Hi all,
I’m interested in a specific fiber connecting two nuclei. The goal is to have a connectivity measure (e.g., the sum of post-SIFT2 weights) for further statistical analysis.
I’ve gone through tons of issues/guides posted on the forum, such as Reconstruction of bundles of interest, Streamlines between two ROI, Extract pathway after Sift2, and Tckedit after targeted tracking, BUT still, the problem persists.
The tract of interest is small; thus, it is hard to extract it from whole-brain tractography. So, I did a 50M whole-brain + 15K targeted tractography. Interestingly, when I combine these two files and try to extract the same 15K fibers from this combined tractography file using the same criteria used in the initial tractography phase, it just gives me 13.5K fibers. This is thoroughly talked about in Tckedit after targeted tracking. Upsampling the targeted tractography helps (post-combination extraction fibers go above 14K), however, when the “sum of weights” is the main measure of interest, different fiber counts after tckedit substantially affect the summation.
I ended up with a solution (based on Tckedit after targeted tracking - #5 by rsmith), and I would highly appreciate any comments. I would also like to know what you -MRtrix lovers- think of my approach.
The steps were:
- Upsample the targeted tractography by factor of 3
- Refine the fiber selection by including the same criteria used for generating the targeted tractography with the addition of two filters:
count = 7500and-maxlengthset to the “median” (extracted fromtckstats targeted_tractography_upsampled.tck) - Combine targeted with whole-brain
- Apply SIFT2
- Tckedit the combined tractography with the same criteria used in step 2
This approach ensures (I guess):
- Constant fiber count across subjects
- Include the majority of the fibers connecting the two tiny nuclei
However, as I said above, I’m not 100% sure whether this is a reasonable approach to this issue or not, and would love to hear your comments 
Best,
Amir