[payamsash]

Reference issue (if any)

Following #14130

What does this implement/fix?

Now subcortical meshes are drawn as surfaces next to the cortex, and their colors follow the time slider and the colormap limits.

Additional information

Opus 5.5 helped implementing, I reviewed, did some modification and added an example.

[larsoner]

@ttaiv do you want to look?

[ttaiv]

@larsoner yes I do! I tried plotting data on the cerebellar mesh produced with CMB. Using the Brain class like in the new example worked nicely like this:

src = mne.setup_subcortical_source_space(
    subject="sample",
    surface=cerebellum_mesh_file,
    subjects_dir=subjects_dir,
    keep_largest_component=False,
)
data = np.random.rand(len(src[0]["vertno"]))

brain = mne.viz.Brain(
    "sample", subjects_dir=subjects_dir, hemi="both", surf="white", **brain_kwargs
)
brain.add_data(
    data,
    hemi="vol",
    vertices=src[0]["vertno"],
    src=src,
    fmin=0,
    fmax=1,
    colormap="viridis",
)
brain.show_view(azimuth=150, elevation=90, distance=400, focalpoint=(0, -10, 10))

But when I tried running the CMB example script and plot the source estimate like this

# Plotting MixedSourceEstimate
estimate.plot(
    subject=subject,
    subjects_dir=subjects_dir,
    src=fwd["src"],
    surface="white",
    hemi="lh",
)

the result looks weird:

[payamsash]

My guess is that the brain is in MRI coordinate, but the cerebellum is in head coordinate ... becasue you put fwd["src"] ???

[ttaiv]

You were correct. Adding coordinate transformation before plotting fixed the location:

# Plotting MixedSourceEstimate
src_plot = fwd["src"].copy()
mri_head_t = mne.read_trans(trans)
for s in src_plot:
    mne.transform_surface_to(s, "mri", mri_head_t)

estimate.plot(
    subject=subject,
    subjects_dir=subjects_dir,
    src=src_plot,
    surface="white",
    hemi="both",
)

But the estimate still floats in the air because the cerebellar surface is invisible. I think this is because the subcortical mesh has no opaque base layer, so its color comes only from the data layer. Wherever the data is transparent, the mesh disappears. Passing transparent=False and colormap="coolwarm" fixed this for me, because together they give a colormap with no transparent values.

[larsoner]

You were correct. Adding coordinate transformation before plotting fixed the location:

Good to know. This is a bug in the code.... you shouldn't have to do that transformation yourself, our plotting function should checking and converting to the destination coordinate frame if needed

the cerebellar surface is invisible. I think this is because the subcortical mesh has no opaque base layer, so its color comes only from the data layer.

We should fix this, too. Our cortical ones have a base gray layer that sits underneath everything. I think all surfaces (cortical and subcortical) should behave like this. (This actually has an alpha of its own that you can set separately.)