draw topomap with a custom ch_type

elvisjade · March 25, 2022, 1:39pm

Is it possible to draw topomap with a custom data?

I am 306channel meg data, then each channel is [-1,1] r value, I want to draw the topomap of this r value, is there any recommended way?
I use this code:

evoke = mne.EvokedArray(data, info)
env_topo = evoke.plot_topomap(ch_type="mag")

But this has to specify a ch_type, and my data is not either eeg or meg, so the plot is a bit strange with the specified ch_type.

env_cor_topo

If you have a question or issue with MNE-Python, please include the following info:

MNE version: 0.24.0
operating system: Ubuntu 20.04

alexrockhill · March 25, 2022, 3:10pm

Hi @elvisjade! Unfortunately, as far as I know, that is not currently a supported use case; in the documentation (mne.Evoked — MNE 1.0.0 documentation) it specifies that the only accepted channel types are in the list for ch_types.

I have had the need to grab the head outline and sensor positions and plot something in my work so I can point you to:

picks, pos, merge_channels, names, ch_type, sphere, clip_origin = \
    mne.viz.topomap._prepare_topomap_plot(evoked, ch_type)
outlines = mne.viz.topomap._make_head_outlines(
    sphere, pos, outlines='head', clip_origin=clip_origin)

This will give you pos and outlines from which you can plot the lines with ax.plot and show an image based on the values at each position with ax.imshow using the interpolation parameter (this is with maplotlib). It’s a bit of an advanced so might be quite tricky to implement though!

agramfort · March 25, 2022, 3:10pm

you can just pass colorbar=False

if the unit is meaningless

Alex

elvisjade · March 31, 2022, 8:21am

Thank you very much for your suggestion, I have tried it as you said.
But it seems that different ch_types do not present the data in the same way.

elvisjade · March 31, 2022, 8:25am

Thank you very much for your suggestion.

I have extracted picks, pos, merge_channels, names, ch_type, sphere, clip_origin ,outlines and other parameters according to your code.

But it is true that these are only the relevant values, and I need to completely redraw the graph with matplotlib, which seems too difficult for me, I don’t have a good foundation in matplotlib.

mscheltienne · March 31, 2022, 2:41pm

You are trying to plot the vector of r (floats) with a shape of (306, ) on a topomap, correct? Did you try using mne.viz.plot_topomap? mne.viz.plot_topomap — MNE 1.0.0 documentation

If so, what was the problem encountered with this one?

elvisjade · March 31, 2022, 2:48pm

Thank you very much for your reply, I have tried this before.

But this seems to be similar to evoke.plot_topomap.

When I use mne.viz.plot_topomap directly myself, I have been unable to get the correct pos parameter.
I refer to this tutorial for the process of getting the pos parameter, but it keeps reporting errors in the middle.
I’m sorry I didn’t log the error in time, but I did try this function. (How to plot topomaps the way EEGLAB does — MNE 1.0.0 documentation sphere-py)

mscheltienne · March 31, 2022, 2:51pm

How about providing pos as:

import mne

info = create_info(306, 1, 'misc')
info.set_montage(whatever_your_montage_is)

Basically the info you provide in EvokedArray, but with all channels set to misc.

elvisjade · April 1, 2022, 2:54am

Thank you very much for your advice, I just tried to do it.
This is my code:
This is my error:

data = mne.io.read_raw_fif("/sss.fif",preload=False)
data.drop_channels(
["EOG001","ECG003","IASX+","IASX-","IASY+","IASY-","IASZ+","IASZ-","IAS_DX","IAS_DY","IAS_X","IAS_Y","IAS_Z","MISC001","STI001","STI002","STI003","STI004","STI005","STI006","STI007","STI008","STI009","STI010","STI011","STI012","STI013","STI014","STI015","STI016","STI101","SYS201","EOG001","ECG003","IASX+","IASX-","IASY+","IASY-","IASZ+","IASZ-","IAS_DX","IAS_DY","IAS_X","IAS_Y","IAS_Z","MISC001","STI001","STI002","STI003","STI004","STI005","STI006","STI007","STI008","STI009","STI010","STI011","STI012","STI013","STI014","STI015","STI016","STI101","SYS201"]
info = data.info
df = pd.read_csv("/r.csv",)
finfo = mne.create_info(info.ch_names,1,'misc')
mne.viz.plot_topomap(df, finfo)

it will report bug:

File “/mne/lib/python3.9/site-packages/mne/io/pick.py”, line 538, in pick_info
raise ValueError(‘No channels match the selection.’)
No channels match the selection.

finfo = mne.create_info(info.ch_names,1,‘eeg’)
mne.viz.plot_topomap(df, finfo)

it will report bug:

File “/mne/lib/python3.9/site-packages/mne/channels/layout.py”, line 690, in _auto_topomap_coords
raise RuntimeError(‘No digitization points found.’)
RuntimeError: No digitization points found.

mscheltienne · April 1, 2022, 8:45am

Yes, you forgot the set_montage() line. you can add: finfo.set_montage(info.get_montage()) before the topomap plot.

elvisjade · April 2, 2022, 2:24am

I tried it as you suggested, but it gives the same error.

finfo = mne.create_info(info.ch_names, 1, “misc”)

finfo = finfo.set_montage(data.get_montage())

mne.viz.plot_topomap(df_r, finfo)

bug:

File “/mne24/lib/python3.9/site-packages/mne/io/pick.py”, line 538, in pick_info

raise ValueError(‘No channels match the selection.’)

