Dear Gaetan,
First it is very good to hear that you got sensible and significant
results. It's indeed my feeling that the last version of the routines
is really on the right track. Regarding interpretation, I understand
that in both cases you used averaged ERP/ERF rather than epoched file
as input. In this case the difference between using a Gaussian window
and wavelets over your whole bandwidth is like the difference between
average and RMS. So if you have some activity that changes polarity in
your time window if you average over it, it can get attenuated or even
average to zero. So looking at total power retains more and if you can
do all your analysis just on that, perhaps that's what you should do.
Averaging is just more conceptually similar and standard thing to do.
Regarding limiting your analysis to a band, that's a question of
whether you can justify it. Also it's perhaps something that makes
more sense when looking at induced activity (i.e. inverting an epoched
file).
Best,
Vladimir
On Wed, Feb 2, 2011 at 3:18 PM, Gaetan Yvert <[log in to unmask]> wrote:
> Dear SPM experts,
>
> I have used the group source reconstruction toolbox provided by SPM.
> It worked very well, but know it's for me to interpret my results !
>
> I have used the both type of temporal contrast (ie gaussian window and morlet wavelet) and I founded slightly different results.
> I used a time windows for the inversion between 0 to 800 ms and a temporal contrast between 0 and 400 ms to highlight the different sources involved in all my process.
> I am working on a langage task which is known to imply broca's and wernicke's area and it is what I found !
>
> with the gaussian window I found Broca's area and with the series of wavelet I found broca's and Wernicke's areas. both type of contrast are strongly significant even with FWE correction.
>
> And know I have to interpret it!
>
> Can I say that using a gaussian windows I highlight more continuous activation whereas using a series of wavelet I will more highlight transient activation in upper frequency band?
>
> should I refine my study and doing it in the different well known frequency band?
>
> thank you for your attention,
>
> bests regards,
>
> Gaetan Yvert
>
|