not detecting any actual EEG signals ... ??

edited April 2016 in Headware
Hi,

Sorry not sure if this should go into another discussion bin. Please move it as appropriate.

I don't think my setup is detecting any actual EEG signals. I say that because I've tried it now with nothing actually physically touching my head except the earclip electrodes. The FFTs from that run pretty much look the same as when the whole headset is on.

I'm using the main + daisy to detect 16 channels EEG. Using a Mark III headset that we printed and assembled using the online instructions. Not using any conductive paste, just the spiky electrodes (which are not particularly comfortable, let me tell you!). The FFTs all pretty much just look like a big slope down from DC (0 Hz) with no other curve shape. So ... DC + Gaussian noise?? It seems like they're not detecting anything!

For the BIAS: FpZ electrode is used (this one is not spiky ... but also don't use paste). SRB2 on each board we have tied together with a wire that was soldered to the pins. So one each ear clips plug into one board or the other's SRB2 pin, but they're really linked. Then we have channels as:

Main - 1 PZ, 2 C4, 3 P4, 4 P8, 5 T8, 6 OZ, 7 F4, 8 F8
Daisy - 9 F7, 10 F3, 11 T7, 12 P7, 13 CZ, 14 FZ, 15 C3, 16 P3


When I look at the signals using the OpenBCI GUI, they all seem to more or less look the same in real time. All around 10 uV. Certainly not DC. But they all seem to behave in the same fashion, in real time.

Is there something dumb I'm doing here???

Comments

  • wjcroftwjcroft Mount Shasta, CA
    edited April 2016
    Leka, hi. Have you gone through the steps in the tutorial, and were you able to see your alpha with eyes closed?

    http://docs.openbci.com/tutorials/01-GettingStarted#getting-started-w-openbci-vi-check-out-your-bodys-electrical-signals-4-brain-waves-alpha-with-openbci

    Another test you can try is to connect all the channel pins together with paste or alligator clips or jumper wires. Then connect this to SRB2. That should result in signal approximately 0 microvolts.

    Do the individual channel electrode graphs jump as you tap on them individually?

  • edited April 2016
    @wjcroft , thanks again for your continued responses. You must work for OBCI?? Otherwise, you're incredibly dedicated.

    Anyhow, I'll try the tutorial to see if I can get Alpha waves. However, I suspect (but may well be proven wrong) that the problem is the spiky electrodes. I think they aren't good as EEG sensors. And they're quite painful, particularly after wearing the headset for a while!

    Thus, a question: can the OBCI boards be used just as well with a more "commercial" type of EEG headcap?? I think the caps are only a few thousand, while I assume the real expense of commercial EEG setups are in the racks of electronics and the proprietary software. The OBCI boards themselves seem to do a nice job and I've already invested a lot of time getting them working and writing Python code to interface with them. It'd be a shame to throw it all away...


    Yes, the sensor graphs do respond to physical touching the electrodes. So I know everything is good from the electrode wires to the board to the computer.
  • wjcroftwjcroft Mount Shasta, CA
    I'm an "engineering consultant" listed on the staff page for the latest Kickstarter. As far as I'm aware, most open source projects are volunteer efforts. Hardware open source is bit different in that various overhead costs must be covered.

    If you have some cup electrodes and paste, that is the way to go for your alpha test. The FRI dry combs have shown good results for many, but they do require significant pressure. There are a couple threads on elastic caps and other headset variations,

    http://openbci.com/forum/index.php?p=/discussion/138/buying-dry-active-electrodes

    http://openbci.com/forum/index.php?p=/discussion/327/3d-printed-electrode-holders-for-g-tec-g-gamma-cap

    http://openbci.com/forum/index.php?p=/discussion/247/velcro-saline-sensor-headset

    http://openbci.com/forum/index.php?p=/discussion/191/elastic-head-caps-electro-cap-g-gammacap-etc
  • edited April 2016
    Trying to detect Mu waves, actually. Now I *think* the problem is the filter I was using on the raw sample values (time series) for the electrodes. I took that off -- obviously the FFT blew up closer to DC, but I just zoomed the graph in on the 8-12Hz band and viola! The averaged FFT response for "rest" periods did show some higher spikes than during the "hand motion" periods on C3 and C4.

