Hi Kristen,
It seems much easier to convert the data to a percent of the
mean before regression, rather than trying convert to percent
of baseline afterwards.
The basic problem is that the motion regressors devour the
baseline. You would really need to add the mean of each
motion regressor times its beta weight back to the constant
baseline term (per run).
Maybe we should consider removing the mean from each
run of each motion regressor, either in 3dDeconvolve or
beforehand. But that seems like work... :)
- rick