Ali Khan

Looks like these modifications might need widespread testing for robustness first -- When I tried the above `laminar_coords_res` modification (changing hipp to 0.1x0.1x0.1), it ends up fixing the holes for...

btw, I've just added the mentioned commits (the unrelated cherry-pick from another branch, and the flag to make pymeshfix only perform on hipp, not dentate).. this still needs testing on...

Sounds good I agree #503 is a good solution (once we confirm it), as it is the least "destructive".. For posterity - the laminar_coords_res 0.1 approach did work to solve...

Ahh-- looks like this doesn't resolve the holes in the test dataset (HCP sub-129028), one hole still remains.. I'm trying now to see whether a combination of upsampled resolution and...

Having a closer look at it, it seems the holes may actually be a red herring (in this case, and perhaps others too) -- looks like the underlying problem is...

correction from my last message -- the holes and src/sink discontinuities are two separate issues, what was confusing things in my test case is that the changes to fix the...

ok some updates so far: I've run all 100 hcpUR100 subjects with the aforementioned PRs, but with the original 0.3mm resampling, out of these, only 2 structures failed: sub-129028_hemi-L_label-hipp (the...

Ok, the run with 0.2mm resampling ended up with just the single error ( sub-190031_hemi-L_label-dentate) as hoped (dentate was unchanged so didn't expect this to be fixed).. I think we...

Yes new NNs should hopefully help.. Another addition I was about to make is to check for holes as a QC output -- in the get_boundary_vertices.py script we just about...

For your first issue, it could be related to how the CLI is being formed, as snakebids will just use that as the target rule. You could try using verbose...