Hello community,

I am currently working on a project to identify any correlation between resting EEG pattern (relative power of each frequency bandwidth: delta, theta, alpha, beta) and propofol requirement to anesthetize a patient. I have collected individual's raw EEG data using Masimo Sedline, which only provides four channel data (Fp1, Fp2, F7, F8).

I have tried to follow several guidelines that EEGLAB and other experts provided online, but I am not sure if I am on the right track.

Here are the steps that I took to pre-process my EEG data:

1. baseline removal

2. High-pass & low-pass filter (exclude < 0.1 Hz, > 50 Hz data)

3. visual inspection & remove any artifacts (eye movement, etc)

According to EEGLab guideline, re-reference data using reference channel data or CAR (common average reference) is a necessary step for preprocessing but since my EEG data do not include a reference channel, I've decided not to re-reference the data. Is it acceptable to skip this step?

Another question that I had was about independent component analysis. To my knowledge ICA is a great tool to remove artifacts related to eye movement, muscle twitching, etc, if there are enough channels and EOG channel data. Unfortunately I have only four channels to work with and no EOG channel data. I went ahead and just applied ICA on my data set, but it seems to me that ICA excludes out too much of EEG signal along with artifacts. What is the most appropriate way to remove artifacts when you have a limited number of EEG channel?

Also, am I missing any steps in preprocessing EEG data?

Thank you in advance.