Preprocessing - Filtering - Epoching

[prosper...@gmail.com]

Hi Mike,

I am a novice to both EEG and signal processing. I thank you for the videos and the book, they are really helpful. I have the following doubts, which may be very elementary – Excuse me for that.

I am doing an experiment with Biosemi 64 channel EEG and analyzing the data with EEGLAB. The data was not acquired through the Biosemi software rather they are acquired through ‘Motion Monitor’ where the trigger is signal is combined with actual data. Moreover, each trial data is acquired as a single file (*.EXP). I have added them together sequentially and converted to a BDF file to be analyzed using EEGLAB. (Please suggest me if there is any other better approach)

i)                    Is this acceptable? Will it adversely affect the signal while filtering? (If I Epoch 1 second before the trigger and 2.5 Seconds after the trigger). - (Please find the attached image)

ii)                   I find there are many filtering tools available in EEGLAB and I need to choose one from it. For example for removing line noise (Cleanline, PrepLine… etc). What would you suggest for me to read to get the knowledge to choose the best filter?

 

Thanks and Regards,

Prosper

[Mike X Cohen]

Hi Prosper. The epoching is fine. It will introduce edge effects during filtering, but that's no different from the effects in time-frequency analysis. I don't know about those two line noise filters you mentioned. I guess that any 50-Hz notch filter should do the trick. Or you can apply a low-pass filter at 45 Hz (or 55 Hz if you are in US), and then you also wouldn't need to worry about harmonics.

Mike

--
You received this message because you are subscribed to the Google Groups "AnalyzingNeuralTimeSeriesData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to analyzingneuraltimeseriesdata+unsub...@googlegroups.com.
Visit this group at https://groups.google.com/group/analyzingneuraltimeseriesdata.
For more options, visit https://groups.google.com/d/optout.

--

Mike X Cohen, PhD
New online courses: mikexcohen.com

[prosper...@gmail.com]

Thank you very much for the speedy reply!

[prosper...@gmail.com]

Hi Mike,

I have four queries regarding epoching, ICA and ERP and I have listed them down with appropriate screenshots in the attached document. It would be of great help if you could have a look and pass on your suggestions.

Thanks in advance,

Prosper

[Mike X Cohen]

Hi Prosper. Here are some answers:

1) Yes, try ICA. I guess it will identify the EMG artifact. I also discuss these kinds of artifacts in the book in the section on baseline normalization. If the artifact is equal in the trial and pre-trial periods, then it shouldn't have such a negative influence on the results.

2) The ICA maps do look a bit noisy, but I'd be hesitant to do too much to the data just to get nicer IC maps, unless you are planning on analyzing the components instead of the channel data. It looks like the EMG artifacts are heterogeneous and are represented by many components.

3) That "band" is one trial with a potential shift. You could check out that one trial in the channel data to see if there is an artifact, but somethings things look big and scary in an IC but are hardly visible in the channel data.

4) The ERP contains a lot of high-frequency activity (possibly noise). You could try low-pass filtering at, e.g., 20 Hz.

Hope that helps,

Mike

--

You received this message because you are subscribed to the Google Groups "AnalyzingNeuralTimeSeriesData" group.
To unsubscribe from this group and stop receiving emails from it, send an email to analyzingneuraltimeseriesdata+unsub...@googlegroups.com.
Visit this group at https://groups.google.com/group/analyzingneuraltimeseriesdata.
For more options, visit https://groups.google.com/d/optout.

[prosper...@gmail.com]

Thank you so much for your prompt suggestions!