streamline count between two ROIs

I am new to Mrtrix3. I thoroughly studied the BATMAN Tutorial and questions asked by other mrtrix users however I am still a bit confused - maybe you can help …:slight_smile:

For my project I need to get the streamline count between a seed and a target region (which are both labels from the DKT atlas of Freesurfer). The streamlines do not have to end in the target region, it is enough if they traverse it. The direction, whether the streamlines go from seed to target or from target to seed is also irrelevant.

After reading through all sorts of posts on this topic, I feel like there are several possibilities for this and am unsure which one is the “correct” one (or if there are several correct ones).

  1. One possibility seems to be to make a whole-brain tractography using tckgen, then tcksift, and then construct a connectome using tck2connectome. Using connectome2tck afterwards, I should be able to get a trackfile between the two nodes of the DKT parcellation atlas. Using tckinfo I should get the streamline count, right?
    However as I understand, only the streamlines that end in the ROIs are counted here and not the ones that pass through it. Is this correct? Would I get around this by using the -assignment_all_voxels option?

  2. Another option seems to be to get the streamlines between two regions using tckgen by specifying a seed image and inclusion mask, where you don’t just get the streamlines that end in the inclusion mask, but also the ones that go through them.
    2a) I read, that you can apply SIFT only to whole-brain tractography, which means it wouldn´t be applicable in this case?
    2b) In the tck2connectome command you can specify the scale_invnodevol option. Is there a way I can consider the size of my ROIs when I do not use tck2connectome that has that option??

  3. Then there seems to be a possibility to use tckedit to edit trackfiles already generated with tckgen - is it possible to extract streamlines between two ROIs from a whole-brain tractography using tckedit and use the -include option twice (so -include ROI1 -include ROI2)?

One last related question: doing functional connectivity analysis we mostly use dilated atlas-labels. Would you also recommend to dilated ROIs in structural connectivity analysis? If yes: I guess you cannot dilate a whole parcellation-image, therefore getting the streamline count between two dilated ROIs by using connectome2tck isn´t possible in that case, right?

I would really appreciate some guidance on how you would approach this analysis.

Thanks in advance,

Hi Pia,

Maybe I can provide some advice.

In my point of view, Method 3 is the most direct way.
Method 1 is also a choice, I’m interested in the difference between these two methods, and I’d appreciate it if you can share your practical experience.
For Method 2, sure SIFT cannot be used in this case and I think the whole-brain tractography should be generated because we always need to prepare for the rainy days. :relieved:

To me, I think there is no need to use dilated ROIs but if the dilated ROIs are pre-defined, it’s possible to get the streamline count between two dilated ROIs.
I would like to point out that if the FCs and SCs are analyzed federally, it’s necessary to align the ROIs used for FCs and SCs. A potential problem (maybe not correct and I would like to hear your advice) is that FCs are mainly analyzed in grey matter and SCs in white matter (through 5ttgen and 5tt2gmwmi), how to deal with it?

Hope to your reply.