Hi Lucia,
No, what you need is store the PDD(s) as calculated by either dtifit
(dti_V1) or bedpostx (dyads1, dyads2, etc.) using spherical coordinate
angles.
You can use the attached matlab script for this. It will output the
merged_thsamples and merged_phsamples files that are used in
probtrackx, with one sample per file. You may then use these within
probtrackx with one sample per voxel.
The only problem with doing this is that by default probtrackx will
perform a probabilistic interpolation of the PDD while tracking. There
is no quick fix for this unless you are prepared to change the code a
little bit.
The interpolation is calculated in tractvolsx.h, and you can easily
change it to nearest neighbour interpolation for your streamline
tractography.
Cheers,
Saad.
On 13 Jul 2009, at 10:59, Lucia Garrido wrote:
> Dear all,
>
> I've been using probabilistic tracking in FSL to identify two
> tracts. Now I
> wanted to use deterministic tracking to identify the same tracts to
> be able
> to compare my results directly with published results from another
> group.
> Can I just run probtrackx again changing the number of samples to 1?
> Or
> should I run bedpostx again to just define one direction per voxel
> (I used
> the defaults in bedpostx)?
>
> Thank you very much.
>
> Best regards,
>
> Lucia
>
--
Saad Jbabdi
Oxford University FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222545 (fax 717)
www.fmrib.ox.ac.uk/~saad
|