Brain GUI modernization (Phase 4)

[payamsash]

Reference issue (if any)

Phase 4 of #14042 and following #14085

What does this implement/fix?

add surf and cortex options to GUI

Additional information

Claude reviewed the code and I understand what has been modified.

[wmvanvliet]

this doesn't test whether the new colors are actually shown, like you do below with changing surf

[wmvanvliet]

Why a new smooth parameter? We already had controls for the amount of smoothing, which you could set to 1 to turn it off.

[payamsash]

smooth_mat lives on the whole mesh but the new smooth just tracks per-overlay.

[payamsash]

ready to be reviewed @wmvanvliet , tnx in advance, phase 5 is almost ready ...

[wmvanvliet]

When selecting some vertices, the spheres don't update when switching surfaces.

Have you tried this with a VectorSourceEstimate?

[payamsash]

Sharp eyes! you can now test it with this small script:

import mne
from mne.datasets import sample
from mne.minimum_norm import apply_inverse, make_inverse_operator

data_path = sample.data_path()
subjects_dir = data_path / "subjects"
meg_path = data_path / "MEG" / "sample"

fwd = mne.read_forward_solution(meg_path / "sample_audvis-meg-oct-6-fwd.fif")
fwd = mne.pick_types_forward(fwd, meg=True, eeg=False)
evoked = mne.read_evokeds(meg_path / "sample_audvis-ave.fif", baseline=(None, 0))[0]
noise_cov = mne.read_cov(meg_path / "sample_audvis-cov.fif")

inv = make_inverse_operator(evoked.info, fwd, noise_cov, loose=1.0)
stc = apply_inverse(evoked, inv, pick_ori="vector")

brain = stc.plot(
    subject="sample",
    hemi="lh",
    subjects_dir=subjects_dir,
    colormap="auto",
    show_traces=True,
    initial_time=0.1,
)

[wmvanvliet]

Does this need to be an entire separate method if it's only called once?

[payamsash]

Fair point, if it was only one method, I would fold it into set_surf but now we have two (_reposition_vertex_glyphs + _reposition_vector_glyphs), just for better readability ...

[wmvanvliet]

Sorry, but I disagree that splitting off that functionality in two private method, both of which are only called once, is more readable. _brain.py currently sits at 4859 lines with 65 helper functions. It's fair to say this thing has gotten a bit out of control, hasn't it? This is why I'm not a big fan of adding even more helper functions if not really necessary. In this case, the functions only encapsulate a few lines of code that I don't think hurt much if just left in brain.set_surf.

[payamsash]

Agreed. Done.
Maybe we can check and trim some helper functions later (in another PR).

[wmvanvliet]

Looks like we just need a changelog entry and then it's good to go from my end.

[payamsash]

Thanks @wmvanvliet

[wmvanvliet]

Thanks, @payamsash!