    PS - appreciate you clarifying your position. Regardless, thanks a lot for the multiple responses you've given me! (as well as others)

    Also appreciate the links. I understand the DIY/"on a budget" spirit that obviously goes hand-in-hand with this community. I guess though, once we get a bit more serious and move past the concept feasibility stage of the research, then we'll invest in some more "commercial grade" EEG hardware. Though the electronics racks are like $30k - $50k !! That's why I like the OBCI boards.
  • wjcroftwjcroft Mount Shasta, CA
  • Well, I don't doubt you. But we need Mu waves. Alpha doesn't do anything for us.
  • wjcroftwjcroft Mount Shasta, CA
    I thought first you were trying to verify the headset and mainboard are working. For that alpha is a sure thing. See Chip's posts for the difficulty of mu detection.

  • edited April 2016
    Reason I thought it was broke is I did a run with the headset off and the FFT results I got looked like everything else I was getting. But now I think that was because of the time response of the filter output. I looked at the raw channel time series and they were as flat as you could get (which makes sense, with the headset off).

    So once I removed the filter from the code and just took the FFT of the raw channel time series, think it's giving me the results I want to see. Maybe.

    I also know that I can tap my finger on any electrode and see the response in the OBCI GUI program. So at least that sense is working.

    But again, thanks a lot for the helpful posts and suggestions!
  • @wjcroft ;  (or anyone ...)

    Is there any reason that if we purchased a 64 ch cap like this: http://www.egi.com/research-division/geodesic-eeg-system-components/geodesic-sensor-nets

    that we wouldn't be able to plug 16 of the cables into the OBCI boards and run the system otherwise as we currently do with the Mark III headset? It looks like it's just a bunch of (passive) saline-sponge electrodes.
  • wjcroftwjcroft Mount Shasta, CA
    Some friends at the Transformative Technology Lab last fall acquired one of the 64 channel EGI systems. Total cost was in the tens of thousands of dollars, I think closer to $50K. Not sure of the cap costs, but since Electro-caps and g.GAMMAcap's are about $250, I'd guess the EGI net would be well over that. It also comes with a very expensive and unique connector that mates with the EGI amp.

    You can make your own saline system with the link I gave above,

    http://openbci.com/forum/index.php?p=/discussion/247/velcro-saline-sensor-headset

    Which might be where you saw the EGI listed on the "commercial systems" tab. Some other commercial systems such as the Pamel bands would be much lower in price. You could also rig up something pretty easily using that idea of Jeremy's, with the g.GAMMAcap and 3D printed holders.

    > It looks like it's just a bunch of (passive) saline-sponge electrodes.

    I have their manual, the electrode holder is a post with a Ag-AgCl electrode and a sponge, correct. The expensive part for any electrode cap is in the wiring, expand-ability and ease of application.

    William


  • @wjcroft

    I believe it, for a whole setup. I think the electronics are what really cost a ton on the commercial setups. But I think the cap is "only" $2k.

    In regards to having a proprietary connector on the end of the cable/wire from the sensor .... let's pretend for a moment that it was just a free wire and I crimped a standard square pin socket, so that I could slide it onto the OBCI board pin.

    Is there any reason it wouldn't work, like with the FRI dry passive metal spike electrodes?
  • wjcroftwjcroft Mount Shasta, CA
    The whole setup is very heavy. The EGI cap is first dipped in a bucket of saline, then applied. Takes 5 to 10 minutes to resettle all the posts by hand for good contact. (Done by the EEG Tech, not you.) Occasionally squirting in more saline as needed. It's not as fast and easy as might first appear from their website. The cap is connected to the amp through an actual physical support arm that extends from the wall, to hold up the weight of the cable and not drag down on the cap.

    My understanding is that the caps can be refurbished at the factory if needed, in which case the original connector would be required to not be cut off. They probably checkout signals in that case with their equipment.

    You're best bet is with some type of g.GAMMAcap adaptor, or the saline velcro. Or the Pamel.

