Eye artifact removal using ICA

External Email - Use Caution

Hi all,

I am trying to remove all eye artifacts from an EEG recording with an EOG
channel using ICA (following the tutorial on MNE). I applied a bandpass
filter (1 and 40Hz) to the signal and removed the default average
reference. Then I fit ICA as follows:

picks = mne.pick_types(raw.info, eeg= True, eog= True, stim= False,
exclude= 'bads')
ica= ICA(n_components=25, method='fastica', random_state=1)
ica.fit(raw, picks=picks, decim=3)

n_max_eog=2
eog_epochs=create_eog_epochs(raw, ch_name='EXG1')
eog_inds, scores= ica.find_bads_eog(eog_epochs)

As an output, this says only 2 EOG events detected (when there are many
more blinks). I can add the reject criteria with reject=dict(eog=threshold)
but I do not know what threshold to choose. I am relatively new to MNE.

I am attaching the link to the file, if it helps.
Giannis_72518_Active.bdf
<https://drive.google.com/file/d/1Jyh_BBPGrnghIcNo1bDYD4yxevu_B4UI/view?usp=drive_web>
Thanks for all your help! Really appreciate it.

-Neelesh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20180727/e1ee7747/attachment.html

1 Like

External Email - Use Caution

hi Neelesh,

you look at ICA components and one looks like the EOG artifacts
just exclude it manually by setting ica.exclude attribute.

HTH
Alex