Dear Michaël,
I guess the meaning of "effect of interest" is in the eye of the
beholder; it depends on what you want the representative time course of
a ROI to contain and what you want to do with it later on (DCM?). The
computation takes place in spm_regions.m:
> Responses are adjusted by removing variance that can be predicted by
> the null space of the F contrast specified (usually an F-contrast
> testing for all effects of interest).
If you use option 1) (ie leave NOGO responses in the representative time
course) and enter the extracted time series in a model that is not aware
of NOGO responses, then their contribution to the signal will remain
unexplained and they will be attributed to the error term.
Best regards,
Guillaume.
On 22/12/16 12:54, Mouthon Michaël wrote:
> Good morning, I would need a clarification of the exact meaning of the "effect of interest" use to adjust the extracted eigenvariates.
>
> Is it all conditions which are convoluted with the HRF or only the one of interest regarding to the task?
>
> For exemple: I have three modelized conditions : GO, NOGO, Wrong responses. The last one doesn't interest me (no hypothesis to test). The F-contrast of "effect of interest" used for adjustment should be
>
> 1) [1 0 0;0 1 0; 0 0 1]
> or
> 2) [1 0 0; 0 1 0; 0 0 0]
>
> ?
>
> Thanks
>
--
Guillaume Flandin, PhD
Wellcome Trust Centre for Neuroimaging
University College London
12 Queen Square
London WC1N 3BG
|