EEG Cap Kit noise issue / EEG+EMG+EOG / AGND [resolved]
stuartjake
UK
We've received our spanking new EEG Cap Kit, Cyton Board and Daisy Board and have signals coming from all the electrodes.
We've set an EEG, EOG and EMG montage for a drowsiness study. We've run into a bit of difficulty as the output in the OpenBCI GUI isn't a signal we expected and at a bit of a loss as to what to check, change or try next. We contacted support and Sam suggested a post on here. Hopefully you can help.
@wjcroft @retiutut @evaesteban
We're trying to determine if:
1) We have the EEG Cap setup correctly (we are getting output)
2) We have Notch 50Hz (UK) switched on, are there any other settings we can try. (I've described the setup below)
3) We are ideally looking at the frequency up to ~30Hz but note the BP filters don't come with a dropdown for that option. Noted that other posts have the option for programming. As non-programmers, looking for a pointer towards a library we could add?
I've attached 3 screen shots of the output in the GUI and an extract of the raw data. We'd really appreciate any insight please. When I run the sample data, the output is more of what we would expect. Our just doesn't look right.
As it is a drowsiness study, we are using the following setup:
On the EEG cap we are using (syringe used to inject gel into each electrode "well") :
REF (Y splitter plugged into bottom pin of both Cyton SRB and Daisy SRB)
GND (Bottom BIAS pin of Cyton board)
O1 (N1P)
O2 (N2P)
F3 (N3P)
F4 (N4P)
C3 (N5P)
C4 (N6P)
We have additional electrodes from the OpenBCI shop for the EMG and EOG
The EMG config is using the OpenBCI EMG guide, I have added:
2 electrodes on the daisy board. Top and bottom of N1P of the daisy board. (We have toggled "OFF" SRB2 for the EMG channel in the GUI). And one electrode on the daisy board AGND.
The EOG config is on N2P and N3P and have relied on the GND of the EEG Cap.
Comments
Stuart, hi.
Your last GUI screenshot "6 electrodes", shows substantial noise on channel 2. And the appearance suggests mains noise, 50 Hz in the UK. Even though you have the notch filter turned on, this would be a sign that you are getting interference from nearby AC electrical equipment. These can be in the forms of AC transformers, extension cords, conduits running inside walls / floors / ceilings, power adapters, monitors, etc.
The suggestion would be to try looking for above conditions, and move EEG setup away from such gear.
No electrodes should be attached to 'AGND' pin. Only channel pins, SRBx pin, Bias. Daisy board does use a 'Y-cable' between SRB2 pins on Daisy and mainboard. Perhaps you mis-connected to AGND. Y cable not needed for Bias since that is bridged automatically.
William
Thanks @wjcroft. Will try for a cleaner electrical environment.
Just a clarification. The OpenBCI EMG config adds "...After you have done this, connect one electrode cable to the Cyton bottom AGND pin." https://docs.openbci.com/docs/01GettingStarted/02-Biosensing-Setups/EMGSetup. Sam from support quoted the same.
Instead of the AGND pin. Where do I attach the 3rd electrode of the EMG config? I have 2 electrodes on chin muscle NP1 (top and bottom pins) plus the reference electrode on the chin bone.
Thanks again.
The page you linked to, labels AGND as 'reference'. That is not an accurate term. It is a Ground connection. In differential mode (using both plus and minus channel input pins), they act as their own self reference. In this case the Ground connection acts to help center the differential amplifier, and for the ADS1299, this is generally done with the Bias pin. AGND may introduce noise into the system because it is not a high impedance connection. AGND is a direct connection to internal reference voltage sources.
