Hi,
If you are really concerned then, one way to do this is to not use fsl_regfilt and then add the relevant columns from melodic_mix as confound regressors in the Feat design matrix.
Note, however, that both wrt DOFs and wrt denoising this wont be the same. in fsl_regfilt the outer product of the time series and the spatial maps is being subtracted, so per removed time series you are not loosing a DOF in fsl_regfilt but you would if you add these as confound regressors. In this sense fsl_regfilt is more like an intelligent filter (which would normally not be adjusted for in the DOFs).
hth
Christian
On 11 Oct 2011, at 21:00, Marina Shpaner wrote:
> Thank you for your response. So, to clarify, when we run the first level Feat on the "output" of the fsl_regfilt, how do we adjust the degrees of freedom or is it already part of the "output" file?
>
> Thank you,
> -Marina
|