3-byte raw data buffers - best way to deal with eeg data interference

✉️ Mailing List Archive (read-only)
1

Hi again,

Every problem I solve just seems to bring me a new one in its stead.
I'm trying to use mne_compute_raw_inverse
Thanks to Matti & Alex, I managed to solve the trigger line problem.
But now I'm getting the error:
Reading data Cannot process buffers of data type: 3

This seems to refer to data buffers in the raw EEG data (at least
that's the only reference to buffers I can find in the manual). I'm
not actually interested in EEG data. We recorded from a few reference
electrodes so we could use the data to remove eye blinks, but
otherwise we don't want to use the EEG channels at all. Is there some
way to tell mne_compute_raw_inverse to use MEG data only and ignore
any EEG buffers (just adding --megonly didn't work)? Or is there
something I have to do convert the 3-byte buffers to 4? The EEG
channels were recorded by the CTF system, and bundled with the MEG
data.

Thanks to anyone who has an idea of how to fix this.

Linnaea

2

Hi Linnaea,

Data type 3 does not mean three bytes but rather 32-bit integers. The
mne_compute_raw_inverse in the new MNE 2.6 handles 32-bit data. If you
wan to continue using the MNE version 2.5.2 you can convert the CTF
data to floats (and maybe downsample):

mne_process_raw --raw <orig> --projoff --save <converted file>

This also does the default 40-Hz lowpass. You can include --lowpass, --
highpass, --filteroff, and --decim options as desired to control the
output.

- Matti