Unknown(and unexcpected) noise

[henrik.eijsink]

Hello,

I have designed an experiment where subjects exucutes(and imagines) handmovements while recording their EEG. The goal is to use machine learning to make a Brain computer interface that eneables robot hand control with EEG. In order to achieve the presise timestamps from stimuli that i wanted for the machine learning algorithms, I implemented the data collection through LabStreamingLayer, using their built-in tools for Biosemi, and Labrecorder.

The data has a very high amplitude noise with a spike shape(that makes it hard to remove). I (stupidly) did not notice this before i collected all the planned data, so im planning to collect new data. I hope to get some input if someone have had a similar problem, both on the origin of the noise, as well as ideas of eliminating the problem.

I am attaching two plots:

EEG_signal.png (22.87 KiB) Viewed 10759 times

EEG_signal - a sample from one of the elektrodes, (time - microvolt). to clearify: the noise is apparent on all 64 electrodes used, so this is just a sample from one of them

EEG_signal_zoom.png (42.96 KiB) Viewed 10759 times

EEG_signal_zoom - zoomed version of EEG_signal, It seems to me that the correct eeg data excists in the signal(when you zoom in enough). Or am i completely off here, and this is just random noise from the equipment?

[Coen]

Blue light blinking ? Common Mode was not in range, confirm by checking the trigger/status channel in the BDF file. For solutions, see https://www.biosemi.com/faq/check_electrodes.htm . Data measured with blinking blue light is meaningless because power to the active electrodes is cut off.

Best regards, Coen (BioSemi)