WIP, ENH: Add sensors to Brain

[GuillaumeFavelier]

This PR follows #8382 (comment) and implements add_sensors to plot MEG and EEG sensors.

For now, I move code and experiment. Refactoring is necessary to avoid duplication between plot_alignment() and add_sensors.

This is still work in progress.

[GuillaumeFavelier]

I play with it using:

import os.path as op

import numpy as np
import nibabel as nib

import mne

data_path = mne.datasets.sample.data_path()
subjects_dir = op.join(data_path, 'subjects')
raw_fname = op.join(data_path, 'MEG', 'sample', 'sample_audvis_raw.fif')
trans_fname = op.join(data_path, 'MEG', 'sample',
                      'sample_audvis_raw-trans.fif')
raw = mne.io.read_raw_fif(raw_fname)
trans = mne.read_trans(trans_fname)
src = mne.read_source_spaces(op.join(subjects_dir, 'sample', 'bem',
                                     'sample-oct-6-src.fif'))

# Load the T1 file and change the header information to the correct units
t1w = nib.load(op.join(data_path, 'subjects', 'sample', 'mri', 'T1.mgz'))
t1w = nib.Nifti1Image(t1w.dataobj, t1w.affine)
t1w.header['xyzt_units'] = np.array(10, dtype='uint8')
t1_mgh = nib.MGHImage(t1w.dataobj, t1w.affine)

Brain = mne.viz.get_brain_class()
brain = Brain(subject_id="sample", hemi="both", surf="white", alpha=0.7,
              units="m")
brain.add_sensors(info=raw.info, trans=trans, coord_frame='meg')

mne.viz.plot_alignment(raw.info, trans=trans, subject='sample',
                       subjects_dir=subjects_dir, surfaces='head-dense',
                       show_axes=True, dig=True, eeg=[], meg='sensors',
                       coord_frame='meg', mri_fiducials='estimated',
                       fig=brain._renderer.scene())

I had to change units to make the brain scale correctly but I still think some transformation is missing.

[larsoner]

brain.add_sensors should not have a coord_frame argument. It's always MRI because brain is always in MRI. This is probably our transform problem

[GuillaumeFavelier]

Working on EEG now.

Where to fetch the sensors' color?

[larsoner]

Same place as plot_alignment, it's some defaults.py / DEFAULTS entry

https://github.com/mne-tools/mne-python/blob/main/mne/viz/_3d.py#L1063-L1065

Eventually we'll want to add a parameter to allow controlling the sensor color on a channel-name basis, but not in the first PR (need to sort out the API). But if you can at least consider this use case it might help guide the initial design.

[GuillaumeFavelier]

Another naive question, should I add support for projected EEG also? On what surface?

Also %(trans)s is not automagically detected in the docstring.

[larsoner]

Another naive question, should I add support for projected EEG also? On what surface?

Not in the first PR, we can always add it later.

Also %(trans)s is not automagically detected in the docstring.

You either need @verbose decorator if there is a verbose=None kwarg or @fill_doc otherwise

[GuillaumeFavelier]

Based on plot_ecog, I got something like the following in this early version:

This slider should fit somewhere nice when time_viewer=True, the default pyvista colormap is used and the range of scalar values is calculated on the fly. Glyph refactor is also required for this.

[alexrockhill]

This seems confusing, could you just transform to head if in "meg", then transform to mri if in originally in "meg" (now "head" because it was just transformed) or "head"? That seems like it might be more straightforward

[alexrockhill]

Ah I guess you might need them multiple times so that might not work