Using frontal EEG channels as a proxy for - EOG. Whats happens to captured frontal activity?

💬 Support & Discussions
,
1

Dear MNE community,

I just piloted my experiment on monday with a 32 channel EEG cap from brainvision and I am now exploring the data. Note: I am a bit familiar with EEG, but have not worked with actual data before.

I have encountered one problem. After fitting and inspecting an ICA

ica.plot_components();

it becomes very obvious that there are artifacts in the data due to eye movements/blinks.
However we failed to set up a EOG electrode. According to the documentation

If you don’t have an EOG channel, find_bads_eog has a ch_name parameter that you can use as a proxy for EOG. You can use a single channel, or create a bipolar reference from frontal EEG sensors and use that as virtual EOG channel. This carries a risk however: you must hope that the frontal EEG channels only reflect EOG and not brain dynamics in the prefrontal cortex (or you must not care about those prefrontal signals).

it is possible to use a frontal electrode as a proxy. However, we actually hypothesize increased activity in frontal areas. If I were to use Fp1 or F8 as a proxy, would I not lose information about neural activtiy in frontal areas? We used a 32 channel EEG from brainvision for the pilot.

I am a bit lost at this point. What would you guys recommend? One possibility would obviously be to re-run the experiment with EOG electrodes.

Furthermore it is apparently recommended to set up a highpass and lowpass frequency before the experiment. When loading the data

raw = mne.io.read_raw_brainvision('Tesr.vhdr')

and inspecting the object

raw.info

I notice that the sampling frequency is set to 1000hz. The Highpass is set to 0.02 and lowpass is set to 1000.
Is this setup already kind of false?

Please apologize those novice questions. I am really just starting to figure all things out.

Looking forward to your replies! Please let me know if I can provide more information.

Best

2

I usually use ICA to correct occular artifacts (Clemens Brunner - Removing eye activity from EEG via ICA), which does not require any EOG channels. In general, you will have a mixture of brain and ocular activity in both (frontal) EEG as well as EOG channels – there is no difference except for their locations (and possibly the way you mount them on the scalp). No method for removing ocular artifacts is perfect, so you will remove some amount of brain activity. However, if you’d rather wanted to use a regression-based approach (Clemens Brunner - Removing eye activity from EEG via regression), having two dedicated bipolar EOG channels is strongly recommended.

3

Hi,

Thanks a lot! I was not fully aware of this. I will thus continue using an ICA to remove the artifacts.

Do you have any remarks on the low and highpass settings? Unfortunately I did not know that you should set up the lowpass filter to prevent aliasing before actually doing the EEG. The researcher that helped me with the setup did also not mention anything in that direction.

Will this be an issue when analyzing the data?

Best

Edit:

Just to clarify: I obviously filtered the data before fitting the ICA. It just seems like the anti aliasing filter is missing here.

4

Are you sure the amplifier does not set an appropriate antialiasing filter automatically as mentioned in this article? What does the corresponding entry in your .vhdr file say?