Dear Anette,
On Tue, Nov 30, 2010 at 9:19 AM, Anette Giani
<[log in to unmask]> wrote:
> Dear Vladimir,
>
>
>
> I have a couple of questions concerning fieldtrip beamformer source
> extraction (spm_eeg_ft_beamformer_source.m). It will be great if you can
> comment on them.
>
> 1. Is it correct that the option ‘regional sources’ lets you choose between
> (1) selecting a dipole location and (2) an average across a ROI? For what
> cases would you suggest to use one or the other option?
>
I wrote this function for myself and tried out different options. I
didn't find a difference for using a ROI vs single source in my data
so I can't tell you when such a difference may exist.
> 2. Does it make a difference if I convert all conditions at once? Or can I
> extract sources for each condition separately?
>
If you are planning to compare these conditions the recommended good
practice is to extract them using the same filters, i.e. together.
> 3. Does it matter if I input an averaged file or do I need to input single
> trial data?
>
Beamforming does not work well on averaged data in general and the
amount of data in an ERF is not sufficient for reliable estimation of
a covariance matrix so you should use single trials.
> 4. Knowing my frequency of interest, would it be advantageous to narrowband
> filter my data before source extraction?
>
As long as you compute the filters based on the same data that you
apply them to, this is OK. If you remove some irrelevant noise by
restricting your frequency band this can make your beamfomer more
spatially selective.
Best,
Vladimir
>
>
> Thanks a lot for your help,
>
> LG,
>
> Anette
>
>
|