Even if the channel settings screen is used to disconnect Bias from certain channels, it should still serve as Ground pin connection, with appropriate high impedance. 'Bias' when it is connected to an EEG channel (the default condition) has the additional function of inserting an inverse (180 degree out of phase) mains small current to counter the ubiquitous mains EMF interference present in most environments. The default connection of Bias is what is used in EEG situations. The control panel can be used to disconnect Bias from channels used for EMG / ECG, since these are such high amplitude sources. In that case Bias still serves as a Ground connection; the control panel just turns off the current injection feature.
William
Cheers William. Got it.
Give this a try and see if your noise level comes down somewhat. Also check all AC wiring in the area.
Also see this related issue. The link inside the issue page, goes to the old Cyton tutorial, showing how the Channel Settings screen can adjust various ADS1299 internal configurations. Including disconnecting from SRBx buses, or Bias.
https://github.com/OpenBCI/Documentation/issues/16
Since your EEG cap already has connections for Ground (Bias), you can omit the EMG 3rd wire. Now that the tutorials are all split up for EEG, EMG, ECG, they don't show how a combined EEG, EMG, ECG setup is done. Which the (previous link) tutorial DID show.
You can use the GUI Channel Settings sub-panel to turn off Bias and SRBx for your EMG / ECG channels. They will still utilize Bias as a Ground lead.
Thank you
Let us know your resolution of the issue. The goal is to have simultaneous EEG, EMG, EOG, without distorting the EEG channels with the high amplitude EMG, EOG.
Made the changes you recommended. Removed the 3rd EMG lead. Toggled off SRBx and Bias for EMG. EEG signal looking good. EMG signal looking good. We've just gone through some calibration exercises. My colleague looking at the outputs. Next step will be filter per channel.
Mentioning Richard @retiutut. I think this underlines that our Docs regarding AGND connections are potentially misleading people. Will work if doing EMG or ECG ONLY. (Because signals are so strong.) But if combined with EEG, the AGND connection will introduce noise. Better to just use Bias and disconnect it from EMG ECG channels. It then still functions as a Ground, but turns off the inverse out of phase mains cancellation feature.
@wjcroft could I please check the config for the EOG part of this montage please. It seems to be the only element (channels 9 and 10) that aren't registering as expected. We are using 2 electrodes attached to N2P and N3P on the Daisy Board. No other reference other than the existing Cap REF and GND.
I've attached an extract of the raw data (time stamp 23:57:40 to 23:57:59 is an example of 5 "hard" blinks). It feels like a reference is required e.g. M1 and M2? But that is guessing. We've tried 3 times ensuring eye electrode placement is as accurate as we can make it. Will keep trying different options.
@wjcroft could I please check the config for the EOG part of this montage please. It seems to be the only element (channels 9 and 10) that aren't registering as expected. We are using 2 electrodes attached to N2P and N3P on the Daisy Board. No other reference other than the existing Cap REF and GND.
It feels like a reference is required e.g. M1 and M2? But that is guessing. We've tried 3 times ensuring eye electrode placement is as accurate as we can make it. Will keep trying different options.
I believe EOG is really the same as EMG, just for eye muscles. So a suggestion would be a differential as shown, NOT referential. Labeling on mainboard 'NxP' is a bit of an abbreviation. The actual pin names are IN1N to IN8N for the 'bottom' pins. And IN1P to IN8P for 'top' pins.
https://www.google.com/search?q=eog+electrodes&rlz=1C5CHFA_enUS577US577&source=lnms&tbm=isch&sa=X
https://en.wikipedia.org/wiki/Electrooculography
@wjcroft If you think it would help to change the docs, can you condense this into an Issue on the Docs GitHub repo?
We do still intend to create a Hybrid setup how-to (EEG/EMG/EKG) in the near future.
Here is the nutshell: it's a mistake to suggest using AGND as a 'Ground' connection. Bias instead acts as Ground.
If desired, the ADS1299 control panel can be used to disconnect the inverse current injection from Bias. But it still acts as Ground. If you look at Conor's original doc with the Cyton combined EEG / ECG / EMG, you can see this.
I'll try to remember to create a Github issue.