[Eeglablist] 16 Hz artifact with lower harmonics in data
Makoto Miyakoshi
mmiyakoshi at ucsd.edu
Mon Apr 11 11:00:06 PDT 2022
Dear Malte and Bene,
Thank you Malte for the detailed explanation. I appreciate you took time to
detail it for us.
ZapLine and ZapLine plus uses a spatial filter. You want to make a good
number of electrodes. If you have only two electrodes, it is probably
severely suboptimal. Use other methods instead, such as a notch filter, a
spectral interpolation (Fieldtrip supports it I believe), or I personally
recommend CleanLine (debugged version). CleanLine works a univariate manner
and it takes advantage of the difference in the time constants of amplitude
envelopes between the environmental noise (slower changes) and the
endogenic (faster changes).
Please keep us posted Malte. 'The whole world wants to know where is
the 16-Hz signal from?'
Makoto
On Sun, Apr 10, 2022 at 10:46 AM Malte Anders <malteanders at gmail.com> wrote:
> Benedikt you are a lifesaver. Zapline plus workd for me when I enter "17
> Hz" in the frequency field. It doesn't really work when I enter 16 Hz. I
> think (?) this supports the hypothesis that this artifact is indeed at
> 16.66 Hz and may thus be coming from Deutsche Bahn. Unfortunately, zapline
> does not work when I enter "16.66" in the frequency field. In any case, see
> these images - those were very very bad recordings which were previously
> unusable. They aren't 100% perfect now, but it's as good as it's gonna get.
>
> Before zapline:
> https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_d23dP1W&d=DwIFaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=TQ0LZx6aEl5Fe8TH8maW5njPnvkyX4dlPZzCMeWmp49Kd6BG7JFCRo3LXvMcnwB-&s=7ys4Mm6dzbNs5sPtpVa-u2VM1y2lT45mv5_1C8DqXS0&e=
> After zapline:
> https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_4Wz8w5s&d=DwIFaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=TQ0LZx6aEl5Fe8TH8maW5njPnvkyX4dlPZzCMeWmp49Kd6BG7JFCRo3LXvMcnwB-&s=Fu-jSM1uR1jTJDK7dCDAa6E6jV_k_yGm9wAiHDWdMZ8&e=
> Raw EEG before/after:
> https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_Brm8ddt&d=DwIFaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=TQ0LZx6aEl5Fe8TH8maW5njPnvkyX4dlPZzCMeWmp49Kd6BG7JFCRo3LXvMcnwB-&s=QXHwLs0sdx-q8Vu4zYipZrDBsdDrFFMo9Glwo4XnX_k&e=
>
> Thanks thanks thanks! This has helped me a lot!
> Malte
>
> Am Sa., 9. Apr. 2022 um 20:53 Uhr schrieb Benedikt V. Ehinger <
> benedikt.ehinger at vis.uni-stuttgart.de>:
>
> > Hi Malte,
> >
> > one more thing: zapline worked wonders for me on the 16Hz artefact,
> > completly removing it - if you use matlab there is a nice toolbox
> > zapline-plus that optimizes some further parameters datadriven. Of course
> > reducing the artefact is the much better option, but sometimes you simply
> > cant.
> >
> > A pitty the U-Bahn also runs with 16Hz :|
> >
> > Cheers, Bene
> > Am 09.04.2022 um 10:10 schrieb Malte Anders:
> >
> > Dear Makoto, dear everybody who engaged in this,
> >
> > Thank you all for the time you've put into this discussion. This is
> highly
> > appreciated.
> >
> > Makoto, you pointed out some things that with 20/20 hindsight I should
> have
> > explained earlier. To shed some more light onto this, those are chicken
> > embryo EEGs at different stages before hatching, like day 09 and day 18.
