Dear SPM
I have been running periodic block design fMRI expts with visual stimulation (on-off) and have tried adding realignment parameters to the design matrix. The realignment parameters seem to take quite a bit of activation away from visual cortex and I understand from previous correspondence on this site that it is not possible to distinguish the task correlated motion from 'real activation'.
I have noticed that after estimation of the model the design orthogonality changes (i.e. some of the rp columns become more collinear e.g.0.28 goes up to 0.68). I have been told that this results from the temporal filtering that is applied to the model.
My question is does filtering need to be applied to the realignment parameters since they are really just nuisance variables? Is it valid to just apply it to the paradigm? If so then maybe SPM will be slightly more leniant in removing activation.
Ahmed Toosy
|