Content-type: text/html
Computes a connectivity based segmentation of a seed region. A connectivity vector is output for each seed point detailing streamline counts to each target. Additionally, two images are produced, one containing the target region with the largest streamline count to each seed, and one containing the respective value.
Input tracts should be in the same physical space as the seed and target images.
Three files will be produced as output, with extensions
sc.csv - The connectivity matrix. Each row in this matrix corresponds to the connectivity from one seed point. The first five columns describe the seed point X,Y,Z coordinates in physical space, the label of the seed ROI at that point, and the total number of input streamlines for this seed point. The remaining columns list the number of streamline that connect from the seed point to each of the target regions.
cbs_labels.nii.gz - The connectivity segmentation. Each seed point is labeled with the target region with the largest streamline count.
cbs_sc.nii.gz - The streamline counts corresponding to the target region with the largest streamline count.
If there are multiple seed points per voxel, earlier seeds will be overwritten in the output images by later ones, though both will exist in the matrix. In the event of two or more targets having the same streamline count for a given seed, the first one (lowest label index) is chosen.
The output of cbsmat has some similarity to that produced by procstreamlines and targetprobs2txt. There are some important differences between the output you would get from the two routes. The main difference is in normalization and the handling of seed points in voxels containing multiple principal directions (PDs). cbsmat
- Takes all streamlines from a seed point and processes them together.
- Does not perform any normalization of streamline counts, however the total number of streamlines input for each seed point is recorded, which can be used for normalization similar to that of procstreamlines
- Outputs seed locations in physical space
- Allows the user to label the target regions, or does so automatically
- Outputs the seed ROI label, which facilitates clustering voxelwise connectivity stats over larger regions.
The main benefit to treating all streamlines from a single seed point as one unit is that the user is free to pre-process the tracts with one or more runs of procstreamlines. There is no requirement to perform all pre-processing in one pass, as there is when producing a connectivity segmentation with procstreamlines.
The cost of not treating each principal direction separately is that it may complicate interpretation of the results, especially in cases where the connectivity of streamlines seeded along two or more principal directions overlap. This may be resolved in a future release by incorporating the principal direction index into the streamlines and thus into the matrix.
cat tracts.Bdouble | cbsmat -seedfile seeds.nii.gz -targetfile targets.nii.gz -outputroot cbsmat_
1 some_brain_region
2 some_other_region
These names will be used in the output. The names themselves should not contain spaces or commas. The labels may be in any order but the output matrices will be ordered by label intensity.