Tck2connectome in lesioned brains

1

Hello everyone,

I am doing probabilistic tractography in lesioned brains and was wondering how fibers with an endpoint within the lesion are treated in tck2connectome? As they do not connect 2 parcels but are only connected to one parcel, do they get discarded within the matrix?

I would appreciate your help on this.

Thank you,
Franziska

2

Hi Franziska,

Streamlines that terminate within or at the lesion boundary are generally excluded from the connectome matrix because they do not connect two distinct nodes in the parcellation image. However, if a lesion overlaps with a node, it may still be included in the matrix.

To include streamlines terminating within or at the lesion boundary in the connectome, the lesion can be incorporated into the connectome by assigning it a unique label and merging it with the parcellation scheme. This can be achieved by indexing the lesion mask with a unique number, and adding it to the parcellation image (take care to not combine it with an already existing node). mrcalc can be used for both the reindexing of the lesion mask and its integration with the parcellation image.

Hope this helps!

Cheers,
Arkiev

1 Like
3

Hello Arkiev,

Thanks a lot for your helpful explanation and suggestion!

I am using the AAL116 atlas and added the lesion mask as an additional 117th node to the atlas. There is an overlap between the lesion mask node and other atlas nodes. I noticed that in this scenario, the previous node label, e.g. label 30 gets reassigned a new label (old label plus new label, thus 30 + 117 = 147). This means that my connectome matrix will change its format from 116x116 to a new format with the number of rows and columns depending on the number of nodes overlapping with the newly added lesion mask node? In my case, the matrix extended to 202x202 with lots of empty rows between row 118 and 202…

What I would like to do, is extracting a node-specific tractogram of all fibers connecting node A to the rest of the brain. This tractogram should include all streamlines connecting node A to node B, node A to node C, etc. as well as node A to the lesion/lesion border (node X). The overlap between node X and other atlas nodes should not cause an issue in the extraction of this node-specific tractogram, as all streamlines that have node A listed as one of their endpoints get extracted, is that correct?

Thanks a lot for your help!!
Franziska

4

Hi Franziska,

Yes, that’s correct—though I’m not entirely convinced that this is the intended approach.

Generating a whole-brain connectome matrix could be useful here, as it would provide connectivity (and corresponding streamlines) between nodes A-B, A-C, …, A-X, and more generally, could be used to extract streamlines between any two nodes in the parcellation image.

I might need some additional clarification about the lesion… Is the goal to:

  1. Treat the lesion as its own unique node? In this case, the connectome matrix would be 117x117, with the lesion indexed as 117. Areas where the lesion overlaps with another AAL node would be defined solely as lesion.
  2. Consider the lesion as several discrete nodes, where each subregion of the lesion corresponds to an overlapping region with the AAL atlas? Here, the non-overlapping region of the lesion could be indexed as 117, while a lesion subregion overlapping with node 30 (which has been indexed as 147) could be reindexed as node 118. In this scenario, the connectome matrix would be larger than 117, with nodes greater than 117 representing regions of the lesion that intersect with AAL nodes.

Keep in mind that with approach 1, all participants in the connectome matrix will have a connectome matrix of size 117x117. Conversely, in 2, it is possible that the connectome matrix will not have a consistent size across all participants…

Best,
Arkiev

1 Like
5

Hello Arkiev,

The goal is approach 1, to treat the lesion as its unique node, labelled as 117.

After successfully adding the lesion mask as 117th node in the AAL-atlas, I received the desired 117x117 connectome matrix:
tck2connectome -symmetric -zero_diagonal -scale_invnodevol -tck_weights_in ifod2_sift2.txt
ifod2_MNI_space.tck aal116_lesion_added.nii tck_connectome_aal116.csv -out_assignment assignment_tck_connectome_aal116.csv

To extract all streamlines of this whole-brain connectome matrix that connect node A to any other node in the atlas (including the newly added lesion mask node), I run connectome2tck ifod2_MNI_space.tck assignment_tck_connectome_aal116.csv -nodes 01 -files -per_node AAL116

Does this seem correct to you? Is it appropriate to use tcksift2 after running the tractography to refine the streamlines or would this be biased by the presence of the lesion?

Thanks a lot for all your help,
Franziska

6

Hi Franziska,

This seems sensible, although it might be worth including the option -prefix_tck_weights_out. This will extract not only the streamlines from node 01, but also, the corresponding weights for each streamline (which could be useful).

Best,
Arkiev

1 Like
7

Hi Arkiev,

Great - thanks again a lot for your help!! You gave me a much better idea of what is happening behind the scenes.

Best,
Franziska

1 Like