1) I guess I should type in the command line probtrackx --network... etc and at the end put --omatrix1 as you said. Then the output as you seed is a matrix with the number os stramlines between any two voxels. But then you said from that matix it can be stracted the number of tracks between each mask (ROI) summing the appropiate rows... but how do I know which are the appropiate ROIs??
probtrackx will also produce a lookup table (in the form of a nifti file) that you can use to figure out which voxels in your ROIs correspond to which entry in the matrix. I would recommend using matlab to post-process the matrix and get ROI-based connectivities.
2) If I have n(i) number of voxels in each mask, what would be the approximate execution time? I mean, aproximately how long it takes per voxel?
The exact amount of time spent per voxel will depend on the number of samples you are seeding per voxel (as well as your computer's performance!). I would suggest doing a test run with few samples (e.g. 1 per voxel), look at the time it took, and multiply that time by the number of samples you will use in the real experiment.
3) in the porbtrackx -help it doesnt appear --omatrix1 ... so i dont know if after eat I should but the name i desire for the output matrix file or just put the flag and the program will put a name... also i didnt understand what --s2tastext means.
It is a hidden option for now (so undocumented and unsupported!), that is why it does not appear. It is quite straightforward to use though, all you need to do is add the --omatrix1 flag.
--s2tastext is unrelated. It is useful if you want seed-to-target results as a ascii text file (on top of the nifti files seeds_to_<targets> that are produced if you use the --os2t option - which is documented).
Cheers,
Saad.
Thanks,
2011/3/11 Saad Jbabdi
<[log in to unmask]>
Hi Jose
If you add the --omatrix1 flag to probtrackx, it will output a matrix that counts the number of streamlines between any pair of voxels in your seed mask. So if your seed mask is the summation of all your masks, you will be able to extract the connectivity information between your N masks by summing across the appropriate rows of the matrix, without the need to run probtrackx N times.
What you will *not* be able to do is have these masks as termination masks as well (i.e. inforce that the streamlines do terminate when they reach one of the masks), because then none of the streamlines will even leave the seed voxels.
cheers,
Saad.
On 11 Mar 2011, at 12:13, Jose Angel Pineda Pardo wrote:
> Hi everyone,
>
> Im working with brain connectivity, and Im trying to build a connectivity matrix between a number of structures of the brain, all of them defined as region of interest in different nifti files.
>
> My question is about the execution time, because I want to be sure that Im doing it well. The pipeline is the next:
>
> Execution of BedpostX (takes about 4 hours for 1 fiber orientation)
> I build my different ROIs on the diffusion space, and generate the nifti files.
> And finally, execution of ProbTrackX as follows.
> From Multiple masks (all ROIs) to a termination mask. I will perform that N times, 1 for each possible termination mask. And finally, merging all the numbers from the "waytotal" file in the ProbTrackX I create my connectivity matrix. The problem is that each execution for a single termination mask takes about 24 hours, so it will take 90 days for creating the matrix between 90 masks. Is it that correct? Another cheaper possibility for performing that process.
>
> Thanks a lot
>
> Cheers
>
> Jose
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
--
Monica Giraldo
Master student in Biomedical Engineering
Universidad de Barcelona - Universidad Politécnica de Cataluña