> > Literature shows that chicken embryo develop a continous EEG somewhere
> > between day 15-16 before hatching and the earliest signatures can be
> > recorded somewhere around d14 (but only intermittend signal and not a
> > continous eeg). They hatch at day 21. The "common" spectra of a continous
> > chicken embryo EEG is not that easy to compile. What I know is that we
> are
> > searching for delta waves mostly, and some high frequent stuff in the
> > higher days (d18, d19). However, amplitudes in literature range from 50
> to
> > a couple of 100's microvolts, depending on the paper. A lot of the
> original
> > research of chicken embryo eeg is from the 1960's and sometimes lacks
> some
> > crucial information (the papers itself are good but I am sometimes
> missing
> > stuff like the range of the Y axis in the figures...).
> >
> > The EEG I sent in the original post is a d11 EEG. This basically
> translates
> > to: no brain acticity (as per existing literature) and thus, amplitudes
> > should be a couple of microvolts at max. On top of that, it is NOT a
> > recording from one single chicken. Those embryos are damn small and we
> are
> > currently getting 2 electrodes placed on the head until we run out of
> > space. A 16 ch chicken embryo recording would be a project for itself.
> > Hence, the image I shared shows recordings from 8 different embryos, in
> > pairs of 2. Electrode 1 and 2 are embryo #1, 3 and 4 embryo #2 and so on.
> > Of course they don't share the same reference, and they have been
> recorded
> > at different points in time. The only reason I visualized it like this is
> > so that I can analyze the data of 8 embryos at once. This should answer
> > your point number 2 and 3.
> >
> > Answer to 1: After a lot of more trying, I 99% agree with you.
> > We tried a lot of things. First, we re-grounded the whole equipment to an
> > alternative grounding post. Then, we
> >
> > -switched the phase (i.e., took another power outlet by using an
> extension
> > cord. The power outlet is used for other EEG recordings in the same
> > facility where the artifact is not present.
> > -we added another grounding electrode to the embryo to avoid any possible
> > ground loops.
> > -we combined 1 and 2
> >
> > Right after re-grounding in the recent recordings, the artifacts where
> NOT
> > present (or only very very slightly visible in the FFT). The
> abovementioned
> > 3 others points did not improve the situation any further.
> > However, in the raw data, there were occasional episodes (~10 seconds)
> with
> > clearly visible 16 Hz noise. That means that the artifact is not gone
> once
> > and for all, but I can for now just cut the parts of the data out. In a
> > d19, I now see mostly delta stuff (<150 microvolts) and in a d12 embryo I
> > see no activity with a random EEG around the baseline in the amplitude
> > range of -10 to 10 microvolts (where it should be). For the first time I
> > got an "idle EEG" for a dead-brain embyro, after 120 recordings where
> over
> > 90% (!!!) of the recordings where contaminated with the 16 Hz artifact
> that
> > has a higher power than our delta waves.
> > Image:
> https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_JzXK6pG&d=DwIFaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=Zq5xPs9WDjJeoF7zJqfjYA95JujAN9tRm60Kjxz-ovscUMc9IU9ox-29kfCQ6Axo&s=k-UBc3GAcYzORyYr8P0Yiha48mJ1aB6JbEH-nevIooA&e=
> >
> > What has changed? Maybe, the timetable or whatever... The U-Bahn runs
> > directly (!) below the basement where the recordings are carried out. On
> > top of that, there is currently very heavy construction work going on.
> > Other EEG machines in the building however do not pickup that artifact. I
> > will have to look at further recordings and see if anything changes, for
> > now, we can work the way it is. I am unsure if one of our measures
> > mentioned above improved the situation.
> >
> > 4,5,6,7: this is very good information, I will definitely dig deeper into
> > this next week. For now, we are gambling and rolling dices that more
> often
> > than not land on the "16 Hz" side.
> >
> > I will keep you guys updated!
> > Malte
> >
> >
> >
> > Makoto Miyakoshi via eeglablist <eeglablist at sccn.ucsd.edu> <
> eeglablist at sccn.ucsd.edu> schrieb am Fr.,
> > 8. Apr. 2022, 19:28:
> >
> >
> > Dear Malte,
> >
> > I read this discussion with great interest. I read the blog linked by
> Bene.
> > I left a comment there but let me repeat it here with more detail.
