[Eeglablist] Reference Biosemi's electrode and ICA decomposition

[Seyed Yahya Shirazi]

Hi Nastassja,

I don't think that this is an obvious question. If you are using bioSemi
ActiveII, both bioSemi and EEGLAB strongly suggest we pick a reference,
otherwise, we would lose 40dB of data. EEGLAB goes a step further and
suggests that we can re-reference to a different channel at a later time.
I found both claims not accurate at least for the type of study I usually
do (which includes ICA, DIPFIT, and ERSP). That 40dB of data loss probably
refers to the common-mode rejection ratio (CMRR) and the fact that CMS and
DRL are located close to each other. While common-mode rejection can
effectively decrease common-mode artifacts such as line noise, eyeblinks or
motion, it can also decrease widespread cortical behavior, such as alpha or
beta signatures of different tasks. If we can re-reference to another
channel at a later time, then re-referencing to the same channel (or the
common average) for multiple times should not create any problem. However,
as you will see in the following presentation, re-referencing basically
removes the activity with respect to the referenced channel, and so
re-referencing multiple times to the common average only degrades the
signals. You can find the presentation for re-referencing from this link:
https://www.dropbox.com/s/v7dxulx0udljb18/re-referencing.pdf?dl=0

What should we do? The harder one is to use a reference that is
theoretically very far from the brain sources and so it is electrically
neutral, this method is called REST (Dong et.al. 2017). Maybe an easier one
is not to re-reference before ICA, and delay re-referencing until having
the ICA weighting matrix. PLEASE NOTE that these solutions are for bioSemi
or any other so-called "reference-less" systems because the signals are
already recorded with respect to a reference, but the reference itself is
not recorded.

Also, there is a good paper for comparing the influence of different
references: Federico Chella et al 2016,* Impact of the reference choice on
scalp EEG connectivity estimation,* J. Neural Eng. 13 036016.

I hope that I could help.

Best,
Seyed

__
Seyed Yahya Shirazi
PhD Candidate, BRaIN lab
University of Central Florida

On Wed, Jan 29, 2020 at 8:01 AM Nastassja Lopes Fischer <fasnlf at nus.edu.sg>
wrote:

> Hi all!
>
> I have a question regarding the referencing electrode. I noticed that the
> initial reference electrode I assign when I import the .bdf files to EEGLAB
> influences significantly the ICA decomposition later on, even when I
> perform the common average re-referencing (before ICA). Therefore, it
> impacts artifact rejection. Have anyone experienced this effect? Would
> anyone know why it happens? And what reference electrode would be
> recommended to use (given the fact mastoid electrodes are quite noisy in
> some subjects because of head movement issues)?
>
> Sorry if the question is too obvious, but I could not find a previous
> explanation for what I am facing.
>
> Best regards,
>
> Nastassja
>
> ________________________________
>
> Important: This email is confidential and may be privileged. If you are
> not the intended recipient, please delete it and notify us immediately; you
> should not copy or use it for any purpose, nor disclose its contents to any
> other person. Thank you.
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>