EEG channel positions

Hello,
I'm trying to interpolate the bad channels but it throws the following
error since chs['loc'][:3] for a few channels is [0,0,0]:
ValueError: 'Could not extract channel positions for 4 channels'

I was wondering what could be the reason for this?

And whether or not it would make sense to use positions from raw.info['dig']
(since I have measured all the EEG channel positions when digitizing)
together with set_channel_positions to set the missing positions?
If so, where can I get more info about the structures inside raw.info['dig']
to make sure which points correspond to the EEG channels
(e.g. don't know what does "kind" value mean there.)

Many thanks,
Rezvan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20160419/f7b336f8/attachment.html

Hi Rezvan,

can you figure out what these channels are. My suspicion would be that some
channels are labeled EEG but are in fact EMG, or ECG or some other
physiology readout. These often have no channel positions, hence 0. If this
is the case your solution would be to exclude those channels.

Denis

Hi Denis,
Thanks a lot. I thought so too but no, EEG071-EEG074 have locations [0,0,0]
saved for them.
I thought it might have something to do with the numbering jump from 60 to
65 but ECG and EOG channels don't have any positions either.
Probably I'll have to use the digitization points but couldn't find a
description of parameters inside raw.info['dig']? Are they in the same
order as I have digitized during the acquisition? does it include fiducial
and HPI locations too?

Best,
Rezvan

So difficult to guess. But if they have 0, 0, 0 it does not look like a
real position. Usually you don't have electrodes in the origin with
noninvasive techniques :wink:
Could you share a small file or just the info with us? See
mne.io.write_info. And is this the Neuromag EEG? Any chance you can figure
out the exact montage of your EEG?

Denis