Assignment of streamlines between two regions using "connectome2tck"

connectomics

#1

Dear all experts,

I want to see the streamlines connecting two regions.

So after generating global tractography using tckgen , I used tck2connectome to get the connectome matrix of some ROIs. Then I used connectome2tck to extract the streamlines between any two nodes of those ROIs.

However,it seems likely that there are some wrong assignments of streamlines. What it means that when I open the track between nodes1 and nodes2 using mrview GUI, for example, “streamlineofnodes1-2.tck” , there are some streamlines assigned in nodes3. In fact, the nodes3 is rather far away from nodes1 and nodes2.

I don’t know if there are wrong during I processed.

Here are the commands I used:

tckgen -algorithm Tensor_Prob -select 200000 -angle 20 -cutoff 0.1 -seed_image b0_mask.mif -fslgrad bvecs bvals DWI_200K.tck

tck2connectome DWI_200K.tck ROIs_labelconvert.mif tensor_200K_connectome.csv -assignment_radial_search 0.4 -out_assignments streamlineofnodes

connectome2tck DWI_200K.tck streamlineofnodes streamlineofnodes_out -files per_edge

By the way, the datas I analyzed were rat brain, the voulme is 0.001 times of human brain. So the assignment_radial_search I chose was 0.4

Thanks very much for your patience.
Zhaoqing Li


#2

Hi Zhaoqing,

I would expect that if tck2connectome were making gross mis-assignments of streamlines to parcellation nodes, it would have been spotted by now; so I’m hoping that there is some other explanation to what you’re observing. However I’d need to see at the very least a screenshot demonstrating the issue, but preferably the data with which I can re-produce and examine the described effect.

Rob


#3

Sorry for my ambiguous demonstration. I think the mis-assignment is obvious.
The screenshot shows the issue I confronted. There is a wrong streamlines connected to node 3.
By the way, where can I send my raw data to you?
Thanks very much.
screenshot


#4

If I had to guess, I’d say that the streamline looks like it’s heading towards node 3, but if you actually trace the streamline along its entire trajectory, it somehow loops back and terminates at/near node 2. There’s a couple of things you could experiment with here to see whether or not this is the case:

  • Use tckedit with a manually-defined ROI to extract just the streamline(s) that you believe to be problematic.

  • Use tckresample -endpoints, combined with the “geometry” setting within the mrview Tractography tool, to better visualise where the streamlines actually terminate (as opposed to where they traverse).

If you remain convinced that there is in fact a software issue, you can send data to:
robert.smith AT florey.edu.au


#5

Thanks very much. I used tckedit to extract the streamlins. It likes what you said exactly. It loops back and terminates at node 2.