We are seeing the 16 Hz skin conductance signal contaminating EEG. It appears when the impedance of the skin conductance electrodes is higher than about 40K-ohm at the time of application. If the impedance falls below this value, the 16 Hz signal will disappear from EEG. This interference appearing in EEG is actually spikes corresponding to the leading and trailing edges of the 16Hz square wave, indicative of capacitive coupling into the EEG leads or circuitry. We had our amplifier checked out, it's ok. SC electrodes/leads are ok. Blue CM and green GSR lights are on when this happens, and SC recordings look ok.
1. Is it normal for 40K impedance and above in skin conductance electrodes to cause coupling of SC signal into EEG? If not, what is the highest value of SC impedance tolerable before seeing any coupling? (We are measuring this impedance using 10Hz 3uA, prior to connecting the electrodes to BioSemi).
2. We often have subjects with >40K SC electrode impedance on fingers, even folks with desk jobs. We have used various electrode pastes, gels, and have not seen much variation. Sometime abrasion helps, sometimes not, and it is undesirable to abraid for SC anyway. At this point we are not able to routinely acquire EEG and SC simultaneously without seeing the 16 Hz contamination.
Skin Conductance 16Hz contaminating EEG
-
Coen
- Site Admin
- Posts: 1125
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
Interference between the GSR and EEG measurements has not been reported earlier. The described interference is not normal
We have not had an amplifier sent to us for checking with the above complaint, so an hardware problem is still possible (who exactly checked your AD-box ?).
In theory, there can be some interference between GSR and EEG when there is a large difference in impedance between the two GSR electrodes. In a situation where one GSR electrode has a very high impedance (hundreds of kOhm) and the other has a very low impedance (several kOhm), current sourced by one electrode may not be sinked completely by the other. The difference flows to the DRL electrode. This causes a small 16 Hz Common Mode signal, and may cause small differential voltages along the current path from GSR electrode to DRL electrode. The Common Mode interference should disappear after referencing. The differential 16 Hz interference can be decreased by relocating the DRL electrode to a place between the GSR electrodes and the EEG electrodes. For example, with the GSR electrodes on the left hand, a DRL location on the left mastoid or left ear is in theory better than the standard location near the center of the headcap.
As said, we have not heart of situations where changing the DRL location was actually needed to prevent 16 Hz interference, but please let me know whether relocating the DRL electrode does indeed influence the amount of 16 Hz interference in your particular measurement situation.
Finally, it should be noted that the 16 Hz is locked to the sample rate. So, off line removal of the 16 Hz and its harmonics from the EEG signal can be performed with very high accuracy.
Best regards, Coen (BioSemi)
We have not had an amplifier sent to us for checking with the above complaint, so an hardware problem is still possible (who exactly checked your AD-box ?).
In theory, there can be some interference between GSR and EEG when there is a large difference in impedance between the two GSR electrodes. In a situation where one GSR electrode has a very high impedance (hundreds of kOhm) and the other has a very low impedance (several kOhm), current sourced by one electrode may not be sinked completely by the other. The difference flows to the DRL electrode. This causes a small 16 Hz Common Mode signal, and may cause small differential voltages along the current path from GSR electrode to DRL electrode. The Common Mode interference should disappear after referencing. The differential 16 Hz interference can be decreased by relocating the DRL electrode to a place between the GSR electrodes and the EEG electrodes. For example, with the GSR electrodes on the left hand, a DRL location on the left mastoid or left ear is in theory better than the standard location near the center of the headcap.
As said, we have not heart of situations where changing the DRL location was actually needed to prevent 16 Hz interference, but please let me know whether relocating the DRL electrode does indeed influence the amount of 16 Hz interference in your particular measurement situation.
Finally, it should be noted that the 16 Hz is locked to the sample rate. So, off line removal of the 16 Hz and its harmonics from the EEG signal can be performed with very high accuracy.
Best regards, Coen (BioSemi)
Thanks for your response Coen. We had our system checked out by Lloyd for an unrelated issue just before the SC prob started. When we first encountered the SC 16 Hz leaking into EEG, we checked the SC electrodes/lead wires and found a wire was broken inside the insulation. This was repaired, and it seemed like all issues were resolved. However, our experimenters have since periodically encountered the 16 Hz, affectionately dubbed "barbed wire", for no apparent reason other than possible high electrode impedance. However, electrodes and lead wires check out ok.
