Hi everyone,
I’m currently investigating the optimization of parcellation atlases and attempting to recreate MRtrix connectomes using Python. Specifically, I’m using Dipy to load tractography data and Nibabel to work with the HCPMMP atlas. Although my Python-generated connectomes are similar to those produced by MRtrix, I’m unable to achieve an exact 100% match.
I suspect that the discrepancy might be related to the -assignment_end_voxels option. When comparing the assignment file from MRtrix with the one generated through Python, I noticed that some tracts which remain unassigned to any parcel in the MRtrix output are being assigned in my Python version.
Could anyone provide detailed information on how the -assignment_end_voxels option works in MRtrix? Any insights or clarifications on this behavior would be greatly appreciated as it might be key to resolving the differences I’m encountering.
Thanks in advance for your help!
Best regards