Saad,
Thanks for the reply. So the initialization of the streamline doesn't
count towards a hit, that makes sense. For these strange seeds, is the
most likely explanation that they quickly step outside of the brain,
hitting the mask and hence none survive?
--Scott
On 1 May 2014, at 4:32, Saad Jbabdi wrote:
> Hi
> If your seed is a surface, then a streamline starting from a vertex on
> the surface won’t count as having traversed the surface unless it
> actually does step through it. The first step of the streamline will
> not produce a count. That may explain what you are seeing (provided
> I understood your question correctly)
>
> Saad
>
> On 30 Apr 2014, at 15:29, Scott Burns
> <[log in to unmask]<mailto:[log in to unmask]>>
> wrote:
>
> Hi FSL folks,
>
> I'm probtrack'ing between all Freesurfer cortical labels from the
> Destrieux atlas. For most of the probtrack runs, things look fine.
> That is to say, the most intense region of fdt_paths.nii.gz intersects
> strongly with seed area, which is to be expected.
>
> For some seeds however, the matrix_seeds_to_all_targets.nii.gz is
> completely blank and waytotal is 0. Even if none of the streamlines
> hit another target, since each streamline starts within one of the
> voxels of the seed (by definition), that initial "hit" should get
> counted towards the seeds row the output matrix, correct?
>
> I'm confident in my registrations, having overlaid both the structural
> image in diffusion space & the freesurfer orig image in diffusion
> space and verifying nothing strange occurred.
>
> For one of the bad seeds, I converted the FS label to a volume using
> mri_label2vol and applied the fs2dif transform. This volume overlays
> nicely on the non-diffusion weighted image. I've confirmed it doesn't
> intersect with the ventricles which I'm using as the --avoid argument.
>
> I generated GIFTI seeds with a loop along the lines of...
>
> for hemi in lh, rh
> mris_convert $hemi.white $hemi.white.asc
> for label in <labels>
> area = $hemi.$label.label
> seed = $hemi.$label.gii
> if the seed doesn't exist:
> echo $area > tmpfile
> label2surf -s $hemi.white.asc -o $seed -l tmpfile
> rm tmpfile
>
> Any help/explanation you can provide would be most appreciated!
>
> --Scott
>
> --
> Saad Jbabdi
> University of Oxford, FMRIB Centre
>
> JR Hospital, Headington, OX3 9DU, UK
> (+44)1865-222466 (fax 717)
> www.ndcn.ox.ac.uk/team/researchers/saad-jbabdi<http://www.ndcn.ox.ac.uk/team/researchers/saad-jbabdi>
|