Codecov Report

Merging #6797 into master will increase coverage by 0.04%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #6797      +/-   ##
==========================================
+ Coverage   89.59%   89.63%   +0.04%     
==========================================
  Files         422      422              
  Lines       76406    76486      +80     
  Branches    12490    12500      +10     
==========================================
+ Hits        68458    68561     +103     
+ Misses       5138     5122      -16     
+ Partials     2810     2803       -7

I've been trying to find what where this 'GND' and 'REF' locations in the doc and googling but I couldn't find it.

@agramfort suggested to plot the postitions and if they are truly the same as 'EEG000' they will appear behind the ear. Well they are not:

import os.path as op
from mne.channels import (Montage, read_montage, read_dig_montage,
                          get_builtin_montages, DigMontage,
                          read_dig_egi, read_dig_captrack, read_dig_fif,
                          make_standard_montage)
from mne import create_info
import mne
import matplotlib.pyplot as plt
from mne.viz import plot_alignment, set_3d_view
from mne.datasets import fetch_fsaverage

plt.switch_backend('Qt5Agg')

data_path = mne.datasets.testing.data_path(download=False)
subjects_dir = op.dirname(fetch_fsaverage())

montage = read_dig_captrack(
    fname=op.join(data_path, 'montage', 'captrak_coords.bvct')
)
info = create_info(ch_names=['GND', 'REF'], sfreq=1, ch_types='eeg',
                   montage=montage)

fig = plot_alignment(
    info=info,
    show_axes=True, dig=True, surfaces='head', trans=None,
    subject='fsaverage', subjects_dir=subjects_dir,
)
set_3d_view(figure=fig, azimuth=135, elevation=80)

What I would do is what this PR is doing, add those channels and positions to the montage.

it's weird to have both GND and REF. One should correspond to EEG000 with neuromag/fif conventions. If it's not called EEG000 then set_montage should be modified to understand that GND or REF are the reference.

@jona-sassenhagen @mmagnuski any idea what GND or REF correspond to here and why there are two things? Should we ignore GND and just take REF as the EEG000 electrode?

This is why:

https://pressrelease.brainproducts.com/referencing/#11

In the end we can ignore the ground and use REF as EEG000

ok this changes one of the dig['ident'] with respect of the fif reference file. Not really sure why.

Any clues?

ok this changes one of the dig['ident'] with respect of the fif reference file. Not really sure why.

It is possible that in this instance we are actually doing something better/more correct than the C code, for example if it does not have an ident=0 and/or does not set chs['loc'][3:6] and we do.

Ok then, I'll ignore info['dig'] and make sure that chs loc is propper.

@larsoner I did it for one of the two. The other one is more annoying.

Please merge if green and happy

Thanks @massich @agramfort