[Eeglablist] ICA and phase

[Makoto Miyakoshi]

Dear Otte,

This is cited from the paper pp1376.

%%%
Qualitative inspection of unwrap phase from real EEG data, Fig. 2, shows in
light blue that ICA removals introduce phase changes. Phase data was unwrap
for Fig. 2 but not for analysis of Fig. 3. As seen in Fig. 2, the
reconstructed data from independent components shows a big disruption in
phase when the identified ocular ICA components have been removed before
reconstruction. This was expected to occur due to the contribution of these
components to the averaged signal of the EEG at electrode level, especially
in the low frequencies range, where the blink artifacts have an important
contribution. This evidence shows that ICA removal has affected phase
differentially across frequencies.
%%%

So he says when he removed eye IC, it changed the phase of the data. Am I
right? Then I don't see any problem. In order not to change the phase info,
then the decomposed activity and the original activity should have the same
phase, but what does that mean??

Makoto

On Sat, Nov 1, 2014 at 10:43 AM, otte georges <georges.otte at pandora.be>
wrote:

> Dear all
>
>
>
> If we reject some artefacts in eeg recordings by zeroing their components
> and then reconstruct the “original” EEG by multiplying the rest of the
> components matrix with the inverted unmixing matrix, the time points of the
> “reconstructed” eeg are a weighted linear interpolation and hence there
> will be changes in phase angle between channels.
>
>
>
> This is a topic that is rarely mentioned in ICA literature although some
> simulation studies have been done (Montefusco at al)
> http://ieeexplore.ieee.org/xpl/abstractAuthors.jsp?arnumber=6696198
> showing that the phase distortion is largest at the origin of the artefact
> (due to weighting coefficients) and within its frequency range , yet the
> effect is not stationary.
>
>
>
> Is there a way, a strategy, a method , when using ICA for rejecting EEG
> artefacts , to avoid this phase distortion problem ?
>
>
>
> Sincerely
>
>
>
> Georges
>
>
>
>
> _______________________________________________
> 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/20141104/0b3b10f9/attachment.html>