Hey MRtrix community,
I’m essentially trying to generate an intersection mask from my spherical ROIs and the grey matter-white matter interface mask. The ROIs were generated using coordinates from a Freesurfer white matter image, and I’ve overlaid them onto the 5ttgen output image to make sure they actually do overlap. However, when I pump both files into 5ttgmwmi like so:
5tt2gmwmi -mask_in ROI.nii.gz 5ttgen_output.mif ROI_gmwmi.nii.gz
I get the error:
5tt2gmwmi: [SYSTEM FATAL CODE: SIGSEGV (11)] Segmentation fault: Invalid memory access
Any clues on how I might deal with this? Can I use a combination of mask and gmwmi input options to tckgen to work around this?
This is clearly not a commonly used feature; there’s a bug with
-mask option that’s been there for quite a while.
The fix for this particular issue will hopefully be pushed out soon, at which point updating your MRtrix3 installation (assuming you’re on the
master branch) and re-compiling should make the command behave as expected.
Thanks for the report
Thanks for the follow-up, Rob! I’ll keep my eye on the GitHub page.
Since I’m apparently off the beaten trail, here, is there a more commonly used approach to seeding from the intersection between an ROI (spherical, anatomical, etc) and the grey matter-white matter interface?
Historically, i.e. without ACT, such seeding was simply done using the ROIs as volumes, and there was no issue. Unfortunately as it stands now ACT is slightly clunky in its handling of manual grey matter ROIs / seeds. I’m working on a raft of changes around the tracking code, which will include more robust assignment of streamline terminations at the GM-WM interface to GM ROIs / parcellations. I’m thinking about how to make similar alterations to the GM-WM interface such that ROIs can be drawn within the GM and the seed will be placed at the corresponding interface location. Eventually it will also be possible to define such seeds using surface-based ROIs, but this is ongoing development also.