Dear Leanne,
> I am having a few problems setting up a model for a study with variable
> epoch lengths.
>
> I have a resting breathing condition (12 x 30 second periods) and a
> voluntary hyperventilation condition (12 x 30 second periods)
>
> In the first analysis I set up a model in which I specified two
> conditions (rest and voluntary), however I have since realised that I
> maybe should have treated the study as having only one condition - vol,
> and treated the resting periods as baseline.
It should not make any difference with box-car regressors and one basis
fuction per condition.
> In the second model I have specified 1 condition (treating the rest as
> baseline), epoch - fixed response box car, convolved with hrf and
> temporal derivative.
> The problem arises when I specify the epoch lengths (some are 5 scans
> in length instead of 6 because of variability in breathing). After I
> enter vector of onsets, SPM asks 'variable duration'. At this point I
> enter yes, then I am prompted by SPM to enter 'duration (scans)'. If I
> enter the variable epoch lengths at this point, I am also asked at a
> later point in the model set-up to enter 'Epoch length (scans) for
> trials'. However, SPM will only accept scalars and not a string of
> vectors (which specify the variable epoch lengths). I don't understand
> why it asks me twice to enter epoch length and will only accept scalars
> on the second prompt (when I set up the first analysis it accepted a
> string of vectors) Also, should I be specifying each condition in the
> design matrix or only the vol condition?
Variable length epochs are dealt with using the event-related options.
Select event-related, not epoch-related when chosing your basis set.
The simpler alternative would be to have two 'vol' conditions (one of 5
scans and one of 6 scans) and simply take to average using contrasts
later on.
With very best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|