Feature request: view electrodes in brain

[bendichter]

We have some new IBL data that uses the new ndx-anatomical-localization to specify CCF coordinates for electrodes.

https://sandbox.dandiarchive.org/dandiset/217678

We can view these electrodes using brainglobe to render a nice 3D visualizations:

import numpy as np

from brainrender import Scene
from brainrender.actors import Points
import pynwb
import lindi

scene = Scene(atlas_name="allen_mouse_100um")

# Load https://api.sandbox.dandiarchive.org/api/assets/e714625e-cf07-4cea-b4b7-504526b3a9a3/download/
f = lindi.LindiH5pyFile.from_hdf5_file("https://api.sandbox.dandiarchive.org/api/assets/e714625e-cf07-4cea-b4b7-504526b3a9a3/download/")
nwb = pynwb.NWBHDF5IO(file=f, mode='r').read()

x = nwb.lab_meta_data['localization'].anatomical_coordinates_tables['AnatomicalCoordinatesTableElectrodesCCFv3']['x'][:]
y = nwb.lab_meta_data['localization'].anatomical_coordinates_tables['AnatomicalCoordinatesTableElectrodesCCFv3']['y'][:]
z = nwb.lab_meta_data['localization'].anatomical_coordinates_tables['AnatomicalCoordinatesTableElectrodesCCFv3']['z'][:]

electrodes = np.c_[y, z, x]

pts = Points(electrodes, colors="red", radius=30, name="electrodes")
scene.add(pts)

scene.render()

(we're still sorting out why we needed to rotate the axes)

Would it be possible to automate this in neurosift?

cc @adamltyson, @h-mayorquin @magland

[adamltyson]

(we're still sorting out why we needed to rotate the axes)

I imagine this is due to two issues:

• All BrainGlobe atlases have the same origin (most anterior, superior, right voxel), regardless of the origin of the original version of the atlas (additional origins will be available for use in V2, sometime in 2026)
• brainrender expects a left handed coordinate system, which other tools may not

[bendichter]

From a recent convo with @adamltyson , they are planning to do a big refactor of the way atlases are stored to move to a multi-resolution Zarr format. This will help us in a few ways:

1. They will have metadata about what atlases correspond to the same image
2. They will have a formal association of different images that correspond to the same coordinate space at different resolutions
3. It should be more efficient to download the image on the fly

[adamltyson]

FYI @IgorTatarnikov