using SSP projections for eye blink correction and mne_process raw

✉️ Mailing List Archive (read-only)
1

Hi All,

Below is a snippet of code that I use: for finding EOG and ECG event and correcting for them:

2

hi David,

My question is regarding the settings : " --projmagrej 4000 --projgradrej
3000 --projeegrej 500" what do these setting lmits mean:
calling the help I get the following:

it's like reject param in Python. Peak to peak difference rejection values
by channel type.

--projmagrej 4000

means igore the epoch/event during which a magnetometer not marked
as bad as a peak to peak difference bigger than 4000 fT

does this mean rejecting an an eyeblink event above the listed value? If so
they seem low. Given that these levels are subjecto the recording
environment/ noise, Are my limits (--projmagrej 4000 --projgradrej 3000
--projeegrej 500) good ?

yes it sounds reasonable. You need to try on your data and see with
mne_analyze if the SSP vector computed does the job.

btw you can compute these SSP vectors from Python.

see:

mne.preprocessing.compute_proj_eog

and the command line tool which usage could be like this:

mne compute_proj_eog -i sample_audvis_raw.fif --l-freq 1 --h-freq 35
--rej-grad 3000 --rej-mag 4000 --rej-eeg 500

HTH
Alex

3

Thanks
Alex this helps