We tried an experiment yesterday. We applied the SC electrodes to two adjacent fingers, on the pads, and observed nice SC readings and clean EEG. We then moved one electrode to the topside of the finger, and left the other electrode on the pad of the other finger. We then observed the same SC reading, but this time also observed the 16 Hz barbed wire in EEG. We did not try moving the DRL as you suggested however, since we need to have a tight routine for getting experiments done without adjusting electrode locations on the fly.
It seems like the SC barbed wire leakage is very sensitive to electrode position. Would this translate into an issue regarding variations in subjects' finger anatomy, e.g., placing electrodes on the balls of fingers with enough difference in length could cause SC leakage?
We tried an experiment yesterday. We applied the SC electrodes to two adjacent fingers, on the pads, and observed nice SC readings and clean EEG. We then moved one electrode to the topside of the finger, and left the other electrode on the pad of the other finger. We then observed the same SC reading, but this time also observed the 16 Hz barbed wire in EEG. We did not try moving the DRL as you suggested however, since we need to have a tight routine for getting experiments done without adjusting electrode locations on the fly.
It seems like the SC barbed wire leakage is very sensitive to electrode position. Would this translate into an issue regarding variations in subjects' finger anatomy, e.g., placing electrodes on the balls of fingers with enough difference in length could cause SC leakage?
-
Coen
- Site Admin
- Posts: 1125
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
The location of the DRL does not affect EEG data.
GSR excitation current leakage can be caused by a large difference in impedance between the two GSR electrodes. So, it is not unexpected that the best results are achieved with the electrodes on similar pieces of skin (e.g. both on the finger pads), whereas worse results are achieved when the skin differs (e.g. one electrode on the finger pad and the other on the finger top). The skin-electrode impedance is usually much large then the impedance across the body. So, it is unlikely that finger anatomy should make a significant difference.
Best regards, Coen (BioSemi)
GSR excitation current leakage can be caused by a large difference in impedance between the two GSR electrodes. So, it is not unexpected that the best results are achieved with the electrodes on similar pieces of skin (e.g. both on the finger pads), whereas worse results are achieved when the skin differs (e.g. one electrode on the finger pad and the other on the finger top). The skin-electrode impedance is usually much large then the impedance across the body. So, it is unlikely that finger anatomy should make a significant difference.
Best regards, Coen (BioSemi)
Today we ran a subject with 128 EEG electrodes and skin conductance measurement. Both GSR lights were on. 16 Hz artifact appeared in electrode B2 and to a lesser degree in some of the other closest EEG electrodes surrounding DRL (located at its standard location on the scalp). The SC electrodes were carefully applied to the balls of fingers, on a subject with soft hands. Unfortunately my experimenters did not get an independent impedance measurement for the SC electrodes, but it was unlikely that one electrode was grossly mismatched vs. the other electrode. We checked out the SC electrodes, cables and plug; all ok. Is there any diagnostic we could run on the SC inputs to check equipment functioning? (We already tried the NaCl bucket test to verify SC is working).
-
Coen
- Site Admin
- Posts: 1125
- Joined: Fri Mar 26, 2004 7:00 pm
- Location: Amsterdam, Netherlands
- Contact:
Test with the GSR electrodes close together and a deliberate gel bridge between the electrodes (or even the electrodes touching each other). Does this setup still generates crosstalk to the EEG electrodes ? If yes, measure the (16 Hz) current in the GSR leads with the GSR electrodes and CMS/DRL in water. The currents should not differ more than a few percent. In case of a too large current difference, sent the AD-box to us for repair.
If the currents are within spec, then change the DRL location as described earlier. Use (purchase) a CMS/DRL set with a pin CMS and a flat DRL. The DRL location affects interference. Changing the DRL location does not affect the referenced EEG data.
Best regards, Coen (BioSemi)
If the currents are within spec, then change the DRL location as described earlier. Use (purchase) a CMS/DRL set with a pin CMS and a flat DRL. The DRL location affects interference. Changing the DRL location does not affect the referenced EEG data.
Best regards, Coen (BioSemi)
Hi Coen,
Today we measured current in the skin conductance leads, in water (with CMS/DRL electrodes also). The current in the leads matches. So, we want to try your suggestion of using a CMS/DRL set with a pin on CMS and flat on DRL, shifting DRL location to the ear, which would let our experimenters keep running with minimal disruption to their protocol. However, I do not see this combination electrode set offered. Is this a stock item?
Today we measured current in the skin conductance leads, in water (with CMS/DRL electrodes also). The current in the leads matches. So, we want to try your suggestion of using a CMS/DRL set with a pin on CMS and flat on DRL, shifting DRL location to the ear, which would let our experimenters keep running with minimal disruption to their protocol. However, I do not see this combination electrode set offered. Is this a stock item?