Dear experts,
I was wondering if you can give me an answer to a more general question regarding the effects of bridges on source localization in the EEG signal and their detection.
I have resting-state data from a 128-channel EEG system, and in some subjects the electrodes are may connected by gel bridges (we are not completely sure about that). The aim is to conduct connectivity analysis between ROIs on the source level. Of course, bridges would be really bad for our connectivity analysis, therefore, the two question arose:
Do you have an idea how the gel-bridges influence the source localization step? Is there a way to detect them afterwards (e.g. in the connectivity matrix or by significant events in the model)?
Best regards,
Julia
Hi Julia,
I have had the same problem so I have had to look at it… Even though I haven’t done any source localisation for the collected data what I know is that salt bridges are usually channel-specific and usually have lowered amplitudes than the background with general higher amplitudes. This is the consequence of the combining of electrodes resulting in one potential for the two electrodes, so an amplification based on the difference between electrodes producing a cancellation (rather flat and isoelectric).
Hope this helps if it doesn’t resolve it.
Cheers,
Rodrigo
Maybe @John_Mosher or @Sylvain have an idea of the effect of these bridges on the source estimation?
Bridges provoke changes in sensor signal dynamics which are unpredicted by the EEG forward model. I recommend bridged sensors be marked as bad and not included in subsequent sensor and source data analyses.
To detect bridges, one could compte NxN correlation between sensor time series and identify pairs of sensors with close to perfect correlation scores.
I hope this helps.