[BUG] Fix marching cubes not in the right space when CT is not aligned

[alexrockhill]

Fixes issue related to https://mne.discourse.group/t/wrong-ct-and-t1-alignment-shown-in-gui/7325/2

[alexrockhill]

@larsoner would you happen to know why these tests are failing again?

It might be

Collecting git+https://github.com/pyvista/pyvista@main
  Cloning https://github.com/pyvista/pyvista (to revision main) to /tmp/pip-req-build-d27__lhw
  Running command git clone --filter=blob:none --quiet https://github.com/pyvista/pyvista /tmp/pip-req-build-d27__lhw
  Resolved https://github.com/pyvista/pyvista to commit 098a6beabb4ccc924a789ede86002d9fdf56a8aa
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
  Installing backend dependencies ... done
  Preparing metadata (pyproject.toml) ... done
Collecting PyQt6
  Downloading PyQt6-6.6.1-cp38-abi3-manylinux_2_28_x86_64.whl (7.9 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 7.9/7.9 MB 77.7 MB/s eta 0:00:00

Collecting PyQt6-Qt6!=6.6.1
  Downloading PyQt6_Qt6-6.6.0-py3-none-manylinux_2_28_x86_64.whl (67.4 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 67.4/67.4 MB 38.5 MB/s eta 0:00:00

the PyQt version mismatch here but that's odd that it worked for a bit and then stopped working... I'll try with PyQt6 != 6.6.1

[alexrockhill]

https://output.circle-artifacts.com/output/job/91844d98-d2c7-4156-96da-146da1477aa3/artifacts/0/dev/auto_examples/locate_ieeg_micro.html#sphx-glr-auto-examples-locate-ieeg-micro-py

Looks good to me

[alexrockhill]

Ok, this should handle when images are not LIA now by going through scanner RAS. I also added some very standard boilerplate to the webpage since it was looking very lonely. If there's any issue with that I can split it out. Otherwise, looks good to go.

[alexrockhill]

Well there was actually a larger issue where all the tests were skipped because they didn't have imageio-ffmpeg... there weren't that many failures so it wasn't too hard to fix.

[alexrockhill]

Okay to merge by me

[alexrockhill]

Oh man well I looked again and the 3D rendering of electrodes when the CT is not aligned were off so then I had to think about with no recon-all what space the 3D rendering should be in. On main it's in surface RAS but if it's surface RAS of an unaligned CT this all just seems so unnecessarily complicated so I streamlined the workflow a bit more:

• The base image is loaded
  • If it is aligned to a recon, show the brain and head surfaces in scanner RAS (aligned)
  • Else just do marching cubes on the image and also show that in scanner RAS
• Electrode positions are in head, convert to mri (surface RAS) of base image using trans then convert to scanner RAS

This way everything internally is in scanner RAS and I think it just simplifies a lot all these transforms which caused this bug.

[larsoner]

This way everything internally is in scanner RAS and I think it just simplifies a lot all these transforms which caused this bug.

Agreed sometimes surface RAS is a pain, and scanner RAS is what nibabel uses everywhere so if it makes stuff easier to work in scanner RAS it's a reasonable choice!

Feel free to merge if you're happy, just one comment about 1.6

[alexrockhill]

Ok, thanks for the review, the last commit just finishes moving the 3D surfaces to scanner RAS, good that you approve of doing that.

Sounds good, I'll wait a day or so in case I hear back from the people who reported the bug if they have a chance to test this before merging.

Re 1.6, it's odd that it fails on stable still. I guess it's fine just to wait to move to stable until 1.7...

[alexrockhill]

All the examples look the same: https://output.circle-artifacts.com/output/job/aea15c4c-5071-434b-832d-f56b4ea42749/artifacts/0/dev/auto_examples/index.html

The diff is a bit large to review but will merge tomorrow unless I come upon any other issues merging this into the other two WIP PRs.