Dear Jeroen,

On 17 Oct 2011, at 17:48, Jeroen Van Schependom <[log in to unmask]> wrote:

Many thanks for the quick response. But if I understand it well, it is not possible to integrate a 3D-region ?

No, that's the whole idea of mass univariate. But you can focus on a particular region by doing a small volume correction for that region.

And plot the results in function of time ? Because I would like to perform an analysis on those results.

You average over the time dimension to create images so there is no time-resolved data that you can plot. You can use the source extraction functionality (most conveniently via the batch interface) to export 'source waveforms' summarised over some volume of interest as SPM datasets. This sounds more like what you want. However, these will be 1D and doing stats on them in SPM is tricky and there is no GUI support for that.

Best,

Vladimir


Kind regards,

Jeroen.

2011/10/14 Vladimir Litvak <[log in to unmask]>
Dear Jeroen,

You can find in the manual around pages 124 and 390 some instructions
about how to summarize the results of your source reconstruction as an
image. You can then subject those images to standard statistical
analysis in SPM. For an example of that you can look at 'PET verbal
fluency' example (page 281) and also some additional documentation
starting at page 79. For statistical analysis, however, you will
always need many images so you'll probably need a group of subjects.

Best,

Vladimir

>Fri, Oct 14, 2011 at 8:50 AM, Jeroen Van Schependom <[log in to unmask]> wrote:
> Dear all
>
> I am currently working on a dataset containing EEG-data. I am working with
> SPM8 and I have (for one subject, to start with) completed the (visual) 3D
> source reconstruction. However, the next step would consist of defining a
> Region/Volume of Interest in order to be able to integrate the voxel values
> in order to assess a certain measure of brain activity. I have been trying
> different methods but none of them worked out yet.
>
> The manual mentions that the obtained reconstructed activity in 3D voxel
> space can be further analyzed using mass-univariate analysis. But how does
> this work in practice ? In fact I should only be able to assess the values
> plotted by the 'render' interface of the 3D-source reconstruction, but I
> can't figure out how.
>
> Many thanks in advance,
>
> Jeroen Van Schependom.
>
>