> >
> > 1. I'm convinced that the powerline for the German train system is the
> > source of this artifact.
> > 2. But I am still curious how it can cause such a strong,
> > channel-by-channel different artifact IF all the recording electrodes
> > are
> > unipolarly connected to a reference electrode AND recording equipment
> > and
> > the subject's body (animal or embryo?) is electrically connected (i.e.
> > ground is shared). When you perform a salt water bucket test, do not
> > forget
> > to share a ground between the water and the EEG recorder. Actually,
> you
> > might test what happens when ground is not shared.
> > 3. Your time-series data show same patterns for electrode 1&2, 3&4,
> 5&6,
> > etc. with the same phases. Does it mean these electrode pairs were
> > closely
> > located, or their leads were paired, etc? What is your explanation for
> > this
> > pairing effect if the source of the artifact is the German powerline
> for
> > the train system?
> > 4. Bene's solution reminds me of a paper for reducing TMS-induced
> noise
> > during simultaneous TMS-EEG recordings. If you are going to follow
> > Bene's
> > solution, you might find this paper also useful. However, I still
> think
> > that the case of TMS-EEG is very different from the case of German
> > powerline for the train system on the point that the source of the
> > artifact
> > is VERY close in the former while it is very far for the latter.
> > Sekiguchi H, Takeuchi S, Kadota H, Kohno Y, Nakajima Y. (2011).
> > TMS-induced artifacts on EEG can be reduced by rearrangement of the
> > electrode's lead wire before recording. Clin Neurophysiol.
> 122:984-990.
> > 5. Speaking of a large-amplitude environmental noise, last year I
> > reported a case of nearly +60dB of line noise observed in one of my
> > collaborators' lab. Although I was not there when the data were
> > recorded, I
> > can imagine the ground electrode had a problem, either not connected
> at
> > all
> > or hardware failure. As the Figure 2 shows, in the time domain all you
> > can
> > see is a 60 Hz sine wave with +/- 850 microV when the recovered signal
> > had +/- 10 microV.
> > Miyakoshi M, Schmitt LM, Erickson CA, Sweeney JA, Pedapati EV. (2021).
> > Can We Push the “Quasi-Perfect Artifact Rejection” Even Closer to
> > Perfection? Front Neuroinformatics. 14.
> >
> https://urldefense.proofpoint.com/v2/url?u=https-3A__doi.org_10.3389_fninf.2020.597079&d=DwIFaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=-H0x6lMjc1PsxcfjKoXqDiULAp4L6M0ZepOLvBnAuP5vcOeRDpNaIU6KZ45pZa8W&s=8mbFbxvyeEuI44IUAzMAYzbA1uZBD0uWbqP02F-Mfiw&e=
> > 6. When you think about the ground and reference issue, if you have
> not
> > heard of the word 'infinity reference', I recommend you check out my
> > Wiki
> > page. When measured our EEG or whatever bioelectric potentials from
> our
> > body using the 'infinity reference', it would be completely unusable.
> >
> >
> https://sccn.ucsd.edu/wiki/Makoto's_preprocessing_pipeline#Why_should_we_add_zero-filled_channel_before_average_referencing.3F_.2808.2F09.2F2020_Updated.3B_prayer_for_Nagasaki.29
> > 7. Read also 'Electric Field of the Brain' (Nunez and Srinivasan,
> 2006)
> > p.42 '9. The reference Electrode'. By the way, Fig. 1-21 in p.43 is my
> > favorite figure ever.
> >
> > If you have follow-up info, please share it with us. I'm curious how this
> > issue can be solved and how the result turns out.
> >
> > Makoto
> >
> > On Thu, Mar 31, 2022 at 8:13 PM Malte Anders <malteanders at gmail.com> <
> malteanders at gmail.com>
> > wrote:
> >
> >
> > Hi Bene,
> >
> > Thank you very much!
> >
> > Someone already mentioned the 16 2/3 Hz thing with trains to me via
> >
> > private
> >
> > Email, but your blog post sheds some more light onto this. Indeed, the
> > measurement is taking place in Germany, in some basement at TU Munich...
