Hi William,
Assuming you have one set of diffusion data in a 4D file in e.g.
compressed nifti format, and assuming this file is called
data_raw.nii.gz, you may go through the following steps:
1. Eddycurrent (and movement) correction (if not done yet):
> eddycorrect data_raw data 0
N.B. This assumes the reference volume for eddy-current correction is
the first one, which we would recommend as long as it is a non-
diffusion-weighted image (or weakly diffusion-weighted), i.e. b=0.
2. Extract brain mask:
> fslroi data nodif 0 1
> bet nodif nodif_brain -m
N.B. The first command also assumes that the first volume of your data
is a b0
N.B.2: you need to check the result of bet (load nodif and
nodif_brain_mask in fslview) and potentially play with the -f option
to get a reasonable brain extraction
3. Produce bvecs and bvals files
This is the tricky part, but generally, these files can be produced
using information contained in the DICOM files.
You may want to read the entry for bvals/bvecs in the following page: http://fsl.fmrib.ox.ac.uk/fsl/fdt/fdt_bedpostx.html
In order to avoid i.o. problems, do not use a common editor for
producing bvals and bvecs files. The best way is to use matlab or
emacs. For example, in matlab if you have the gradient directions in
the variable Z then type "save -ascii Z bvecs" at the matlab promps.
The order of entries in these files must match the order of volumes in
the input data series.
4. Run dtifit
If you have done the previous steps correctly, this is easy, just use
the following command:
> dtifit -k data -o dti -r bvecs -b bvals -m nodif_brain_mask
N.B. check the results of this before proceeding. Load dti_FA and
dti_V1 onto fslview, change the display of the second image to "line"
mode, and check whether the corpus callosum on the coronal AND the
axial slices connect the two hemispheres. If they don't, then
something is wrong with your bvecs. The most common thing is a Left-
Right swap. In which case you will need to invert the sign of the
first coordinate in your bvecs file, which you can easily do with
these three matlab commands:
> load bvecs
> bvecs(1,:)=-bvecs(1,:);
> save bvecs bvecs -ascii
5. Run bedpostx
Assuming the folder you were working on so far (which contains data
and dtifit results) is called <subj1>, then run:
> bedpostx_datacheck <subj1>
This is a sanity check for data compatibility before running bedpostx.
Then:
> bedpostx <subj1>
Finally, please read the FDT documentation ( http://fsl.fmrib.ox.ac.uk/fsl/fdt
) for more details.
Cheers,
Saad.
On 10 Feb 2009, at 14:18, William Sohn wrote:
> Dear FSL community,
>
> I am new to FSL and am currently studying Resting State connectivity
> and am
> trying to obtain connectivity maps by seed voxel correlations.
>
> I've read in literature that I can analyze connectivity of the brain
> to a given
> ROI using FSL however I'm not sure which tools to use in order to do
> this.
>
> FDT diffusion seemed like the right tool to use but I was unsure if
> this was
> more related to DTI structural connectivity.
>
> Also I tried to get FDT to work however I seem to have problems even
> when
> using the tutorial data. When using my own data I'm not sure where
> to start.
> In the tutorial for DTIF files such as the bvecs, bvals and brain
> masks are
> provided. However when running my own data I am unsure where to
> obtain
> these files.
>
> Basically what I want to know is if FDT is the right tool to use and
> how do I
> generate data files similar to "subj1 / subj1.bedpost" in the
> tutorial data.
>
> If I am doing this wrong I was wondering if someone who has
> experience in
> this area could point me in the right direction.
>
> Also if there is any other documentation besides the tutorial (fsl
> course) I was
> wondering if someone could tell me where to find it.
>
> Thank you,
>
> -William Sohn
>
Saad Jbabdi
Oxford University FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222545 (fax 717)
www.fmrib.ox.ac.uk/~saad
|