    One other possibility to replace the FRI combs on your Ultracortex with this,

    http://openbci.com/forum/index.php?p=/discussion/comment/3828/#Comment_3828

    Or soften the pointy aspect,

    http://openbci.com/forum/index.php?p=/discussion/comment/3712/#Comment_3712

  • @wjcroft

    There was nothing special about the Geo Desic design that interested me, or saline electrodes for that matter. Ideally what I want in a headset/cap is: 1) 10-10 spatial resolution, 2) comfort and ease of use for the electrodes and 3) high repeatability in getting the electrodes in the correct positions.

    So the g.GAMMAcap might be the ticket.


    But my question still stands: is there any reason that any headcap of passive electrodes can't be used with the OBCI boards?? Obviously the OBCI boards aren't designed to drive fancy active electrodes. But I'm assuming that any headcap can be used like passive electrodes.
  • wjcroftwjcroft Mount Shasta, CA
    Yes, any passives can work with the ADS1299 in OpenBCI. However... Some conductive materials work better than others, obviously.

    The new .6 ohm-cm filament I mention might be a game changer. Conor ordered a spool for testing. But the lab has their hands full with other priorities. Potentially, future 3D printed combs using the .6 ohm-cm material could be swapped in place of the FRI combs. These printed combs could be less pointy and more adaptable to skull curvature.

  • @wjcroft , that sounds like a potentially great improvement for the next 3D printed Mark XX headset design. Especially if it could have 10-10 spatial resolution.

    In looking at the g.tec product line, it looks like the only passive electrode they have for the g.GAMMAcap is the g.LADYbirdPASSIVE which is described as a passive ring and "sintered Ag/AgCl ring (for DC recordings)"

    Two questions, if you know the answers (no worries if not, I'm going to contact g.tec):

    - do you know why they put "(for DC recordings" in the description? Does that mean the passive ring is not suitable for EEG signals?
    - I assume for a ring like this, similar to the OBCI gold cup electrodes, you have to use some type of conductive gel or solution to "fill in" the physical gap between the scalp and the conductor -- do you put the cap on first and then inject the gel/solution into the electrode?


    Thanks again!
  • wjcroftwjcroft Mount Shasta, CA
    edited April 2016
    Check with Cortec Solutions in the U.S., they are reps for g.tec here,

    http://cortechsolutions.com/

    When I emailed with them last year, here's what I was quoted for the g.GAMMAcap,

    Mr Croft,

    Each size of cap is $230 a piece a set of small, medium, and large caps is $685.  I can send a formal quote if you would like.

    Best

    Nicholas Anderson, PhD
    g.tec Sales and Product Manager

    Cortech Solutions, Inc.
    1409 Audubon Blvd., Unit B1
    Wilmington, NC 28403

    At that time it was not necessary to buy the cap with electrodes. You can retrofit any electrodes into the holes the cap provides. These holes are also labeled with each site location. I'd suggest trying various dry passive electrode combs in the holes using snapon connectors to bridge to the cap outer surface.

    Sintered Ag-AgCl electrodes are especially stable for FbEEG full band EEG recordings that include the DC offset, slowly moving baseline. They are the BEST type of silver chloride electrodes available and quite expensive. As well as fragile, they are recommended to be cleaned with distilled water.

    The g.GAMMAcap with g.tec electrodes is used in two types of configuration: dry or gel injected.

    image

  • I'll contact Cortech and see what they have to say.

    Given the discomfort of the Mark III headset electrodes, I'm cautious about using "comb"-style dry electrodes. I'd be quite fine squirting or injecting gel, however that is done, into a flat/cup electrode.

    Thanks once more, for the wealth of info you've shared.
  • Hi
    I have very strange issuse. My open bci cap and board worked when i put it on my head.
    But when i used it for someone else it didnt work.and even didnt show any signals.
    What can i do?
    Please help me
  • wjcroftwjcroft Mount Shasta, CA
    Faeze, hi.

    Are you adjusting the electrode holders for each user, by screwing in or out so that proper contact is made? If you see "didn't show any signals", then likely you have an issue with the reference ear clip. Since it is the common electrode for each channel. Try swapping it with the Bias ear clip, in case there is some damage.

    Also, the ear lobe on your other users may be coated with oil or lotion that is disturbing conduction. Use alcohol to clean the ear lobes before you clip on.

    Regards,

    William

Sign In or Register to comment.