> >
> > I
> >
> > need to have a look at this next week.
> >
> > My idea is to connect an Open BCI for troubleshooting, which can be run
> > from 4 x 1.5 Volt AA Batteries. In that case we would eliminate any
> > external artifacts that are induced via the grounding circuit.
> >
> > Best wishes!
> > Malte
> >
> > Benedikt V. Ehinger <science at benediktehinger.de> <
> science at benediktehinger.de> schrieb am Do., 31.
> >
> > März
> >
> > 2022, 21:25:
> >
> >
> > Hi!
> > We observed this with a railway line (important: in Germany) nearby:
> >
> >
> >
> https://urldefense.proofpoint.com/v2/url?u=https-3A__benediktehinger.de_blog_science_trains-2Dand-2Deeg_&d=DwIFaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=e6pZJC_XEDiwU7oAh8oG3hkF2MrfOHh2eni06_FGdgEvmT5Qcv5j9FM8RxQ80SdH&s=RECakigl7u0Nl4lJBABntcKX-5agXzdxePZ9ayIC_q4&e=
> >
> > . Note that 16
> >
> > 2/3 is exactly 50Hz divided by 3. Hope that helps already :-)
> > Cheers, Bene
> >
> > Am 31.03.2022 um 12:20 schrieb Malte Anders:
> >
> > Dear all,
> >
> > We are currently trying to record EEGs on the open brain of certain
> >
> > animals
> >
> > - so no "standard" human recordings. The gold electrodes are placed
> > directly on the brain, without any conductive gel applied.
> >
> > We are encountering a strange ~16 Hz artifact, with harmonics around
> >
> > ~10.x
> >
> > Hz and ~5.x Hz, see these images:
> >
> >
> >
> https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_NKMcrf0&d=DwIBaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=c88_VOjmnuVwvN-A41WN_LrcUXdxqhHwgozv6g_vBGXNTPB1AS0LrJkqDp8M8B44&s=E8I2mkOB0AmaOPUEUXwPZvKu-BkYWuzhow82gl11YCg&e=
> >
> >
> https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_Z8zkCpp&d=DwIBaQ&c=-35OiAkTchMrZOngvJPOeA&r=kB5f6DjXkuOQpM1bq5OFA9kKiQyNm1p6x6e36h3EglE&m=c88_VOjmnuVwvN-A41WN_LrcUXdxqhHwgozv6g_vBGXNTPB1AS0LrJkqDp8M8B44&s=haiZvyK3aT-jVpA4wpHVfjQQrPRs1O0cx88EnN9NitE&e=
> >
> > Anybody ever saw artifacts in these frequency ranges? I am trying to
> >
> > narrow
> >
> > down if those artifacts stem from our recording setup itself (maybe
> >
> > bad
> >
> > impedance between the brain and the measuring electrode?), or from
> >
> > any
> >
> > other external hardware in our recording setup.
> >
> > Any help is much appreciated!
> > Malte
> > _______________________________________________
> > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> > To unsubscribe, send an empty email to
> >
> > eeglablist-unsubscribe at sccn.ucsd.edu
> >
> > For digest mode, send an email with the subject "set digest mime" to
> >
> > eeglablist-request at sccn.ucsd.edu
> >
> > _______________________________________________
> > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> > To unsubscribe, send an empty email
> toeeglablist-unsubscribe at sccn.ucsd.edu
> > For digest mode, send an email with the subject "set digest mime"
> toeeglablist-request at sccn.ucsd.edu
> >
> > _______________________________________________
> > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> > To unsubscribe, send an empty email
> toeeglablist-unsubscribe at sccn.ucsd.edu
> > For digest mode, send an email with the subject "set digest mime"
> toeeglablist-request at sccn.ucsd.edu
> >
> > _______________________________________________
> > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> > To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> > For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
> >
> >
>
> --
> Mit freundlichen Grüßen,
>
> Malte Anders
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
More information about the eeglablist
mailing list