Hello afni experts,
how can I with afni, having the time of a complex stimulus formed by: text, scene of a film, cross of fixation, eliminate the contribution of the stimulus “text” only?
For each stimulus I have onset and duration time.
In practice I should have a residual timecourse where the contribution of the text has been removed inside. Do you have an example?
Maybe you can do it with “3dDeconvole”? Anyone have an example?
This can be done with 3dDeconvolve in conjunction with 3dSynthesize. When 3dDeconvolve is run, include “-cbucket PREFIX”, which will create a dataset of only betas (for all regressors). With that, 3dSynthesize can construct a combination of any set of regressors (times their voxelwise betas). Removing just one might require a few -select statements, or you could simply -select the regressor to remove, and then subtract the result from the original time series (the 3dDeconvolve input).
Note that this is different than projecting out that one regressor (which would be done with 3dTproject). In the case of fitting and omitting, say (or subtracting as I suggest in this case), the resulting data will (presumably) not be orthogonal to the one regressor of (dis)interest.
Does that seem reasonable?