Dear Thomas,
On 10 Jan 2010, at 14:28, Thomas Fischer wrote:
> I want to compute sources of event related beta with the meg
> toolbox in spm8. However, in the DICS beamformer tool is no option
> to compute event-related or induced activity.
>
Event-related activity is usually just power normalized to baseline.
It is not possible to reconstruct such a thing directly. However, you
can reconstruct activation and baseline time-windows and subtract
them. This functionality is built-in in the DICS function. You can
define several time windows and a contrast and the images that are
written out are with the contrast already applied.
>
> at second:
> I have tested the fieldtrip DICS tool in SPM8, with two conditions
> and the contrast between the two conditions. Baseline vs.
> eventrelated power. The nifti files are written , each file for each
> epoch and each subject.
> Now my question: should I test the significance of event related
> power in the second level with a one sample t-test and with using
> all of the nifti files or should I average the files before second
> level testing subjectwise? Or I am completely wrong?
>
I wrote the function for my own purposes and I needed an image per
trial. In the latest version that I posted on the list there is an
option to produce an average image (not in the GUI). You can also
average the nifti images or your can do something more complicated
like a regression within subject and a single-sample t-test at the
second level.
Best,
Vladimir
|