Thank you Saad,
I will try to contact Moises/Stam from FMRIB to know if the new
version is available (the other two options seem quite complicated to
me). Do you have any suggestion for the --onewaycondition option?
thank you!
Angela
Saad Jbabdi <[log in to unmask]> ha scritto:
> Hi Angela
>
> Depending on the number of seed voxels it can indeed take a while.
>
> You can accelerate it in three ways:
>
> 1) split the seed mask into N seeds and run them in parallel then
> combine the results
>
> 2) split the number of samples into N*n where N run in parallel,
> each job running n samples (don’t forget to include the —rseed
> option so you get different results for each parallel run)
>
> 3) contact Moises/Stam from FMRIB to get hold of our new GPU version
> (still under testing but insanely fast!)
>
> Cheers
> Saad
>
>
>
>
>
> On 23 May 2016, at 08:52, Angela Favaro
> <[log in to unmask]<mailto:[log in to unmask]>> wrote:
>
> Dear all,
> Sorry to repeat my question, but I received no message in reply to
> my previous mail (is it not appropriate in some way?)
> I am testing the command line below and the first matrix needed more
> than 3 days for output.
> I would like to have some suggestion on how to speed out the process
> (if at all possible) or to know if this is due to something wrong in
> my command line
>
> probtrackx2 --network -x ${subject}/masks_surface.txt -V 1
> --omatrix1 -c 0.2 -S 2000 --steplength=0.5 -P 5000 --forcedir --opd
> -s ${subject}/dti.bedpostX/merged -m
> ${subject}/dti.bedpostX/nodif_brain_mask
> --dir=${subject}/track_graph
> --avoid=${subject}/dti/anat/ventricles.nii.gz
> --xfm=${subject}/dti.bedpostX/xfms/fs2fa.mat
> --seedref=${subject}/dti/anat/fs.nii.gz
> --waypoints=${subject}/dti/waypoints.txt --waycond='OR'
> --meshspace=freesurfer
>
> masks.txt, ventricles and wm.lh wm.rh (waypoints) are all in
> Freesurfer's space (fs)
> fs2fa.mat was created as suggested in the fsl site
> the aim is to create a matrix to be used for graph analysis. Do I
> have to use the option --onewaycondition? this point is not clear to
> me
>
> Thank you again for any help!
> Angela
>
> Dear FSL experts,
> I need some help in using probtrackx2.
> I want to use Destrieux parcellation surfaces obtained by using
> Freesurfer (and some subcortical areas) to perform probabilistic
> tracking in order to create a matrix to perform graph analyses.
>
> I looked at the instructions and previous posting and I am trying
> with this command line (in a script):
>
> probtrackx2 --network -x ${dir}/${subject}/dti/masks.txt -V 1
> --omatrix1 -c 0.2 -S 2000 --steplength=0.5 -P 5000 --forcedir --opd
> -s ${dir}/${subject}/dti.bedpostX/merged -m
> ${dir}/${subject}/dti.bedpostX/nodif_brain_mask
> --dir=${dir}/${subject}/track_graph
> --avoid=${dir}/${subject}/dti/anat/ventricles.nii.gz
> --xfm=${dir}/${subject}/dti.bedpostX/xfms/fs2fa.mat
> --seedref=${dir}/${subject}/dti/anat/fs.nii.gz
> --waypoints=${dir}/${subject}/dti/waypoints.txt --waycond='OR'
> --meshspace=freesurfer
>
> masks.txt is a list of all masks created by Freesurfer (with the
> commands mri_annotation2label and mri_label2vol) and so they are in
> Freesurfer's space (fs)
> ventricles and wm.lh wm.rh (as well as subcortical nuclei) were
> obtained with the command mri_binarize (wm.lh are wm.rh are
> waypoints.txt) and are in Freesurfer's space
>
> is this command correct? do I have to also include the inverse
> matrix? ----invxfm=xfms/fa2fs ?
> is it correct the option --waycond='OR'
> do I have to use the option --onewaycondition ?
>
> please give some help! this is a very complex issue and I do not
> want to make mistakes (and loose hours of computation....)
>
> Thank you!
>
> Angela
|