Thanks to your helpful comments. I really appreciate the wide range of
options to analyze CBF data in FSL. Maybe you can insert in the future
a paragraph with group analyses of CBF data within FSL. I must admit
that I analyze my BOLD data by default with SPM ;-) and therefore I'm
not so familiar with the data structure and analyze options in FSL.
Best wishes
Ralf
Thanks to Matthew there, but it does also merited me to say that you
could do analysis with randomise, the argument in favour of that is
less compelling for CBF as the distributional assumptions used in FEAT
seem reasonable for CBF, but permutation testing would be more
conservative.
Michael
On 16 Dec 2015, at 16:53, Matthew Webster <[log in to unmask]> wrote:
Hello Ralf,
randomise is an alternative to flameo, so is probably not the
best choice here to get corrected p-values. To simulate FEAT you need
to run smoothest on the residuals, and then pass those parameters to
cluster ( with a p-threshold )/
Kind Regards
Matthew
Dear Michael,
I tried flameo on the command line using your suggestions and it
worked well. Now I have a logdir directory with zstat1.nii and
zstat2.nii and several other files. Am I right that the zstats images
represent the statistical differences of A>B and B>A? Do I need
another step to correct for multiple comparisons (randomise)?
Best wishes
Ralf
As far as I was aware this should work - it is quite reasonable to
treat your calibrated CBF images as COPE images (as this is
effectively what they are, just run through a different ‘first-level’
analysis to BOLD). However, it is possible that FEAT is still checking
that registration has been done. I have generally run the higher-level
analysis directly on the command line using flameo, the syntax should
be fairly easy for your case, you first merge (using fslmerge) all
your CBF images and then run something like
flameo —cope={aslimages} —mask={mask} —dm={design.mat}
—tc={design.con} —cs={design.grp} —runmode=flame1
where I have assumed that you can get your design matrix (and
associated files) using Glm_gui.
Michael
On 10 Dec 2015, at 16:52, Ralf Veit <[log in to unmask]> wrote:
Dear experts,
I used the oxford_asl tool to create calibrated CBF images.
Registration and warping in MNI space was performed with FLIRT and
FNIRT. Then I created a design matrix (Higher-level analysis) for a
paired t-test with calibrated CBF images (already in MNI space).
Unfortunatley in the Data input section there is only the option for
"lower level FEAT subdirectories" or "Inputs are 3D cope images from
FEAT directories". Because I have neither one thing nor the other I
tried the option "Select cope images" and loaded the corresponding CBF
images (one input for one subject). When I try to run the anaylses I
get the message "Registration has not been run for all of the FEAT
directories that you have selected for group analysis". How can I run
the paired t-test without First-Level FEAT directories.
Best
Ralf
Ralf Veit PhD
> Institute of Medical Psychology and Behavioral Neurobiology
Otfried-Müller str. 47
72076 Tübingen
phone: +49-7071-2987703
fax: +49-7071-295706
---
Michael Chappell MEng DPhil
W: http://www.ibme.ox.ac.uk/qubic
T: +44 1865 617657
Associate Professor, Institute of Biomedical Engineering, University
of Oxford.
http://www.ibme.ox.ac.uk
Director of Training, EPSRC-MRC CDT in Biomedical Imaging
http://www.onbicdt.ox.ac.uk
Governing Body Fellow, Wolfson College, Oxford.
http://www.wolfson.ox.ac.uk
Research Fellow, FMRIB Centre
http://www.fmrib.ox.ac.uk
Ralf Veit PhD
Institute of Medical Psychology and Behavioral Neurobiology
Otfried-Müller str. 47
72076 Tübingen
phone: +49-7071-2987703
fax: +49-7071-295706
|