Dear Marc,
Thanks for the files.
I have figured out what the problem is. The reason is related to the
image resolution, as you guessed. We currently do not compensate
for large changes in voxel volume in tbss_deproject in the same way
that flirt compensates in the affine case. Consequently, it is easy for
voxels in the 1mm^3 space to get missed out when doing the
deprojection (as it only looks at one coordinate per voxel in the
original space).
We will work on a solution to this problem and get back to you
soon. A temporary solution is to backup your image containing
the blobs of interest and substantially dilate those in the original
file prior to running tbss_deproject. Using a 3x3x7 kernel did the
trick for me when I tried it. You can do this with fslcreatehd and
fslmaths. However, we should have a better solution early next
week if you can wait.
All the best,
Mark
On 2 Jan 2008, at 20:13, Marc J. Dubin wrote:
> Mark,
>
> Thanks, This still is not resolved. I will upload the tarred images
> in the
> next 1-2 days.
>
> I really appreciate your help with this!!
>
> Marc
>
> On 1/2/08, Mark Jenkinson <[log in to unmask] > wrote:Hi,
>
> Is this resolved?
> I doubt it is an issue of resolution, as you should get non-zero
> partial volume still. Also, if it works for some of these images
> at this resolution, then it should work for all.
>
> If you are still having problems with this, please upload a tarred
> set of images for one of the offending subjects and we'll have
> a look.
>
> All the best,
> Mark
>
>
> On 22 Dec 2007, at 13:03, Marc J. Dubin wrote:
>
> > I am using the default, which I believe is trilinear....
> >
> > Could it have something to do with the resolution of the
> > native space images, which is 1.7x1.7x5mm?
> >
> > Best,
> > Marc
> >
> > On 12/22/07, Steve Smith <[log in to unmask] > wrote:Hi, this is
> > probably because you are using the -n interpolation option.
> > Have you tried without this option?
> >
> > Cheers, Steve
> >
> ----------------------------------------------------------------------
> > --
> > 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] http://www.fmrib.ox.ac.uk/~steve
> >
> ----------------------------------------------------------------------
> > --
> >
> > On 21 Dec 2007, at 05:34, Marc Dubin <[log in to unmask]> wrote:
> >
> > > Dear FSL Users,
> > >
> > > I am using TBSS_deproject 3 to invert the TBSS from MNI space
> back
> > > to native space.
> > > I create grot file from a 45 pixel blob in a tract of interest in
> > > MNI152. TBSS_deproject 3
> > > then runs without errors and looking at the native space images, I
> > > am able to find the
> > > inverted blob for most subjects, but there are several subjects in
> > > which the blob is nowhere to
> > > be found and in fact running fslstats on these native space images
> > > reveals that in fact there
> > > are no nonzero voxels.
> > >
> > > Any suggestions on what might be going on with these subjects
> would
> > > be greatly appeciated.
> > >
> > > Sincerely,
> > > Marc
> > >
> >
> >
> >
> > --
> > Marc Dubin, MD PhD
> > Resident in Psychiatry - PGY4
> > Weill Cornell Medical College
> >
> > 212-746-3764 (work), 646-831-8886 (cell)
>
>
>
> --
> Marc Dubin, MD PhD
> Resident in Psychiatry - PGY4
> Weill Cornell Medical College
>
> 212-746-3764 (work), 646-831-8886 (cell)
|