Discrepancy between Allen CCFv3 and IBL atlas

[Jiachen24]

Hi all,

I am trying to remap channel_locations (output from IBL ephys GUI) to another brain atlas, kim_mouse_25um through brainglobe api. I found the result is quite different because of the annotations so I tried to remap to allen_mouse_25um through brainglobe first. I converted IBL coordination to Allen CCF coordination using brain_atlas.xyz2ccf, and I reorganized the sequence from mlapdv to apdvml. Surprisingly, the result is still a little bit different (first picture from IBL gui, second one after conversion). I also attached a demo code to show the coordination conversion. If the IBL ephys GUI is using Allen CCFv3 and the origin is same as documented ( "bregma": [5739,5400,332]), I don't know why I am seeing this difference. Please advise! Thank you so much!

demo showing coordination from xyz to ccf

Conversion code

[oliche]

Hi,
Thanks for the feedback !
Happy to look into this, but could you please add a minimal working example of the problem that we can copy/paste ?

[Jiachen24]

Sure! I attached the channel_locations.json from IBL ephys GUI and the one (allen25_channel_locations) after conversion. Below is the code I used. Thanks!

from pathlib import Path
import json
import numpy as np
from brainglobe_atlasapi import BrainGlobeAtlas


atlas = BrainGlobeAtlas("allen_mouse_25um")


bregma_offset= np.array([5400, 332, 5739], dtype=float)


in_file  = Path("F:/channel_locations.json")
out_file = Path("F:/allen25_channel_location.json")

with in_file.open() as f:
    ch_dict = json.load(f)

for ch_name, ch_info in ch_dict.items():
    if not all(k in ch_info for k in ("x", "y", "z")):
        continue
    rel_xyz = np.array([-ch_info["y"], -ch_info["z"], ch_info["x"]], dtype=float)
    abs_xyz = rel_xyz + bregma_offset

    label_id = int(
        atlas.structure_from_coords(tuple(abs_xyz), microns=True)
    )
    label_nm = (
        atlas.structures[label_id]["acronym"] if label_id != 0 else "void"
    )
    
    ch_info.update({
        "x": float(abs_xyz[0]),
        "y": float(abs_xyz[1]),
        "z": float(abs_xyz[2]),
        "brain_region_id": label_id,
        "brain_region":   label_nm,
    })

with out_file.open("w") as f:
    json.dump(ch_dict, f, indent=2)

print(f"Saved   {out_file.resolve()}")

channel_locations.json

allen25_channel_locations.json

[Jiachen24]

Hi @oliche ! Please let me know if you need more information. Thank you!