I am currently trying to construct a parcel to parcel matrix for the HCP subjects. I have constructed surface files for each seed / target. The tractography does not take that much time, but loading in all the surface files (400) takes a long time, and the whole process requires about 60GB of memory. Any ideas about how to speed this up and use less memory? All I need is the number of streamlines between the seed and each target.
Here is the command for 100206 for the first seed:
probtrackx2 -x /home/mbmbertolero/hcp/bedpost/yeo_surfs/100206/lh.17Networks_LH_VisCent_ExStr_1.gii
-s /home/mbmbertolero/hcp/S900/100206/T1w/Diffusion.bedpostX/merged
-m /home/mbmbertolero/hcp/S900/100206/T1w/Diffusion.bedpostX/nodif_brain_mask.nii.gz
--meshspace=freesurfer
--seedref=/home/mbmbertolero/hcp/bedpost/100206/orig.nii.gz
--xfm=/home/mbmbertolero/hcp/bedpost/100206/freesurfer2struct.mat
-V 2
--omatrix1
-P 100
--distthresh=20
--os2t
--s2tastext
--targetmasks=/home/mbmbertolero/hcp/bedpost/100206/yeo_seeds.txt
--forcedir
--dir=/home/mbmbertolero/hcp/bedpost/100206/results/yeo/lh.17Networks_LH_VisCent_ExStr_1/
|