dry vs wet electrodes, quantitative noise & signal quality comparisons

George MartinGeorge Martin
edited November 2015 in Electrodes
I recently checkout out your store and the new kickstarter page.  Early on when a question was posed on the forum regarding dry vs wet electrodes the response was that dry electrodes were not up to the quality signal that wet electrodes provide.  I see that the Ultra Cortex uses dry electrodes.  Has there been any comparisons done regarding signal quality and amount of noise?

Thanks

George

Comments

  • wjcroftwjcroft Mount Shasta, CA
    George, hi.

    I think the question / response ("not up to the signal quality") you may have referred to (on this thread?)


    Might have been an early response before the dry combs were getting much use. However at this time the FRI Ag-AgCl dry electrodes have gotten a good deal of field verification. They might be producing slightly less amplitude than a wet connection, but still very usable.

    Perhaps Karl @KZurn or Conor @conor_obci have done a side by side comparison of signal quality. With Karl's velcro bands this would be easy to do, using side by side combs, one dry and one injected with a bit of gel through the center hole. Then comparing the two channels for amplitude and noise characteristics.

    Cognionics is using similar Ag-AgCl combs in their headsets, although with active amplification in some cases. There have been a number of papers published comparing dry vs wet electrode signal quality. All generally conclude that with good dry sensors, the signals are comparable and usable.


    William



  • George MartinGeorge Martin
    Thanks for the info.

    George
  • manneokokomanneokoko
    Hello,
    can somebody who own a OpenBCI confirm that the signal is as good as with wet electrodes?
    Ive had a conversation with one of the authorities of neurofeedback about OpenBCI, his opinion is that its impossible to make a headset with good dry electrodes in the pricerange of OpenBCI and he concludes that it is not suitable for neurofeedback.
  • wjcroftwjcroft Mount Shasta, CA
    Manneokoko, hi.

    I'm going to mention Karl @KZurn here of FRI, who is distributing the Ag-AgCl dry electrodes and cables. He has been selling many of them to research and educational communities even outside of OpenBCI.

    Did you look at some of the papers cited above in this thread? They have shown quantitatively that dry electrodes can be very compatible with the signal quality received from wet electrodes. Can you please mention the person you refer to who is questioning the usability of dry electrodes in neurofeedback?

    Are you and he aware of the many consumer grade headsets now using dry electrodes successfully for neurofeedback, such as Muse, Emotiv, Neurosky, Versus, etc. The Versus headset, using 5 dry comb sensors, has Dr. Leslie Sherlin as founder and CSO. He has been past president of ISNR, the primary neurofeedback research and publishing organization.

    I'm also mentioning Joel @biomurph and Conor @conor_obci who have had much experience using the Ultracortex with Neuromore, which is an application devoted primarily to neurofeedback and BCI.

    [Quoting forum usernames with the at sign causes them to receive a notification email.]

    Best regards,

    William

  • wjcroftwjcroft Mount Shasta, CA
    There is a type of neurofeedback done where the recorded EEG signals from the client (band, amplitude, coherence, phase, etc. at each 10-20 site) -- are compared to the signals recorded in a normative statistical database. The technical term for that is QEEG, quantitative EEG. All measurements in the database are generally made with an Electro-Cap and conductive gel.

    So yes, if you are talking about QEEG type of measurements, it's always recommended to do those with the same type of electrodes / caps as were used in the collection of the database.

    William

  • George MartinGeorge Martin
    William,

    I checked out the Versus web site, it looks interesting.  Do you know if they use active or passive dry electrodes?

    George
  • wjcroftwjcroft Mount Shasta, CA
    edited December 2015
    George, hi. I believe with all the conductive rubber comb dry sensors, that active amplification is needed. There are several groups getting good results however from Ag-AgCl passive dry sensors. If the passive dry sensors are combined with a high impedance amp (>= 1 gigaohm), and relatively short cables, it appears to be a workable alternative over the more expensive active sensors.

  • George MartinGeorge Martin
    william,

    thanks.  I am not a tech minded guy but I assume the open BCI boards a fall into that category.

    George
  • manneokokomanneokoko
    Hi,
    The person Im refering to is Peter van Deusen of Brain trainer. He has been doing neurofeedback for a long time, and is sceptic to a dry electrodes for Neurofeedback. Since he has seen many "breakthroughs" in dry electrodes over the years, and only the very expensive ones has worked good. He means that they might work good on bare skin but mostly not on areas with hair. He also thinks that it might be difficult with correct electrodeplacement using ultracortex.
    I contacted him because I wanted to ask if he thinks that OpenBCI would be compatible with Brain trainer and BioExplorer. It seems like it wont happen. Too bad, I think the Brain Trainer system looks very good, with automatic generation of training plans from the assessment
    William, you seem like you know alot of neurofeedback; can I email you and maybe get some useful tips on starting Neurofeedback?
  • wjcroftwjcroft Mount Shasta, CA
    edited December 2015
    Pete's Brain-Trainer system is great, I've used it and benefited enormously. Probably the best platform for home trainers like yourself. I've also participated on his Yahoo forum and exchanged emails with Pete. An amazing mentor and indeed neurofeedback guru.

