[FieldTrip] ROI selection in a GC analysis

["Jörn M. Horschig"]

Dear Marco,

I once played around with that as well and tried different approaches. I 
think the cleanest approach is to define some region of interest based 
on anatomy and current literature and take the neighbouring peak voxel 
for each subject. In a similar vein as SPM is doing it, you can first 
spatially blur your source reconstructed activity if you like. Of 
course, instead of taking the initial ROI from literature, you can also 
do it as Arjen suggested take the ROI from some cluster analysis. 
Anyway, by making it subject-specific by taking the peak voxel per 
subject and blurring it, you will end up with a somewhat consistent 
region per subject and will be less sensitive to inter-subject 
variability. Unfortunately, you need to write most of this yourself.

There are tal2mni and mni2tal functions on the internet for transforming 
between these two coordinate systems. Note that this is only a rough 
transformation and might not 100% correct.

For blurring (or smoothing if you want to make it sound fancy) you can 
use spm_smooth. Note that this function is does work like the usual 
Matlab function in that it does not produce an output argument but 
rather changes the input argument inherently (in more technical terms: 
call-by-reference rather than call-by-value).

Good luck! :)
Best,
Jörn


On 12/11/2012 9:41 PM, Marco Rotonda wrote:
> Dear fieldtrip experts,
>
> I'm sorry to ask again to the list the same question but I had no 
> answer and I still have not solved my questions.
>
> I would like to ask you if there is an elegant solution in the 
> selection/definition of the ROI for a Granger Causality analysis.
>
> Just to explain better my doubts.
> Let suppose i've the sources reconstructed in the temporal domain 
> and/or in the frequencies domain.
> If now I wish to perform a GC analysis I should know the ROI or 
> sources on which to perform it.
> As far as I've seen most of them are selected with an a priori knowledge.
>
> I would like to know if there is a way to define the ROI from the 
> data, not from my (eventually with bias) knowledge.
>
> As far as I've searched the only method developed is the Full-brain 
> auto-regressive modeling (FARM).
> http://www.ncbi.nlm.nih.gov/pubmed/21439388
>
> Since there are already ft_mvaranalysis and ft_connectivityanalysis, do
> you think that is possible to integrate the FARM approach into those
> functions?
>
> I'm sorry but I know my limitations and I don't know how to do it.
> I hope this could be useful not for me only.
>
> Regards,
>
> Marco
>
>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip


-- 
Jörn M. Horschig
PhD Student
Donders Institute for Brain, Cognition and Behaviour
Centre for Cognitive Neuroimaging
Radboud University Nijmegen
Neuronal Oscillations Group
FieldTrip Development Team

P.O. Box 9101
NL-6500 HB Nijmegen
The Netherlands

Contact:
E-Mail: jm.horschig at donders.ru.nl
Tel:    +31-(0)24-36-68493
Web: http://www.ru.nl/donders

Visiting address:
Trigon, room 2.30
Kapittelweg 29
NL-6525 EN Nijmegen
The Netherlands

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20121212/b24501cb/attachment-0002.html>