Hello everyone and of course a special HELLO to the one person that will
have the perfect solution :-)
We need some advice in terms of some old data, not designed for SPM
analysis, that we would like to "salvage". The nature of the stimulus
(visceral balloon inflation) did not allow for repeated rest and stimulation
conditions during one session, ie. it is not possible to inflate and deflate
the balloon repeatedly during one session. We know how a block design is
done and are aware of the shortcomings of the data. This said, can we do
something with this data in SPM?
We had 2 stimulation conditions (sensation (S) and pain(P))and rest (R)
conditions.
Each condition was acquired separately, ie the scanner was started and
stopped for each condition. We looked at:
1. the concatenated (cat) data (R P R for example) modelled as one session
(spatial pre-processing done on cat data). Result: the entire brain is
active probably due to a session-specific effect.
2. the non-cat data (spatial pre-processing done on non-cat data), modelled
as three sessions. This looks impossible as each session has no
activation/rest alternation (ie. SOA=activation length)
3. non-cat data and cat data modelled as one session and using user
specified regressors to model the session-specific effect. Result: we have
to convolve the box-car with hrf to get orthogonal columns in the model, and
then the fitted time courses are distorted by large peaks at the beginning
and end of the stimulation condition.
We read the SPM mailing list on concatenation, but everyone deals with
properly acquired sessions of course...
Do you think there is anything salvageable here?
Any suggestions would be sooo very welcomed!!
Thanks everybody
Uta
|