I have a question about the sensor layout map for EGI hydrocel sensor nets.
I have been studying the two figures attached - one is the 128 channel map provided by EGI, the other is the sensor positions determined by using the commands mne.channels.read_montage('GSN-HydroCel-128') and mne.viz.plot_sensors.
While there are similarities, some definite differences exist - notably, 81 has shifted from the occipital midline to close to the vertex. Has this been addressed before? And if so, what is the advice you recommend?
I think we may have resolved this? The updated code is
file = '/Users/delfin/Documents/RPKPILOT/RAW/PILOT002_GONOGO.set'
### read channel names and get raw data
mon = mne.channels.read_montage('GSN-HydroCel-128')
raw = mne.io.read_raw_eeglab(file, preload=True)
raw.set_montage(mon) ## !!
### look at channels
mne.viz.plot_sensors(raw.info, show_names=True)
raw.plot_sensors(show_names=True)
raw.plot_sensors('3d')
Sorry for any confusion! Hopefully we are correct and this is the right layout!
The plot you showed from MNE in your original post is the layout map for
the much-older GSN 200 2.1 nets. For the past many years (decade?) EGI has
sold the HCGSN nets, which have a slightly different layout (as shown in
the "official" EGI layout you attached). As you noticed, a telltale sign is
whether the bottom-most centre electrode in the layout map is 81 (new) or
82 (old).
I haven't tried importing EGI data into MNE yet, so I can't speak directly
to your issue. However I do know that EEGLAB for years (and possibly still)
by default uses the old layout when using their EGI importer. The EEGLAB
folks did, however, post a layout map file for the new HCGSN configuration,
which you can find at https://sccn.ucsd.edu/bugzilla/show_bug.cgi?id=1643
It should be in a standard format that hopefully MNE recognizes, or could
be adapted pretty easily. Getting EGI data into MNE has been on my to-do
list but I haven't tried it yet!
I believe that the problem here was actually the montage reader. The
channel names were read incorrectly in mne versions 0.12 and older, causing
duplicate channel names. So updating mne to the latest version should fix
the problem. The workflow looks correct.
Many thanks to you both! It seems possible that in my first attempt there was some issue with 3d to 2d conversion - the back row of electrodes seemed folded 'under' and ended up towards the center of the projection?
Anyway, I'm moving forward with the current layout - thank you again!
