Noise introduced into EEG readings when others walk around the patient
tcreagan
Arizona
Hello there,
I'm an engineer who has been working on an ADS1299-based EEG device that loosely references the OpenBCI Cyton board -- many of the similarities between the two boards are in the bias filter network and channel filter networks. I've gotten it to stream across USB and Bluetooth, and I generally get clean EEG signals both ways.
However, something I haven't been able to fix yet has to do with noise. While a patient is testing with the EEG cap, if someone else walks around them -- even up to 6ft away -- huge noise spikes will be introduced into the data. This happens both when the design is used on a wired connection or when it is streaming wirelessly. My guess is that this is because of large static potential differences between the patient and the person walking around them, but I'm not entirely sure. I will be doing some testing in different rooms around the office to see if it's related to the floor material, it's possible the carpet here is what's creating so much static electricity.
Have any of you experienced this with the OpenBCI boards, and if so, how did you mitigate it?
More information about the system:
- Uses a ADS1299 for data conversion
- Uses the bias amplifier sourced from all N channels
- Uses the SRB2 reference electrode internally connected to all P side channels.
- Uses a linked ear referential montage with the bias electrode on a mastoid.
- Electrode wires are kept to a short length, but are unshielded.
- Electrodes are dry comb electrodes.
Regards,
Reagan
Comments
UPDATE: Just completed the test with the EEG system in a different room. The same spikes were still present when one of my coworkers walks around the room. Spikes were even seen when he tapped his foot on the floor.
From your description it sounds like the elastic cap is applying downward pressure on dry comb electrodes. Is that correct? Are these the same combs we use?
Generally these dry combs require much greater pressure than typical caps provide. See the Ultracortex headset or headband units in the Shop. Headband allows arbitrary pressure to be applied by tightening the velcro straps.
EEG noise rejection is directly related to skin impedance. The higher the skin impedance, the worse the noise characteristics. Have you measured your impedances?
If you have access to a gel or saline type cap, that is what I would try next. Also be aware that in most clinical settings, the EEG is recorded in a quiet room with only the subject and practitioner. There are not other people walking around the area.
A standard low cost gel cap used in QEEG / neurofeedback is the Electro-cap. They also make the gel.
Thanks for the reply William. I'm working on some code to measure impedances as we speak -- after reading through this forum and some literature, it seems like high impedances and subsequently poor SNR may be a large part of the issue. Once I measure them, I'll post it here.
As for the dry comb electrodes, they're these ones from Neurospec.
https://shop.neurospec.com/flexible-dry-eeg-electrode-tips-brush-10-pack?___store=en_US&___from_store=x_default
My client is using a pretty standard elastic cap. He was able to get good results from an OpenBCI Cyton system he was testing with, but I believe he was using shielded wires. I'm going to inspect his cap tomorrow and see what I can find out about it. He's pretty focused on quick setup times, so gel or saline is going to be a last resort option -- I want to see if we can get better results from dry electrodes first.
Although dry electrodes can be used as 'passive' electrodes with normal EEG amplifiers -- increasingly manufacturers are pairing the dry electrodes with 'active' amplification behind each electrode. As the signal strength of dry/passive is noticeably less than either wet/passive or dry/active.
https://shop.openbci.com/products/thinkpulse-active-electrode-kit?variant=37113738985630