Print

Print


Hi Moran,
If you have a GPU with 3 or more GB of memory, you can use the CUDA version of probtrackx:

You can find here the binaries for different versions of CUDA:
http://users.fmrib.ox.ac.uk/~moisesf/Probtrackx_gpu/

It is about 70 times faster for generating a whole connectivity matrix.

Moises.

On 21 April 2016 at 13:25, Moran Artzi <[log in to unmask]> wrote:
Hi all,
I'm trying to perform DTI connectivity between all seeds in a given template using FSL Fdt tool  (such as Harvard oxford cortical structural atlas, after transfer the DTI & templates VOI to the same space)
What is the most recommended way?
(1 ) Regards the Matrix options, Matrix 1 option seems appropriate  but the running time for multiple masks is very long and not applicable. Is there any way to accelerate the running time / calculate connectivity to the entire VOI (and not to each voxel with in it?
(2) Regards the analysis results, what is the best way to normalize the fdt_network_matrix file results across subjects (given different member of voxels for the different masks across subjects)?

Many thanks
Moran