[Eeglablist] Question on dipole modeling

[Arnaud Delorme]

Dear Muluh,

>1. Is it necessary to do PCA or ICA and get components of interest for
>dipole analysis? (I intend getting    latencies of interest for my
>components).
>  
>
Yes, it is necessary although the tutorial describes a trick to perform 
dipole analysis at a single latency (the dipfit_erpeeg() function you 
mention below). Note that "getting latencies of interest for my 
components" has no meaning as components are static spacial filters 
(their scalp map does not vary with time).

>2. If yes or no in (1) above, can anyone help with the major difference
>or advantage of using ICA over PCA and    finally any advantages or
>disadvantages for using latencies of interest from my ERPs.
>  
>
The difference between ICA and PCA is described in most ICA papers. See 
for example the EEGLAB reference paper at

http://sccn.ucsd.edu/eeglab/download/eeglab_jnm03.pdf

>3. The magnetic resonance imaging (mri) data in EEGLAB defaults to mri
>data from Montreal Neurological       Institute mri data when
>dipfit_erpeeg is called. I have MRI data in analyze format. Can anyone
>help on how      to use this data in the dipfit_erpeeg function please.
>  
>
This will be possible in the new release (hopefully very soon).

Best,

Arno