Tck2connectome output, Structural Connectome values are Non integer values

SC is the count of streamlines from one region to every other region.
I am using tck2connectome command to generate SC from my parcellated image.
The command is given below:

tck2connectome -symmetric -zero_diagonal -scale_length -tck_weights_in sift_10M.txt tracks_10M.tck sub_parcellated.mif SC.csv -out_assignment assignments_SC.csv

Now, SC.csv file has all non-interger values.
So my question is if SC is the count of the number of streamlines, how can it be some fraction (non-integer) value?

I have tried with other -scale_* options. But same result.
What i should do to get the integer count of streamlines for each pair of nodes?

Dear @math_priyanka ,

The non-integer values come from weighting the number of streamlines between regions (with the -tck_weights_in option). These weights were calculated when you ran tcksift2 and are important to account for tractography biases.

Please have a read through this paper which tries to explain what the resulting structural connectivity metric (fibre bundle capacity) represents.

If you still want to use the number of streamlines as your outcome, you should use tcksift instead, and you might want to scale up your tractogram before filtering (approx 10x more streamlines than you want in your filtered tractogram).

Hope this helps,

Ok. I will try with tckshift. I am not biased with integer values of SC rather I want to know if I am doing something wrong with preprocessing. I will read the paper you shared.
Thank you.

1 Like