what is wrong?

[jan-mathijs schoffelen]

Dear Maité,

What do you mean by spatially not located sources?
In general it may be helpful to give more to-the-point information  
besides the input cfg-structures. The output of your beamformer  
analysis depends both on the sanity of the data (which I cannot  
check), and on the sanity of the input configuration.
Two remarks w.r.t the cfg:

-when cfg.lambda is a scalar, the csd-matrix will be regularized with  
an identity matrix, scaled with cfg.lambda. From your value of 0.05  
however I assume that you want to do a relative scaling. True? In that  
case you should specify cfg.lambda = '5%'; The consequence of the 0.05  
could be that if the scale of your data is Volts-squared or microVolts- 
squared (boiling down to the csd having a magnitude of 10^-12 or 1,  
respectively) which at least in the first case leads to inappropriate  
regularisation.
-in the second step I assume you want to re-use the filters computed  
in step 1. Therefore you need to pass source.avg.filter (computed in  
the first step) to the cfg. Does this happen in source2grid? I don't  
know, it could be, but I'd check that if I were you.

Some post-processing probably needs to be done in order to see  
spatially localized sources (even when the whole analysis went fine  
(=data OK and cfgs OK) until this point), so just looking at the  
output of sourceanalysis typically is not optimal.

For the computation of the CSD-matrix I'd like to refer to the  
tutorial http://fieldtrip.fcdonders.nl/tutorial/beamformer
on the fieldtrip wiki.

Best,

Jan-Mathijs


On Jul 6, 2010, at 12:56 PM, Maite Crespo wrote:

> Hello!
>
> I read some previous mails in the discussion list, but I'm still not  
> sure about
> the procedure I'm following. I'm obtaining too flat images (not  
> spatially located
> sources). I would like to localize the source activity within a  
> frequency band
> and a time window. I computed the CSD matrix using ft_freqanalysis.  
> Can I
> average the resulting matrix in frequency and time before using it  
> with
> ft_sourceanalysis?
>
> I used the following steps to compute the filters in the averaged  
> CSD (across
> trials) and then apply it to compute the power in single trial data.  
> Is that
> right?
>
> % 1st run
> cfg=[];
> cfg.frequency=5;
> cfg.latency=.7;
> cfg.method='dics';
> cfg.projectnoise='yes';
> cfg.keepfilter='yes';
> cfg.grid=grid;
> cfg.vol=vol;
> cfg.elecfile=elecfile;
> cfg.lambda=.05;
> source=ft_sourceanalysis(cfg,data);
>
> % 2nd run
> cfg=[];
> cfg.frequency=5;
> cfg.latency=.7;
> cfg.method='dics';
> cfg.projectnoise='yes';
> cfg.rawtrial='yes';
> source.leadfield=grid.leadfield;
> cfg.grid=source2grid(source);
> cfg.vol=vol;
> cfg.elecfile=elecfile;
> cfg.lambda=.05;
> source=ft_sourceanalysis(cfg,data);
>
> I also tried changing the lambda parameter, but it seems there are  
> not many
> changes...
>
> Many thanks in advance for any help.
>
> Best regards,
> Maité
>
> ----------------------------------
> The aim of this list is to facilitate the discussion between users  
> of the FieldTrip  toolbox, to share experiences and to discuss new  
> ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html 
>  and http://www.ru.nl/neuroimaging/fieldtrip.
>

Dr. J.M. (Jan-Mathijs) Schoffelen
Donders Institute for Brain, Cognition and Behaviour,
Centre for Cognitive Neuroimaging,
Radboud University Nijmegen, The Netherlands
J.Schoffelen at donders.ru.nl
Telephone: 0031-24-3668063


----------------------------------
The aim of this list is to facilitate the discussion between users of the FieldTrip  toolbox, to share experiences and to discuss new ideas for MEG and EEG analysis. See also http://listserv.surfnet.nl/archives/fieldtrip.html and http://www.ru.nl/neuroimaging/fieldtrip.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20100706/34d145bf/attachment-0002.html>