Dear Matthew,
> We are working with a very simple on-off boxcar design in FMRI, and
> found ourselves wanting to calculate the percentage signal change in
> the on condition relative to the level of signal in the off condition,
> for a chosen voxel. Is there an easy (or not-so-easy) way of doing this
> in SPM99?
The fitted effect (Y during plotting) is the activation expressed in
adimensional units of the whole brain mean (assuming proportional
scaling was used). If the default of 100 was used for the grand mean,
over scans, then this is the percent activation in terms of whole brain
signal. The percent activation in terms of signal at that voxel would
require the parameter estimate for the constant term at that voxel
(easily accessed wih the appopriate contrast).
The fitted effect can be computed by multiplying the appropriate
parameter estimate by the height of the box car regressor in the design
matrix.
Very best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|