EEG Troubleshooting

EXAMPLE ISSUE OF SOMETHING THAT CAN GO WRONG

PSG Setup Log – Ground Electrode Issue
We encountered challenges during the PSG setup Friday night due to persistent issues with the reference (REF) electrode. The REF impedance remained grey, despite reapplying and repositioning it at least three times. Eventually, we identified that the ground (GND) electrode was affecting all channels, including the reference.Troubleshooting steps taken:

• Reapplied the GND electrode multiple times, ensuring proper skin preparation (cleaning, abrasion, and conductive paste).

• Swapped the electrode to rule out the possibility of dead electrode.

• Tested alternative GND placements:

  • Participant’s hand

  • Area near the collarbone

  • Center of the chest (sternum)

  • Experimenter’s hand

• Asked the participant to remove all metal accessories, including necklaces and bracelets.

• Swapped REF and GND electrode placement

Observations:
The REF electrode turned green and all EEG channels appeared normal in the following conditions:

1. GND applied to the experimenter's hand

2. GND applied to the participant's hand

3. GND not connected or not applied to any person

Additional steps to isolate the issue:

• Restarted the Natus wall-mounted amplifier unit

• Restarted the Natus software on the acquisition computer

• Swapped in a different headbox

• Connected the headbox to the Natus system in a different room

All tests produced the same result, confirming the problem was not due to hardware failure in the amplifier, software, or headbox.

GENERAL GUIDELINES TO DETERMINE CAUSE OF EEG ISSUE:

This is an S tier (i.e. perfect) example of a classic "the EEG is doing something weird and I don't know why" situation. This is such a good example, I asked Shengzi to write it up and we can leverage it as a learning opportunity for the whole group with some general guidelines on how to try to tackle this should you ever find yourself in a similar situation. First, I'll just validate that Shengzi and company did everything I would have done on Friday, so great job there! They also did check-in with me to make sure that they weren't missing anything obvious and they weren't.

So again, using this as a prime example, here is some general guidance if you ever find yourself in a situation where the EEG/PSG is acting similarly weird or weird in a completely different way. The first thing to remember is that for successful EEG recording, the following things must all be connected and functioning properly:

recording software --> computer --> amplifier (hanging on the wall of the bedroom) --> headbox --> electrode caps/electrodes --> participant.

If something like what Shengzi describes is happening, it is time to put on your detective cap and locate where in this system the error is occurring. The following is the order that I would recommend checking things based both on likelihood of it being an issue and how time consuming/difficult it is to do the check:

(1) Double check the basic electrode/cap connections. This includes (a) making sure everything is plugged in correctly, (b) re-cleaning/re-applying electrodes, (c) adding gel/pressure to cap electrodes, (d) swapping out electrodes, where possible. If the trouble persists even after this, you can be pretty sure the issue is not specific to the electrode caps or specific electrodes, and you can move to other potential issues.

recording software --> computer --> amplifier (hanging on the wall of the bedroom) --> headbox --> electrode caps/electrodes --> participant.

(2) From here, the next easiest things to check are the recording software, computer, and amplifier. The hack for doing this most quickly is that if we can access the other bedroom, try plugging the participant into the other HD system and see if the issue still persists! If the issue still occurs when using the other software/computer/amplifier, then you can pretty much rule out that it is something specific to any of these on the original set up. If you don't have access to the other bedroom or if the issue does go away and you need to try to fix it in the original bedroom, you basically turn everything off and back on again: (1) restart the recording software, and if that doesn't work (2) turn the entire computer and the amplifier off and back on again and confirm that all of the cables are plugged in as they should be (e.g., a fairly regular issue is that someone unplugs one of the ethernet cords, either to the computer or the blue Ethernet cord in the room that plugs into the wall). If the issue still persists after restarting all of these, then you can be pretty certain the issue doesn't have anything to do with the recording software, computer, or amplifier.

recording software --> computer --> amplifier (hanging on the wall of the bedroom) --> headbox --> electrode caps/electrodes --> participant.

