[Eeglablist] Dipole localization and ICA

[Iman M.Rezazadeh]

Dear Makoto,
Thanks for your reply!
 I think dipole approach might be  'good' for more or less localized sources
(maps) not distributed brain area. 
On the other hand, how could one say that the dipole which is the outcome of
DIPFIT has physiological meaning considering the fact that ICA maps are only
mathematical plots of a 'mapping matrix' . In other word , how can we find
physiological meaning of each component by saying that each ICA reflects
individual and SEPARATE brain process - which is in my point of view far
from reality of physiology of the brain?
Best,
Iman

Iman M.Rezazadeh, PhD
Postdoctoral Research Fellow
Center for Mind and Brain
University of California, Davis
irezazadeh at ucdavis.edu
Cell:310-490-1808
Skype: Imanmr




-----Original Message-----
From: Makoto Miyakoshi [mailto:mmiyakoshi at ucsd.edu] 
Sent: Wednesday, November 14, 2012 4:08 PM
To: Iman M.Rezazadeh
Cc: eeglablist at sccn.ucsd.edu
Subject: Re: [Eeglablist] Dipole localization and ICA

Dear Iman,

> 1-      As far as I know, EEGLAB uses ICA components for source
localization
> and they claim that each ICA component corresponds to a (or a pair of) 
> dipole?

Yes.

> So, if you choose N ICA components, is  each of them fit separately ?

Yes.

> in other words, for source localization, does the EEGLAB algorithm 
> consider all the selected components at a same time and do 
> localization for all of them

Yes.

> - if the presence of one component affects of the location of the 
> dipole which corresponds to the other component- ?

This never happens. ICs are constrained to be temporally maximally
independent of each other, but no constraint for spatial location.

> if yes, so, how can
> we say one dipole is associated with one ICA component?

ICA computes scalp maps. Dipole location is computed from the map using
dipfit (or equivalent).

> 2-      Is there any study on the fact that how many dipoles should be
> chosen from N-Channel EEG/MEG recording?

I'm not sure choose for what, but we usually use the residual variance of
15% after dipole fitting to evaluate the goodness of ICs. Please see Delorme
et al. (2012) 'Independent components are dipolar' PloS One.

Makoto



2012/11/8 Iman M.Rezazadeh <irezazadeh at ucdavis.edu>:
> Hi everyone,
>
> I have  two questions re dipole localization:
>
>
>
> 1-      As far as I know, EEGLAB uses ICA components for source
localization
> and they claim that each ICA component corresponds to a (or a pair of) 
> dipole? So, if you choose N ICA components, is  each of them fit 
> separately ? in other words, for source localization, does the EEGLAB 
> algorithm consider all the selected components at a same time and do 
> localization for all of them - if the presence of one component 
> affects of the location of the dipole which corresponds to the other 
> component- ? if yes, so, how can we say one dipole is associated with one
ICA component?
>
> 2-      Is there any study on the fact that how many dipoles should be
> chosen from N-Channel EEG/MEG recording?
>
> Best,
>
> Iman
>
>
>
> Iman M.Rezazadeh, PhD
>
> Postdoctoral Research Fellow
>
> Center for Mind and Brain
>
> University of California, Davis
>
> irezazadeh at ucdavis.edu
>
> Cell:310-490-1808
>
> Skype: Imanmr
>
>
>
>
> _______________________________________________
> 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
JSPS Postdoctral Fellow for Research Abroad Swartz Center for Computational
Neuroscience Institute for Neural Computation, University of California San
Diego