Strange loop tracts are generated


#1

Hi Mrtrix experts.

I used mrtrix to build the structural connectivity to analyze the alternation of schizophrenia subjects connectivity.
Then, I created the connectivity matrix by these commands, and I got the fibers like this.
Unfortunately, I found some of the loop strange fibers in some sub-networks, and I think these tracts are not suited from biological aspects.
So, I want to know why these loop fibers were generated, and how to overcome this problem.
My data is single shell low b-value (1000) and 16 direction data, and this color scheme of this visualization is endpoint based.

(result)

(command)

# Diffusion image processing by mrtrix
  mrconvert reg_eddy DWI.mif -fslgrad GE15dir-b1000.bvec GE15dir-b1000.bval
  dwi2response dhollander DWI.mif RF_WM.txt RF_GM.txt RF_CSF.txt
  dwi2fod msmt_csd DWI.mif RF_WM.txt WM_FODs.mif RF_GM.txt GM_FODs.mif RF_CSF.txt CSF_FODs.mif

  # Diffusion tractography processing by mrtrix
  tckgen WM_FODs.mif 5M.tck -act 5TT.mif -backtrack -crop_at_gmwmi -seed_dynamic WM_FODs.mif -maxlength 250 -number 5M

 tck2connectome 5M.tck subj_AAL_fix_no_cellebrum.nii.gz no_cellebrum.csv -zero_diagonal -out_assignments no_cellebrum_assignments.txt
 mkdir connetome
 cd connetome
 connectome2tck -nodes 3,6,7,9,42 ../5M.tck ../no_cellebrum_assignments.txt tracts

#2

It looks like these are just a few spurious fibres. If you’re after a specific bundle (as your screenshot suggests), you can just add a few exclusion regions to get rid of these.

Unless of course everything you’re showing there is what you’re referring to as the looping fibres. In that case, you’ll need to provide some more context to allow people to give you advice… it’s quite hard to see what structure you’re showing there… Maybe add an anatomical volume for reference, or show the situation from a few different angles.

In general though: spurious fibres are never unexpected. In fact, Maxime Descoteaux recently showed, based on extensive data from a tractography challenge, that up to 70% of all bundles generated are typically entirely false positives (I’m talking entire bundles here, not just a few spurious streamlines). With that as a context, I wouldn’t worry about a few spurious streamlines… as long as there’s only a few compared to millions of streamlines in your connectome as a whole.