[FieldTrip] Running the ICA method to remove EOG/ECG artifact on Neuromag(*.fif) data

[Andrew R. Dykstra]

Hi Candida,

An alternative to using fieldtrip would be EEGLAB (I believe Fieldtrip
using the code there anyway).  I was able to run ICA on my combined
EEG/Neuromag-122 data using a combination of the MNE MATLAB toolbox
(specifically, fiff_setup_read_raw & fiff_read_raw_segment) and EEGLAB's
import from MATLAB matrix functionality.  More work would be required to
retain the sensor locations and resulting component topographies.  Then,
along the lines of Alex's concerns, you could run the decomposition
separately for each sensor type (Mags/Grads/EEG).  This may be unnecessary
if the runica algorithm handles this properly (assuming you have imported
the sensor types along with the time series data into EEGLAB).  After
subtracting out the artifact components, you could then replace the
appropriate entires of your original data matrix and
use fiff_write_raw_segment (from the nightly build of the MNE suite).

Hope this helps,
Andy

-- 
Andrew R. Dykstra, PhD
Auditory Cognition Lab
Neurologie und Poliklinik
Universitätsklinikum Heidelberg
Im Neuenheimer Feld 400
69120 Heidelberg

"How small the cosmos.  How paltry and puny compared to human consciousness
. . . to a single individual recollection." - Vladimir Nabokov
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20120413/a0ee1431/attachment-0002.html>