[Eeglablist] rerun ICA to improve ocular artifacts removal

[Tarik S Bel-Bahar]

Rerunning a second Ica should work, but I don't think you should be
removing components. Try ica_based artifactual epoch rejection instead,
using really or fake epochs.

However, with sleep data things might be different.
On Feb 18, 2012 9:58 AM, "Daniela Dentico" <dentico at wisc.edu> wrote:

>
>
>
> Hi! I was wondering if it does make sense on a theoretical point of view
> to rerun ICA on
> data cleaned from artifacts with a previous ICA if the result is not
> satisfactory.
>
>
>
> I'm analyzing continuous waking data
> (256 channels EEG) from a subject with a huge amount of eyes movements.
>
>
>
> Before ICA the topography was
> completely dominated by eyes movements. After removing independent
> components
> that isolated ocular artifacts, there was a big improvement, with a 10
> times reduction
> of the power of ocular artifacts in the theta band for instance. Eyes
> movements were still there
> though.
>
>
>
> I tried to run ICA a second time
> on the clean data. I'm very doubtful about the quality of this second
> decomposition since many source activation time courses (scroll component
> activities) look very weird,, with a lot of high frequency activity, as
> you can see from the enclosed file ('ICactivation') that show the first 10%
> of the sources activities. I also enclosed a screen shot of the first 8
> components of this second run of ICA and a higher magnification image of
> the continuous data for the first component.
>
> Any cue/suggestion/explanation?
>
>
> Thank you!
>
> Daniela
>
> --
> Daniela Dentico, MD PhD
> Center for Sleep and Consciousness
> University of Wisconsin - Madison
> 6001 Research Park Boulevard
> Madison, Wisconsin 53719
>
>
>
>
>
>
>
>
> _______________________________________________
> 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/20120219/360e67e4/attachment.html>