ValueError: No channels match the selection.

And I tried this codes:

finfo = mne.create_info(info.ch_names, 1, “eeg”)

finfo = finfo.set_montage(data.get_montage())

mne.viz.plot_topomap(df_r, finfo)

It will report an error like this.

DigMontage is only a subset of info. There are 306 channel positions not present in the DigMontage. The required channels are:

[‘MEG0111’ all meg channel]

Consider using inst.set_channel_types if these are not EEG channels, or use the on_missing parameter if the channel positions are allowed to be unknown in your analyses.

It seems that this idea is also required to match the type of channel and the type of info, and cannot avoid the channel type.

mscheltienne · April 2, 2022, 9:31am

Arf, I guess setting the montage match with the channel types and not the channel names only. My bad… How about:

import mne

data = mne.io.read_raw_fif("/sss.fif",preload=False)
data.drop_channels(
["EOG001","ECG003","IASX+","IASX-","IASY+","IASY-","IASZ+","IASZ-","IAS_DX","IAS_DY","IAS_X","IAS_Y","IAS_Z","MISC001","STI001","STI002","STI003","STI004","STI005","STI006","STI007","STI008","STI009","STI010","STI011","STI012","STI013","STI014","STI015","STI016","STI101","SYS201","EOG001","ECG003","IASX+","IASX-","IASY+","IASY-","IASZ+","IASZ-","IAS_DX","IAS_DY","IAS_X","IAS_Y","IAS_Z","MISC001","STI001","STI002","STI003","STI004","STI005","STI006","STI007","STI008","STI009","STI010","STI011","STI012","STI013","STI014","STI015","STI016","STI101","SYS201"]
# Set all channels to one type, 'misc' or 'eeg', or whatever
data.set_channel_types({ch_name: 'misc' for ch_name in data.ch_names}) 
df = pd.read_csv("/r.csv")
mne.viz.plot_topomap(df, data.info)

elvisjade · April 3, 2022, 1:51pm

Thank you very much for your patience in helping.

I tried following your latest reply.
data.set_channel_types({ch_name: ‘misc’ for ch_name in data.ch_names})
df = pd.read_csv("/r.csv")
mne.viz.plot_topomap(df, data.info)
but still gives an error
File “/mne24/lib/python3.9/site-packages/mne/io/pick.py”, line 538, in pick_info
raise ValueError(‘No channels match the selection.’)
ValueError: No channels match the selection.

I tried
evoke = mne.EvokedArray(df_r, info)
evoke.plot_topomap()
then it will give an error
File “/mne24/lib/python3.9/site-packages/mne/channels/channels.py”, line 102, in _get_ch_type
raise RuntimeError(‘No plottable channel types found’)

RuntimeError: No plottable channel types found

I also tried this idea

data.set_channel_types({ch_name: ‘eeg’ for ch_name in data.ch_names})
df = pd.read_csv("/r.csv")
mne.viz.plot_topomap(df, data.info)

Then an error will be reported:
File “/mne24/lib/python3.9/site-packages/mne/channels/layout.py”, line 724, in _auto_topomap_coords
raise ValueError(‘The following electrodes have overlapping positions,’

ValueError: The following electrodes have overlapping positions, which causes problems during visualization:
MEG0111…all channel

mscheltienne · April 3, 2022, 5:32pm

I think we are at a point where helping/debugging via discourse becomes difficult.
I suggest you join the next office hours on Discord where you can directly ask your question to MNE developers and share your screen. The office hours take place every 2 weeks on Fridays at 17h CET.

alexrockhill · April 4, 2022, 7:46pm

Just to comment and maybe clarify, misc channel types are not supported for plotting in plot_topomap so you’ll have to hack it a bit and pass eeg like you did. It looks like you’re really close, the error is telling you what’s wrong, some of your sensors have the same position. It looks like they are gradiometers and magnetometers which makes sense. I would suggest separating the data out by gradiometers and then plotting those distances only, it would be the same for magnetometers.

elvisjade · April 5, 2022, 2:38am

Thank you very much for your continued patience and help.
I’ll join mne’s Discord event later and screen share to discuss with the developers.
Many thanks.

elvisjade · April 5, 2022, 2:45am

I understand part of what you are saying.
Thank you very much for your suggestion!
But this is too difficult for me, and I may not be able to do this with my current programming skills.

Topic		Replies	Views
Plot topo maps for raw or epoched data, not evoked 💬 Support & Discussions	12	199	September 9, 2023
Plotting topomap using a custom magnetometer layout 💬 Support & Discussions	2	255	November 18, 2022
mne.viz.plot_topomap - align data to head outline 💬 Support & Discussions meg , visualization	7	1273	September 8, 2021
evoked.plot_topomap 💬 Support & Discussions visualization , evoked	2	865	August 5, 2021
Getting topomap data 💬 Support & Discussions	3	133	August 8, 2023

draw topomap with a custom ch_type

Related Topics