Hi,
If the TR (pixdim4) is being reset then that sounds like a bug in film_gls.
You can fix it by applying fslcpgeom to restore the information about voxel sizes and TR (just copy from the original 4D input).
As for the design matrix, the easiest workaround (as I think there can be some annoying difficulties in working with zero EVs) is to use a single EV that is a column of ones. This will effectively remove the mean from the timeseries and then you can just add it back in later by using fslmaths to add the cope to the res4d file.
All the best,
Mark
> On 12 Jul 2017, at 02:14, SUBSCRIBE FSL US.Choi <[log in to unmask]> wrote:
>
> Whom it may concern,
>
> I run GLM to get the res4d.nii.gz image using flim_gls command as below.
>
> film_gsl --in=EPI --pd=design.mat --rn=stats --vef=pnmev001.nii.gz
>
> Here are two qeustions,
> 1. After running flim_gls command, I checked a time series of res4d.nii.gz and found that a duration between time points was changed.
> The TR of my EPI data was 1.5 (sec) but that of res4d.nii.gz is 1 (sec).
> How can I fix this?
>
> 2. I want to get the res4d.nii.gz image with only physiological regressor confounds.
> How can I make the design.mat ?
> Should I input only 0 or only 1 in one single column?
>
> Best
|