I am trying to perform k-means clustering on my seed mask using the matlab k-means function as described on the FSL website (with 2 clusters). I'm running probtrackx2 from my single seed mask to 7 target masks. When I used the following probtrackx2 call I was able to obtain clusters:
probtrackx2 -x seedmask -l --omatrix2 --target=NODIF_B_MASK_subsamp2.nii.gz --avoid=ventricles --pd -c 0.2 -S 2000 --steplength=0.5 -P 5000 --fibthresh=0.01 --distthresh=1 --sampvox=0.0 --forcedir --opd -s Diffusion.bedpostX/merged -m Diffusion.bedpostX/nodif_brain_mask.nii.gz --dir=$PROBTRACK_DIR_R --targetmasks=targetmasks.txt --os2t
HOWEVER I want to add several options to my command but when I do this and attempt k-means clustering again, I get an error in matlab: "error: kmeans: empty cluster created"
This is my new probtrackx2 command where I added modeuler, made the targets "stop exit masks" with --wtstop, added white matter as a waypoint, and added the onewaycondition. I also added otargetpaths to get individual fdt_paths for each target. I fail to get clusters now. Any other comments regarding these additional options would be appreciated. My goal is to calculate the probability of connectivity from my single seed to 7 targets and to do this for 50 subjects and then look at the seed clustering and correlate that with some behavioral measures. Am I simply not getting enough tracts now with the additional options?
probtrackx2 -x seedmask -l --wtstop=wtstop.txt --otargetpaths --omatrix2 --target=NODIF_B_MASK_subsamp2.nii.gz --avoid=ventricles --modeuler --waypoints=WM_mask.nii.gz --onewaycondition --pd -c 0.2 -S 2000 --steplength=0.5 -P 5000 --fibthresh=0.01 --distthresh=1 --sampvox=0.0 --forcedir --opd -s Diffusion.bedpostX/merged -m Diffusion.bedpostX/nodif_brain_mask.nii.gz --dir=$PROBTRACK_DIR_R --targetmasks=targetmasks.txt --os2t
|