selecting epochs to use in compute_proj_eog

Mailing List Archive (read-only)
1

Dear list,

I'm currently trying to correct EOG artifacts on EEG data. To do so, I'm
using "compute_proj_eog". This function first detects where EOG
artifacts are, by using "find_eog_events". However, in some datasets,
"find_eog_events" clearly picks segments that are not EOG. Using
"create_eog_epochs" and plotting the resulting epochs, one can select
which ones? are really EOG artifacts and save them in the epoch object.
But? how can we indicate to "compute_proj_eog" that it should only use
these ones for later correction??? Indeed, "compute_proj_eog" re-run
"find_eog_events", and hence ends-up with all the original segments,
including the ones not corresponding to EOG artifacts.

Is there a way to tell "compute_proj_eog" which segment it should take
into account to build the projector?

??? Thank you in advance for your help!!

??? ??? Best,

??? ??? ??? Boris.

2

Hi Boris,

You can add the projector manually using something like this:

ssp_blink_proj = mne.compute_proj_epochs(blink_epochs, n_grad=0, n_mag=0,
                                         n_eeg=2, desc_prefix=None,
verbose=None)
raw = raw.add_proj(ssp_blink_proj)

Here "blink_epochs" is your manually curated epochs object, and "raw" is
the raw data you want to apply the projector to. n_grad and n_mag are set
to zero since it's EEG-only data.

-- dan

Daniel McCloy
http://dan.mccloy.info/
Postdoctoral Research Associate
Institute for Learning and Brain Sciences
University of Washington

3

Hi Dan,

Le 25/09/2017 ? 20:51, Dan McCloy a ?crit?:

Hi Boris,

You can add the projector manually using something like this:

ssp_blink_proj = mne.compute_proj_epochs(blink_epochs, n_grad=0, n_mag=0,
??? n_eeg=2, desc_prefix=None,
verbose=None)
raw = raw.add_proj(ssp_blink_proj)

Thanks very much!! It works perfectly !!!

??? B.