Dear Stuart,
The probtrack algorithm has a random component to it, and the default random selection will differ on different machines. Using the -V option in fslstats will tell you about all the voxels that had even a single streamline pass through them, and so is an overly sensitive measure of the differences. You should see that the results are actually quite similar, especially when you threshold to get rid of the very unlikely tracts (with low counts).
If you want to see what options are used by the command line, then look at the output of the GUI in the terminal, as the FSL GUIs display their command line calls so that they can be replicated outside of the GUI.
All the best,
Mark
On 24 Jan 2014, at 00:20, Stuart Traxel <[log in to unmask]> wrote:
> I am trying to run probtrackx2 on two different systems, a Mac and a Linux based cluster, but the results aren’t matching up. When I run "fslstats fdt_pathx.nii.gz -V", different values are printed depending on which system I ran the probtrackx2 command on despite using the same files (seed, waypoints, etc.) for both. The FSL versions are the same (5.0.1), so could a difference in OS or the system cause different results when running probtrackx2? Any other ideas? The commands and flags are the same for both.
>
> Also, what is the command or script used to construct waypoints.txt by the GUI and is there source code for probtrackx2 somewhere?
|