[Eeglablist] Interpolating bad channels while avoiding PCA before ICA

[Mikołaj Magnuski]

Dear all,
thanks for the responses!

Dear:
Simon-Shlomo Poil and Renzo Carlo Lanfranco,
That's true, no need to remove bad channels
if I can perform ICA only on the good ones. Thanks!


Dear Tarik S Bel-Bahar,
I am going to analyse ICs later, that is just the
initial, sensor-level analysis that will save me from
problems with inferior IC decompositions, clustering
and so on. I am planning to use Measure Projection
toolbox later though. (the channel-analysis is just
safer with respect to time-consumption or problems
I may face, so it will constitute my master thesis;
publication-wise I consider other approaches to
data analysis, like analysis on brain-like ICs)

Concerning bad channels: I stated that I have
maximum of four (clearly) bad channels per
subject, not four bad channels summing accross
all subjects or the same four bad channels in
most/all of subjects. Is it bizzare to have max 4 bad
channels? I thought it's within the realms of 'OK' :)

I am aware of the problem of throwing out the baby
with the bathwater in IC rejection, so I was planning
to remove exactly the components you mentioned
and additionally some really obvious artifacts that 'litter'
the electrode-signal (with caution, of course).

But what do you think about removing heart IC?
It contains rhythmic heart-related activity but
for other times (in beetween of the heart beats)
this spatial filter extracts (/creates) some non-heart
signal (this is at least the way I reason), so I am
not sure whether to remove such ICs from EEG
signal or not...


Dear Jeff,
not so long ago Arno mentioned on the list that:
"The function eeg_interp will also perform spherical
interpolation using Perrin's method."
which is either a statement of what is the current
or planned behavior of this function. Unfortunatelly
I am not sure which one is true (current vs planned)
but Andreas Widmann published eeglab plugin for
spherical spline interpolation among his other contri-
butions to eeglab here:
http://www.uni-leipzig.de/~biocog/content/widmann/eeglab-plugins/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20130522/704678cf/attachment.html>