    All Pete's designs are based on Bioexplorer, which is also a wonderful package. Unfortunately the Bioexplorer dev, Larry Janow, is very reluctant to communicate via email or add new drivers. Several OpenBCI owners have tried asking, with no replies. Perhaps if he were offered a payment, don't know how much, I'll try sending him an email regarding that. So up to this point the likelihood of having an OpenBCI driver in Bioexplorer has been slim. Bioexplorer does have an OpenEEG driver, but it runs at 256 sps, not the 250 sps we currently use. (Ganglion will run at 256 sps.) So even emulating OpenEEG, which has been discussed in this forum on other threads -- looks problematic at the moment.

    Pete's system requires flexibility in accessing all 10-20 sites. Most consumer EEG headsets do not offer this. So I can understand his reluctance.

    I'll send you an email, your best route at this point is to just use Bioexplorer with Pete's designs and one of the Pocket-Neurobics amplifiers.

  • manneokokomanneokoko
    Yeah, Pete seems like a great guy and i really would like to start with the Brain Trainer system. What i was hoping for was to be able to use OpenBCI with braintrainer, to avoid the hassle of using wet electrodes.

    I was also askimg the OpenBCI devs on their kickstarter page about the possability for BioExplorer support, and just now i got an answer from them "Our current tech, the 32bit Board does connect with BioEra and BioExplorer", i guess they mean the open EEG emulation?

    I also asked them about the quality of the dry electrodes. Here is their answer
    "We have great results with the dry electrodes. Our current tech can read the impedance of the electrode skin connection, and it is lower than what we get with gold-cup electrodes and paste."
  • wjcroftwjcroft Mount Shasta, CA
    edited December 2015
    Yes, we do connect to BioEra in full 8 or 16 channel mode at 24 bit resolution 250 sps (125 sps for the 16ch). The Bioexplorer OpenEEG emulation experiment was useful, in that it demonstrated that OpenEEG emulation could be done. But again, the sample rate difference, 256 sps vs our 250 sps, throws a significant monkey wrench into the feasibility of using it with QEEG like technology, such as what Pete does with his protocols.


    You could use one of the velcro based setups to avoid paste. You could even use FRI dry combs and give that a try.



  • George MartinGeorge Martin
    I have used a similar product created by 3M  I liked it for it's ease of use but it degraded over time.  Whag I do now is bu a packet of velcroe that has a sticky back.  The package contains both the hook and loop strips.  I cut the strips to the proper length, peel off the paper backing and carefully place them together.   I have multiple lengths so I can use on most any sized head.  If you want to get a clear sense of where the sites are you can take a piece of elastic at 10cm  Mark it off in 1 cm segments.  Stretch from inion to .  Each mark will br 10% from the last giving you a good indicator of the 10-20 sites.  Do the same side to side.

    I use small pockets of the artificial chamois (shamwow being a name brand) hand sewn to just fit the electrode inside. 

    I also wonder of you were worried about the dry electrodes if it would be possible to insert a small piece of sponge or shamwow into the space between the combs.

    On the assessment side there are a couple possibilities that could be done with BIoEra Pro, and perhaps NeuroMore. 
    Paul Swingle"s quick Q is a nice tool for basic assessment.  It uses info from a small set of sites.  An Excel document created in BIoEra could be used to do the assessment.  You have o do some of the math yourself but it is not difficult.   That way you could use your open BCI instead of having to purchase another amp. 

    Regarding Pete's concern about placement,  A person I took some training from many years ago said that consistency in placement is more important than attempting exact placement.
    The OpenBCI headset woud get y9ou very close to exact site with great consistency of placement.

    If your are doing research or working off a normative database this may not be the best method, but I have found for neurofeedback it works very well

    George


  • George MartinGeorge Martin
    If there are users interested in using BioEra Pro for Neurofeedback with the Quick Q I would gladly create and share a design that would gather and process data.

    If enough people (10 or more) are interested In BioEra Pro I can make a bulk order for BioEra and pass the savings on to end users.

    George
  • manneokokomanneokoko
    William, will the OpenEEG emulation work right away for the forthcoming ganglion? Does it support 4 ch?
    Then I'd think that Ganglion already works in Bioexplorer..?

    Very intresting with the velcro and saline .. As a second choice after ultracortex. Will ultracortex IV work with other amplifiers (q-wiz)?

    George,
    Can you tell a bit more about how the Quick Q / clinical Q works? Does the system include protocols and designs for training different areas in the same way as brain trainer? What happens after the assesment is done, how does the practitioner turn the assesment into a training plan? Have you used the system on yourself and gained benefits?
  • George MartinGeorge Martin
    The quick/clinical Q analyzes amplitude data at and between 5 sites.  It will give you a good idea of what sites and bands need training.  It will not autogenerate a training plan like the tlc.  Pete based a lot of the rations and amplitude analysis on the quick q. 

