[Eeglablist] Moving ICA info to a different data set

[Tarik S Bel-Bahar]

Hi Becky,

well, hopefully you have some nice ICs after feeding ICA quality data that
is clean and long enough.
some quick responses to your queries below, good luck with the processing.


1. Wait to reject artifactual ICs till after you transfer them over to
EEG2. Order matters as I believe you'll not have matching channels between
the two files unless you seek out and remove those same channels from EEG2
before copying the ICA fields over from EEG1.


2. Yes, reject those same channels. You can interpolate the channels later
for EEG2.
What I usually do for an erp paradigm is have a file that is totally
cleaned of bad periods, then I drop all bad channels, then I epoch, then I
feed to ICA.
Then EEG1 with ICA can be applied the step right after dropping the bad
channels (if you save a file at that step).
Order matters, because  the channels of the file that is getting the ICA
should match exactly the channels of the file with the ICA.


Regarding running ICA only on good channels, some good rules of thumb are
don't drop more than 20% of your channels, don't replace bad channels with
data from surrounding bad channels, watch out for clusters of bad channels
and note that it's not good to interpolate clusters of adjacent bad
channels. Some of these points of course vary depending on whether you work
with a 16 channel montage or a 256 or better channel montage





On Mon, Dec 2, 2013 at 10:37 AM, Becky Prince <becky.prince at york.ac.uk>wrote:

> Dear EEGLAB users,
>
> I ran ICA on clean, epoched subsets of my individual data sets with the
> goal of using the ICA decomposition from each subject on their larger
> continuous data set.  So for each subject, I have a cleaned, epoched,
> smaller EEG set with ICA (EEG1) and a larger, continuous EEG set without
> ICA (EEG2). Now I'd like to copy each subject's ICA info from EEG1 into
> EEG2, and continue analysis on the EEG2 data sets for each subject.
>
> I have a few questions:
>
> 1. Should I reject the artifactual ICs before or after copying the ICA
> fields into the EEG2 data set?  Does the order of these steps matter?
>
> 2. If there are fewer channels in EEG1 (set with ICA), should I reject the
> same channels in EEG2 before copying the ICA fields into the EEG2 data
> structure?  Again, does the order of these steps matter?
>
> BTW I have been reading the previous posts about rejected/interpolated
> channels and ICA, so I'm somewhat familiar with the issues running ICA on
> only the 'good' channels.
>
> Thanks!
> Becky
> ____________________________________________
>
> Becky Gilbert (nee Prince)
> PhD Researcher
>
> http://www.york.ac.uk/psychology/staff/postgrads/becky.gilbert/
> Office B213
> Lab C120
> Department of Psychology
> University of York
> Heslington, York, YO10 5DD, UK
>
> _______________________________________________
> 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
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20131203/25e1820c/attachment.html>