Thank you a lot. The points 1-3 are clear to me now.
Regarding the design matrix with the VBM as a covariate I would like to make sure I am doing the right thing:
I have two groups that differ in their variances, thus:
Group EV1 EV2
1 1 0
1 1 0
2 0 1
2 0 1
I found in the previous posts (e.g. https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0901&L=FSL&D=0&1=FSL&9=A&I=-3&J=on&K=2&X=79D50F63AE93548029&Y=burzynska%40mpib-berlin.mpg.de&d=No+Match%3BMatch%3BMatches&z=4&P=262183) that I need to add the VBM information as 2 additional voxel-dependent Evs in order to make these Evs “separable”.
Group EV1 EV2 EV3 EV4
1 1 0 vol1 zero_vol
1 1 0 vol2 zero_vol
2 0 1 zero_vol vol30
2 0 1 zero_vol vol31
Therefore, I plan to run the feat_gm_prepare separately for each group (so that they will be demeaned within that group) and then concatenate each file with the “zero” volumes to make the t dimension fit the number of subjects. Is it the right way to deal with this covariate having 2 groups?
On 6/23/11 11:55 PM, "Steve Smith" <[log in to unmask]">[log in to unmask]> wrote:
On 22 Jun 2011, at 10:28, Agnieszka Burzynska wrote:
Hi Cornelius, dear all,
Thank you a lot for your help.
I have some follow-up questions:
1) Regarding the post
Do I need to make the same change in the feat_gm_prepare script?
I use the path with FSLv.4.1.8.
look at the prepared GM images - if they look like WM the script needs fixing…..
2) My fmri data contains 3 runs/subject. Does it mean that for the 1st level
analyses for feat_gm_prepare I can use only 1 run per subject?
Not sure what you're asking - the GM covariates are for use at higher-levels….?
3) I see that feat_gm_prepare uses fast to segment the anatomical images
(with the same parameters as vbm), then smooth, and register to a standard
I have my anatomical images completely pre-processed for vbm, so they are
modulated by the non-linear component of the transformation to the
study-specific template, concatenated, and smoothed.
Both the VBM template and our fMRI data (on the group level) are at 2x2x2mm3
resolution in standard space.
Can I use the 4D file, such as GM_mod_merg_s3, as an input instead of using
feat_gm_prepare? What is exactly the difference (and advantage) of using the
file generated with feat_gm_prepare? I think both Oakes et al. 2007 and
Filippini et al., 2009 used feat_gm_prepare and did not take the files
generated by VBM.
Yes there are subtle technical differences (like having smoothing that matches the smoothness of the FMRI data) so I would not re-use the VBM data here.
4) In the last step, feat_gm_prepare de-means the GM images. If I have 2
groups, should I make this step separately for each group, and then fuse the
two 4D images?
This depends on how you want to use the confounds within your higher-level modelling.
On 6/21/11 6:59 PM, "Cornelius Werner" <[log in to unmask]">[log in to unmask]> wrote:
yes this is available for some time now. For FEAT consult the online
help under http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html and search
for the term fsl_gm_prepare. Randomise has the beta option to include
VBM data as well (type randomise to get the usage) - not sure what the
beta stage means, however.
On Tue, Jun 21, 2011 at 5:58 PM, Aga Burzynska
<[log in to unmask]">[log in to unmask]> wrote:
I would like to follow up on the older post (see message below), as maybe
this option is now available.
Is there a way to integrate structural VBM data with results from FEAT in
I mean something similar to
or the BPM tool in SPM.
I could not find an option in randomise to include a separate 3D matrix as a
covariate or a way to correlate two multimodal volumes/subject, across the
I will be very grateful for your advice!
Previous post on a similar topic:
I'm not sure about FSL, but SPM has.a toolbox -- Biological Parametric
Mapping -- from Wake Forest University that dies exactly what you want
to do. I believevthe reference is Casanova 2007, but I could have the
On Tuesday, September 8, 2009, Andrej Schoeke <[log in to unmask]">[log in to unmask]> wrote:
I recently read a paper titled "Integrating VBM into the General Linear
Model with voxelwise anatomical covariates" by T. Oakes et al. I am
interested in applying this analysis technique to our data. I know about the
VBM tools in FSL, but is there a way to integrate the results into the
analysis as described in the paper?
Any hints are welcome.
 Oakes, T. R., Fox, A. S., Johnstone, T., Chung, M. K., Kalin, N., &
Davidson, R. J. (2007). Integrating VBM into the General Linear Model with
voxelwise anatomical covariates. NeuroImage, 34(2), 500-508. doi:
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask]">[log in to unmask] http://www.fmrib.ox.ac.uk/~steve