[Eeglablist] All red or blue ICA

Tarik S Bel-Bahar tarikbelbahar at gmail.com
Wed Apr 5 14:38:53 PDT 2017

Hello Julia, some notes below.

WIth 20 channels your ICs will likely not be that great, though you should
get a few neural ICs, and also some artifactual ones that you could remove.

Some investigators are quite cautious about IC removal, and only remove
clear artifacts.
Some investigators are focused on only using clearly neural ICs, and/or
rebuilding their data with just clearly neural ICs.
It depends on their understanding and experience and biases.
try to err on the side of caution unless you have clear reasons that you
can report for dropping some ICs.

Also, think about just focusing analysis on just two or three clear neural
ICs rather than "rebuilding" the data after IC dropping.

What you need to determine is whether yours is artifactual or not. If it's
not dipolar, does not have a clear frequency peak, and does not take a lot
of variance,
then you can probably safely remove it.

However as a caveat, with sparse EEG such as yours, you want to think about
trying to retain information that you are not sure is artitifactual or not.

You may also run analyses to compare two different strategies (keeping the
ones you're not sure about) and (removing them), which you may also report

If you have not had a chance to yet:

Review the online EEGLAB ICA detection site, which has more information
about what constitutes weird non-neural ICs, which might be what you are

*http://reaching.ucsd.edu:8000 <http://reaching.ucsd.edu:8000> *

Please review if you have not had a chance the PLos article from the SCCN
team about different IC functions and what makes IC good:


Please also review the Onton and Makeig chapter in Luck's ERP Handbook
about ICA for EEG.

Please also google past eeglab list posts on similar topics.

On Wed, Apr 5, 2017 at 11:24 AM, Julia Basso <jb4590 at nyu.edu> wrote:

> Dear all,
> I have a 20 channel EEG system and after cleaning the data by first
> conducting a band pass filter than cleaning epoched data by visual
> inspection, I am running ICA to detect and remove eye blinks and eye
> movements.
> My question is in regards to ICs that are distinctly all blue or all red.  My
> instinct is to remove these ICs (and when I do, the data looks cleaner)-
> but I wanted to get your opinion.  What does this type of IC indicate?  Is
> it merely noise in the system?
> Best,
> Julia
> --
> *Julia C. Basso, PhD, CYT*
> *Post-doctoral Research Associate*
> *New York University*
> *Center for Neural Science*
> *Suzuki Laboratory*
> *4 Washington Place, Room 809*
> *New York, NY  10003*
> *(212) 998-3969 <(212)%20998-3969>*
> www.juliabasso.com
> *www.suzukilab.com <http://www.suzukilab.com>*
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.uc
> sd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20170405/402080dd/attachment.html>

More information about the eeglablist mailing list