: Thu Dec 23, 2021 3:47 pm
by Douwehorsthuis
Wonderful, happy holidays