Dear Bernard,
On Mon, Jul 28, 2008 at 3:29 PM, Bernhard Spitzer
<[log in to unmask]> wrote:
> Dear SPM experts,
>
> a question regarding the group inversion functionality of
> the SPM EEG 3D source reconstruction module.
>
> For single subject inversion of epoched data (e_file),
> after definition of a time-frequency window, one can
> choose between evoked and induced activity.
> However, using the group inversion routine, one can define
> a time-frequency window, but there's no further selection
> between the type of activity (evoked/induced).
I added the question about evoked/induced to the group inversion
script (see attached). I do not have time to test it though and I
leave it to you. In general localization of induced power is something
we do not have too much experience with and none at all with group
inversion. However, just yesterday someone wrote me off the list
asking practically the same question as yours so it seems there is
interest in this option and we are looking forward to hearing about
your results.
>What type of
> TF-activity is actually imaged when using group
> inversion? Is there a way to choose between
> evoked/induced/total power when using the group inversion
> routine?
The inversion routine does not make this kind of distinction. It just
looks at the raw data. This is true for both group and non-group case
as the same function is called. The distinction only comes about when
you define a 'contrast' to write out a single image that summarizes
the 4-dimensional result of the source reconstruction. This is done
with the 'Window' button of the 3D-source reconstruction interface.
You are asked about time window and frequency window of interest. I
hope time window is clear. If you input 0 for frequency window, SPM
averages the ERP amplitude over the time window specified weighted by
a Gaussian. If you specify some frequency limits, a series of wavelets
is used to summarize the power in the time-frequency window specified.
Note that in your terminology that would be the 'total' power. The
'evoked'/'induced' question refers to what is done if multiple trials
of the same type are present. In the 'evoked' case the trials are
first averaged and then the contrast is applied. In the 'induced' case
the contrast is applied to every trial and the result is averaged.
This is how things should work in theory but I advise you to test
things thoroughly because you are one of the first people actually
using these options.
Good luck,
Vladimir
|