Hello,
I've read some posts on how it might be possible to take an fstat or zfstat
image from a lower-level analysis and pass it up to a higher level analysis.
The reason I'm interested in this is because my fMRI activations use
temporal basis functions, and I want to allow the specific mix of basis
function elements to vary across subjects.
To test this out, I made up a simple simulated data set w/ multiple
subjects, ran my first level analysis, and then tried to pass the zfstat and
fstat images to a "simple OLS" analysis at the higher level. Here is the
error message I received at the "Higher Level Input Preparation" stage:
<<<<<<
/usr/pubsw/packages/fsl/current/bin/fslmaths cope1 -mas mask cope1
echo 1 > design.lcon
/usr/pubsw/packages/fsl/current/bin/fslmerge -t varcope1 0 0 0 0 0 0 0
** ERROR (nifti_image_read): failed to find header file for '0'
** ERROR: nifti_image_open(0): bad header info
Error: failed to open file 0
Cannot open volume 0 for reading!
/usr/pubsw/packages/fsl/current/bin/fslmaths varcope1 -mas mask varcope1
** ERROR (nifti_image_read): failed to find header file for 'varcope1'
** ERROR: nifti_image_open(varcope1): bad header info
Error: failed to open file varcope1
Cannot open volume varcope1 for reading!
>>>>>>
I tried renaming the zfstat and fstat images to look like a cope, and I also
tried using the mask image as the varcope (since it seems like its looking
for a varcope image, even though the OLS processing does not use that
information). I ran an OLS analysis on proper copes, and that ran without
errors.
Any idea what the problem could be?
Thanks for your help!
--Patrick
|