search

sina-mansour / UKB-connectomics

This repository will host scripts used to map structural and functional brain connectivity matrices for the UK biobank dataset.
https://www.biorxiv.org/content/10.1101/2023.03.10.532036v1
62 stars 7 forks source link

Using crop_at_gmwmi #31

Closed sina-mansour closed 2 years ago

sina-mansour commented 2 years ago

Hi Rob (@Lestropie),

I just came across this issue which I think we had not discussed before and wanted to finalize it before running the test pipeline for the first 100 subjects on spartan.

I noticed that currently, we aren't using the -crop_at_gmwmi option with -act. I was not sure what that option provides above what ACT is accomplishing and I wanted to know if you reckon it would be worth adding it relative to the potential extra time that it may consume? In the tckgen documentation, it states that the flag provides more precise cropping at the boundary, but wasn't sure about the extent to which that impacts the connectomes.

Lestropie commented 2 years ago

I'd omit that option.

What it does is take the final step of the streamline, and shorten it so that the termination point occurs precisely at the isocontour between GM and WM, taking into account interpolation. It means that if you visualise just the streamline terminations, they form a more precise surface, rather than having terminations up to one streamline step size deep within the GM. But in the use case here, the 5TT image doesn't possess partial volume fractions, so refining the streamline terminations to occur precisely at that isocontour will just accentuate the binary nature of the tissue segmentation. Plus, it's primarily for visualisation purposes, which isn't of interest here. The maximum magnitude of difference it could have is one streamline step size, which is considerably smaller than the radial search distance for streamline-node assignment, so it shouldn't be consequential in that respect.

sina-mansour commented 2 years ago

Thanks for the clarification Rob!