This a general question regarding DTI based parcellation that I was hoping
to clarify. In the published methodolodies, it is stated that "probabilistic
tractography is run from every voxel in the respective mask to assess
connectivity in every brain voxel".
Now as I understand things, when you run a single mask in probtrackX the
output image (FDT paths) represents the number of paths from the seed voxel
to a particular brain voxel, summed accross all mask voxels that also pass
through that brain voxel (hence can produce values >5000 default). Does this
mean, if you wanted to examine the connectivity probability distribution of
a region of interest (as in the Tomassini et al, Journal of Neuroscience
2007 paper) and then cluster according your method of choice (K-Means,
Spectral Clustering etc.,) you would have to subdivide your region into all
of it's respective voxels (using matlab find to generate a list I presume)
and run single voxel whole brain tracking for each, or is this information
stored somewhere after seeding a mask? If so is there any advice on batching
this sort of process (hints towards useful command lines)?
Many thanks in advance,
- Chris
|