(3) Next is the headbox. This is the most annoying one if you don't have another one already all set/plugged in, and that is the only reason it is last. If you have another headbox all plugged in and ready to go, you can just try swapping them out and seeing if the issue persists. If you don't, you may need to do another set up, and you'll want to try both terminal ends of the cap electrodes to make sure it doesn't have anything to do with that (depending on the issue -- from the sounds of it that would not have been helpful here since the issue was already diagnosed to be a reference/ground issue). If you swap out the headbox and have tried different combos of the electrode connections and the issue still persists, you can pretty safely rule out the headbox as the primary issue as well.

recording software --> computer --> amplifier (hanging on the wall of the bedroom) --> headbox --> electrode caps/electrodes --> participant ✅

(4) This leaves something specific to the participant or something about the way the electrodes/system and participant are interacting. Once you've hit this point, the questions then become how good can we get the signal to be, and how important to the study is having clean EEG data (see below). Shengzi and the team did everything I would have: removed all metal, considered the hairstyle (sounds like that was unlikely to be the issue here), and trying the electrode on different places. You should also do a sweep to make sure no electronic devices nearby are causing an issue (though in my experience, this is never the problem) and depending on the issue, something called "bridging" could happen where you have too much electroconductive gels. At this point, you're going to take all of the information to the PI of the study, and they will direct you on what to do, but here is the decision tree essentially for our current studies:

STUDY SPECIFIC DECISION INFO: For the K23 study, the PSG data is a secondary variable and staying relatively on time with the study protocol is most important. The participant must be in bed no later than midnight, so I would say once you have exhausted all possible options (which usually includes checking in with me), it's fine to proceed even if we're not getting a great signal. You can still put them to bed and hit record even on a bad signal, bc frequently when it's something at the participant level the electrodes will settle and the signal will improve throughout the night. For the tDCS studies (both the nap and Michael's new one) the EEG data is the primary outcome variable. That means every effort should be absolutely exhausted to try to get it to work, and for the nap study, if we aren't able to get a clean-ish EEG signal the session should be canceled and rescheduled (I'll let Michael decide how he wants to handle the overnight versions -- those will probably be case by case depending on what's going on with the signal).

🤔WHAT I THINK WAS HAPPENING HERE🤔: So given that the team was effectively able to get to this point, recording software --> computer --> amplifier (hanging on the wall of the bedroom) --> headbox --> electrode caps/electrodes --> participant ✅, we can take some solace in that it was most likely an individual issue and not a system wide issue. This is why taking a step-wise approach is so important, ruling out potential issues is key to figuring out what is going on and if there might be a larger issue that needs to get addressed. So I can't say with absolute certainty, but based on the description that Shengzi gave I'm about 80% sure that the reference was becoming "oversaturated" for some reason. Oversaturation basically occurs when the EEG signal is too strong for the recording equipment to be able to represent. The two things that are leading me in this direction are (1) the signal worked when the ground was placed further away (which was a great thing to check) and (2) the reference channel grayed out when it wasn't working (graying out is a very common effect that EEG systems use to denote oversaturation). Why it was happening with this specific participant -- I have no idea 🤷‍♀️. It's possible it could be hair or skin products, but cleaning the sites should take care of this, especially with the different placements. We can rule out jewelry, etc. My best guess is that this person might just carry more static electricity than the average person, which boils down to an individual difference in their skin type/moisture/sweat glands, etc., basically it is causing some level of bridging that is not typically experienced. I asked ChatGPT a few leading questions and it did essentially agree with all my suspicions (but obviously AI will say yes if it might ever be the case -- even if super unlikely -- so informative but take the certainty with a grain of salt).

All this to say that we can't say with absolute certainty, what the issue was, but the crew was able to rule out a lot, and as such I felt confident that there wouldn't be an EEG-issue with the participant's we ran last night, and I was proven correct as they crushed it!