Different apparent channel locations between standard_1020 and biosemi64 montages

Support & Discussions
1

Hi everyone,

I have a question about the channel locations in different built-in montages.

I noticed that channels with the same standard labels can appear in quite different positions when using different MNE built-in montages. For example, when I plot standard_1020 and biosemi64, some channels with the same labels, such as frontal or central electrodes, appear to have noticeably different locations.

In particular, in the standard_1020 topomap layout, several channels seem to be plotted outside or very close to the head outline, which makes the layout difficult for me to interpret. I am not sure whether this is expected behaviour due to differences in the coordinate systems/projection methods, or whether I am misunderstanding how these montage layouts should be used.

Here is the code I used:

import mne

mont = mne.channels.make_standard_montage("standard_1020")
fig = mont.plot(
    kind="topomap",
    show_names=True,
    sphere="auto"
)

mont = mne.channels.make_standard_montage("biosemi64")
fig = mont.plot(
    kind="topomap",
    show_names=True,
    sphere="auto"
)

image
image1272Ă—613 70.2 KB

My questions are:

Why can channels with the same labels have different apparent locations across these built-in montages?
Why do some channels in standard_1020 appear outside the head outline in the topomap plot?

Any clarification would be greatly appreciated.

Best,
Jianyang

2

The standard_1020 defines locations for many electrodes, also for those not on the top of the head, but rather on the sides. These electrodes are located at the height of, and sometimes even lower than, the ears. If the head were a sphere, they would be on the underside of it. By default, such electrode locations are shown in “skirt” mode, meaning we place them outside the head circle. To mentally visualize their location, imagine wrapping them around the sphere like one would wrap a present.

With biomsemi_64, I would say the default plot is wrong, as it draws the head circle such that even electrodes such as P9 and P10 are still drawn “on top” of the head-sphere instead of being at the underside of it.

3

I get a slightly different output for "biosemi64":

import matplotlib.pyplot as plt

import mne

mont = mne.channels.make_standard_montage("standard_1020")

fig, axes = plt.subplots(1, 2, figsize=(18, 8))

fig = mont.plot(
    kind="topomap",
    show_names=True,
    sphere="auto",
    axes=axes[0],
)
axes[0].set_title("standard_1020")

mont = mne.channels.make_standard_montage("biosemi64")
fig = mont.plot(
    kind="topomap",
    show_names=True,
    sphere="auto",
    axes=axes[1],
)
axes[1].set_title("biosemi64")

plt.show()

Figure1
Figure11799Ă—799 38.1 KB

P9/P10 are actually outside the circle. You can influence which electrodes are drawn on the circle with the sphere argument, but leaving them at "auto" produces correct (comparable) electrode locations for me.

However, I am more concerned about the weird “curved” arrangement of horizontally aligned electrodes in the "standard_1020" montage (e.g., T7-C5-C3-C1-Cz-C2-C4-C6-T8). IMO, these should be straight lines like in the "biosemi64" montage, which is why I prefer that over the former.

4

PS: If you want to create montage plots with theoretically derived 10-05 locations check out the eeg_positions package (and this example). The 2D plot created with this package matches the "biosemi64" plot.

5

Hello, this is intentional behavior and known to cause confusion. The montages whose names start with “standard” were fitted on the fsaverage head and are suitable for source estimation, but look oddly unfamiliar when working at the sensor level. You may want to use one that doesn’t start with “standard” and take a look at this section of the documentation to achieve a more familiar representation:

I’d suggest to only apply a “standard” montage immediately before performing source localization without individually measured sensor locations. Otherwise stick with the biosemi or similar templates.

Richard

6

PPS: Maybe there’s something off with the coordinate origin, because when setting sphere=[0, -0.025, 0, 0.1] at least the electrodes seem to be aligned correctly (but then all electrodes are shifted towards the front).

Figure1
Figure11799Ă—799 41.2 KB

7

Thanks @richard! There is already an issue, let’s continue the discussion there and maybe try to fix this! Default standard_1020 montage contains realistic, not template positions · Issue #7141 · mne-tools/mne-python · GitHub