Random Perturbations in EKG Signal

juderbjuderb
in Hardware
Took my first EKG measurements today for testing purposes, and am observing some significant random perturbations in the signal.

OpenBCI_8Bit
Electrode 1: SRB2
Electrode 2: 1N
Applied 10th Order, 58-62Hz, Butterworth Bandstop Filter

Has anyone else observed this in their measurements? Any ideas on the source of the perturbation?

Comments

  • wjcroftwjcroft Mount Shasta, CA
    edited April 2015
    It looks like some kind of EMG spiking to me. Could it correspond with movement, sharp respiration, etc.? You say the period is totally random, and not these 15 second intervals? The high frequency nature of it sure points to some kind of artifact. Are your electrodes directly on the chest or on the arms? Do you see this artifact only with OpenBCI / do you have other ECG instrumentation that shows cleaner signals?

    If you are monitoring the signal during recording, do you see any correlations of these spikes to movements?

    I'm not sure of the 'normal' bandpass filtering applied in EKG, but you may want to apply that instead or in addition to your notch. E.g. .5 to 40 hz bandpass. You must be doing some highpass already to remove your baseline? Also other upper band limits I've seen are 40, 100 or 150 Hz.

    https://www.google.com/search?q=ecg+bandpass+filter

    http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3388307/

    Also 10th order filter seems somewhat high order, usually higher orders cause more delays / phase shifts. There are alternatives to Butterworth if you just want notch at 60hz.

    https://www.google.com/search?q=best+dsp+notch+filter

  • biomurphbiomurph Brooklyn, NY
    Please also try relocating your setup. You may be seeing EMF or proximal powerline noise.
  • wjcroftwjcroft Mount Shasta, CA
    @juderb , did you find a resolution?
  • juderbjuderb
    No luck yet. It shows up when the leads are attached, but not connected to the body, so I am sure it is not from any kind of muscle activation. I have also isolated it as much as possible.

    I am trying to record the data to display here, but I am having a problem where the data displayed in the GUI doesn't match the appearance in the saved text file (accounting for filtering).
  • wjcroftwjcroft Mount Shasta, CA
    The leads plugged in, but no body connection is the worst case. Since the leads will be acting as antennas for any noise. Without benefit of the differential canceling or Bias. So perhaps you do have some kind of EMF pulses going on in your lab.

    Have you tried moving to another location? Is this a home environment, industrial, academic? How are your electrodes setup? Differential as shown in the tutorial (IN*P paired with IN*N) or using SRB2 and one IN*N channel? 

    When you have the electrodes on the body, where are they located? Where is your Bias? Have you tried a more standard EKG setup with foam electrodes on the chest? Is this showing up for EEG tests as well as EKG?

    You say the GUI display differs from what you generate from the txt file. Do both show the same noise pulse? Have you changed your filtering as suggested on previous post?
  • wjcroftwjcroft Mount Shasta, CA
    Here's a related post on measuring the self-noise of the ADS1299 A/D converter. You might want to try some of those tests as described by Chip or Joel. The "loopback" can happen either internal to the ADS1299, or with the three electrodes touching each other via some paste.


  • juderbjuderb
    Actually, that post was very helpful. Using the technique described, I confirmed the self-noise for all 8 channels is about 0.13 - 0.18 uV RMS. Thanks for the suggestion.

    I will hook the system up to a function generator and see what I can learn from a well-defined signal.


  • wjcroftwjcroft Mount Shasta, CA
    Did you use the internal self-noise test, or the external with cables?

    What do you see measuring the EEG as described in the tutorial with eyes closed alpha?

  • biomurphbiomurph Brooklyn, NY
    @juderb,
    You can also use the internal test signal in the ADS1299. 
    Please see this page for the specific commands that we have revealed for the user
Sign In or Register to comment.

© Vanilla Keystone Theme 2026