[Eeglablist] help for ICA on thermal signals

[giulia.rizza at tiscali.it]

Dear EEGLab users 

I know this is not a usual topic but maybe you
can give me some advice, I would like to ask you some help for the
correction of ocular artifacts on facial thermal imaging signals. The
temperature of the periorbital areas is very informative about the
sympathetic system activity, but we realized that blinks and saccades
also influence the temperature in these regions 

Our signal is sampled
at 10 Hz and the time course is extracted from 10 regions of interest,
one of which is placed exactly on the right eye. Thus we have 10
channels (ROIs). 

We preprocessed the signal by using a band pass
filter (Butt 0.01 - 4 Hz) and by demeaning it. 

Then we used ICA
(runica or fastica) to detect the ocular components. A component was
defined as "artefactual ocular" when its face distribution was mostly
around the eye and the periorbital channels. 

After subtracting that
component from the original signal, we saw that the signal was very well
corrected on the eye channel itself but it got worst in the channel of
interest (the periorbitals) by introducing some artefacts. Further we
noticed that the unmixing matrix was close to singular (det(A) =
10^-13). 

We thought that this could be due to the number of channels
so we also tried to export more (64) ROIs but nothing changed. 

Do you
have any idea about why this would happen? We thought that ICA could
work on any type of wave, but maybe there is something we are missing.


Should we use another method? Do you have any suggestion? 

Any
suggestion is pretty much appreciated 

Thanks in advance for any help.


Giulia   

Scopri istella, il nuovo motore per il web italiano.

Istella garantisce risultati di qualità e la possibilità di condividere,
in modo semplice e veloce, documenti, immagini, audio e video.
 Usa
istella, vai su http://www.istella.it?wtk=amc138614816829636
[1]



Links:
------
[1] http://www.istella.it?wtk=amc138614816829636
 

Scopri istella, il nuovo motore per il web italiano.
Istella garantisce risultati di qualità e la possibilità di condividere, in modo semplice e veloce, documenti, immagini, audio e video.
Usa istella, vai su http://www.istella.it?wtk=amc138614816829636

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20140606/b4e6828e/attachment.html>