Dear experts,
I’m using the TractoPrep container TractoPrep Processing Manual — TractoPrep v1.0.6 documentation to obtain connectome from my DWI data.
It leads from preprocessing to the final connectome across different steps, mostly automatically (using several mrtrix and fsl commands).
I have different DWI acquisitions, single- and multi-shell, but the container seems to handle both of them.
However, for a considerable proportion of subjects acquired with a specific sequence, the final connectome presents a lot of empty nodes (totally disconnected ROIs). This specific acquisition is single-shell and includes 35 directions and b=900. Here attach an example of the QC after the tracto step (which basically works with dwi2fod, tckgen and tcksift).
My first question is: what could be the problem that creates several disconnected nodes?
And then: what is a kind of cutoff to decide whether to use or exclude a connectome for analyses (in general, also when working with the standard mrtrix pipeline)?
Last: is it really appropriate to also use single-shell data to build a connectome?
Thank you so much in advance for any help!
Best,
Beatrice