[Eeglablist] multi-subject ICA

[Makoto Miyakoshi]

Dear Russell,

If you concatenate all the subject's EEG to run a one-step ICA, I believe
that approach is called group ICA. EEGLAB does not support this and instead
direct user to (much more complicated) single-subject ICA and subsequent
ICA-clustering at the group level. Why do we want to bother ourselves in
this way, given that the group ICA is much more simple and straightforward
(at least computation-wise)? The reason is because our recommended method
appreciates individual differences. Your Fz is not his Fz because the
underlying brains have different gyrification patterns etc...

I've used conn and GIFT in fMRI years ago. Quite a straightforward
group-ICA approach, using PCA several times to reduce dimensions. Results
were robust and good.

Makoto

On Fri, Jan 15, 2016 at 2:59 PM, Russell Butler <
Russell.Buttler at usherbrooke.ca> wrote:

> hey all, i'd like to get group-level components from the ICA in eeglab...i
> see the same components showing up over and over across subjects, so i was
> wondering if its possible to perform a multi subject ICA using
> pop_runica...similar to whats done with FSL's melodic.
>
> i was thinking of just merging all the datasets from all my subjects,
> running ICA on the merged set, and then applying the weights on the single
> subjects. does it make sense to do this, or is there another/better way?
>
> thanks,
>
> Russell
>
>
>
>
> _______________________________________________
> 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
>



-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20160119/e2b63551/attachment.html>