    I used the assessment for several years with good results.  I would rather look at data myself, use a good clinical interview and Pete's subjective assessment to generate a more personalized plan for each client.

    I still often use Pete's older 2 channel assessment which is very similar to the quick q with the addition of coherence values.  
    I also use a measure of cortical arousal called brain-rate which gives a good idea of problems in arousal between eyes open,closed and task.

    I think the ultra cortex will work with any amplifier.  You just would not be able to place it into the headset.


    George

  • manneokokomanneokoko
    Very nice if Ultracortex works with Q-wiz. But I read somewhere that dry sensors need to have very short wires to give good signal. If using Ultracortex with Q wiz, its not possible to place the amp on the headset, which means that long cables would be needed. Will this be a problem?
  • wjcroftwjcroft Mount Shasta, CA
    > But I read somewhere that dry sensors need to have very short wires to give good signal.

    No, that is not true. FRI sells the dry combs with conventional touchproof cables and gets good results. By 'short' I think cables in the range of a few feet to a few inches would qualify. What you don't want are cables reaching down to the floor.

    I'm pretty sure you could also rig up a velcro based holder 'strap' such that the Q-Wiz could attach to your shoulder or upper arm. Cables then again would be short enough.

  • KZurnKZurn

    We have a discounted on our AgCl Ear Clip electrodes from
    $19.95 to $14.97.  It is good until April
    30.  See this link  http://fri-fl-shop.com/product/td-430-silver-disc-electrode-ear-clip/

    Karl Zurn

  • ratlabguyratlabguy
    The latest generation, and future upcoming dry electrodes are much better than what used to be available.
    The good news is that, based on what I've seen in the research world, it;'s a good bet that within 2 years dry electrodes will be readily available.
    The bad news is that the primary challenge with dry electrodes is NOT so much finding the right materials and designs, but that they are highly susceptible to motion artifacts and capatitive discharch artifacts etc from movement.
    In other words.they work great as long as you are motionless. But once you start moving around the SNR can quickly get way worse than standard ones.
  • HirokiHiroki
    I'm looking for research that compare the correlation of the signals simultaneously obtained by a OpenBCI and another research/medical grade wet EEG.

    I found a paper that compare the amplifiers and shows the signal correlation is very good.

    What I want to check is the correlation of signals obtained both of a dry OpenBCI and another wet-type EEG at the same time.
    ( Subject should wear both of an OpenBCI and a wet EEG )

    Do anyone know such a paper?

    with best regards,
  • ratlabguyratlabguy

    Hiroki, you need to please clarify what it is that you really want to know.

    Do you mean that you want to see the correlation between dry and wet electrodes, using the same amplifier?

    You really should never compare the data from one amplifier (e.g. OpenBCI) using a dry electrode to a different amp using wet electrodes. This is meaningless b/c you have two different variables.

    Although, generally speaking, these days the electronics inside of amplifiers are not very different between manufacturers, especially if both happen to be based on the TI ADS1299 chipset (which is very common today).

    Here is an example of a recent study comparing a COTS wet to a COT metal pin to a new polymer elastomer electrode. All are using the same amp (g.tec). It isn't OpenBCI, but the important result is that you can see how the electrode material really matters. And more importantly... a carbon-silicone electrode can work just as well, and even better than metal pins, which are uncomfortable...

    https://ieeexplore.ieee.org/abstract/document/8513428



  • HirokiHiroki
    What I want to know is, yes, the correlation of the signals from the dry OpenBCI and another wet EEG device.

    Now I'm doing experiment using wet EEG. My experiment is to measure an ERP.
    If the OpenBCI can capture the same ERP, I want to replace the wet EEG with OpenBCI.
    ( "same" means statistically significant )

    So I want to have an evidence that the same ERP can be captured using dry OpenBCI and wet EEG.
  • ratlabguyratlabguy

    It's not that simple.

    Overall, in a static condition, the signal characteristics of COTS dry electrodes are fairly similar to wet gel based electrodes. This is assuming you use the same amplifier for both.

    But this statement is not universal. Where dry electrodes are dramatically different is in (1) the frequency response curve, particularly in the very low frequencies, due to the different material properties oat the capacitive junction of the skin, and (2) when any motion occurs. Almost all conventional dry electrodes have a HUGE and long capacitive discharge and rebound when there is motion.

    so the point of all this is, generally, it will be close to the same unless you have any motion at all or have frequency-dependent effects. It really depends on the scenario and what signal you want to get - so you cannot just make a blanket prediction.

    It is definitely possible to collect basic ERPs w/ an OpenBCI and dry electrodes. We have done it in our lab. We have done is w/ OpenBCI, Emotiv, G.Tec. Enobio, Quasar, ABM, Cognionics..... just don't move, lol.

  • ratlabguyratlabguy

    By the way, overall, straight up correlation is a terrible statistic to tell you if two electrodes are similar.

    The problem is that large amplitude, low-frequency trends that are in common will greatly dominate the statistic. It is better to use something more complicated, either with binning and averaging or better yet, use RMS difference or other things.

Sign In or Register to comment.

© Vanilla Keystone Theme 2026