How can I change the style of my topographic maps?

đź’¬ Support & Discussions
1

MNE 1.3.0
MacOS BigSur 11.6.7

My topographic maps are looking a little weird, for example when visualizing ICA components with the following code:

raw = mne.io.read_raw_eeglab('data/participant/sub-01_task-rsvp_continuous.set', preload=True)
filt_raw = raw.copy().filter(l_freq=1., h_freq=None)
ica = ICA(n_components=62, max_iter='auto', random_state=97)
ica.fit(filt_raw)

They look like this:

image
image1902Ă—1138 118 KB

I’m not familiar with this type of topographic map and don’t know how to interpret it. I would like to generate maps of the type in the tutorials which I’m familiar with, which look like this:

image
image1538Ă—1022 142 KB

Does anyone know how to change this? Thanks in advance :slight_smile:

2

Hello,

That’s some very artistic maps :slight_smile:
Could you share a complete code snippet all the way to the plot and the data file that is causing these plots?
I suspect a scaling issue on the electrode location (montage) which then messes up the interpolation on the maps.

Mathieu

3

Hi! Thanks for the quick response!

Here is the minimal code to recreate the weirdo maps:

import mne
from mne.preprocessing import (ICA)

raw = mne.io.read_raw_eeglab('data/participant/sub-01_task-rsvp_continuous.set', preload=True)
filt_raw = raw.copy().filter(l_freq=1., h_freq=None)
ica = ICA(n_components=62, max_iter='auto', random_state=97)
ica.fit(filt_raw)
ica.plot_components()

How can I also upload the data? I can tell you where it’s from, which is here:
https://openneuro.org/datasets/ds003825/versions/1.2.0
If you navigate to derivatives/eeglab then you have a collection of .set and .fdt files. My example uses the first one.

Cheers!

4

So, it’s a bit difficult to reproduce because those files tend to crash my interpreter for some reason… especially when trying to plot the raw time-series. But anyway, it is a scaling issue of the montage (the electrode location).

When loading the file, you are greeted with:

RuntimeWarning: Estimated head radius (0.2 cm) is below the 3rd percentile for infant head size. Check if the montage_units argument is correct (the default is “mm”, but your channel positions may be in different units).

And indeed, here is the loaded montage compared to the default head:

raw.get_montage().plot()

image
image660Ă—648 10 KB

Instead, you should provide the argument montage_units to the reader to correctly load the montage.

from mne.io import read_raw_eeglab

raw = read_raw_eeglab(file, montage_units="dm")
raw.get_montage().plot()

image
image659Ă—569 18.1 KB

Note that I’m not convinced that "dm" is the correct unit, and since:

  • the montage seems to be a template and not an actual measured montage (which are usually not that symmetric)
  • the electrode seems to respect the standard 10/05

You could simply replace the loaded montage with a template provided by MNE.

from mne.io import read_raw_eeglab

raw = read_raw_eeglab(file, montage_units="dm")
raw.set_montage("standard_1005")
raw.get_montage().plot()

image
image660Ă—648 19.1 KB

That should fix the topographic maps produce from those files.

Mathieu

1 Like
5

Thanks! This has fixed it!