Polarity of ERP signal

JkempelJkempel United States

Hi, 

I am using a cyton board with a daisy board attachment (see pictures in the end). This is for an ERP based study. 

The processing steps involve band pass filtering the signal to stay
between 0.05 - 50 Hz using an FIR filter, followed by averaging of CP channels,
baseline correction and detrending. The resulting ERP signal for neutral visual
stimuli (see below) looks reversed.

I was looking at a previous forum post about the polarity of the signals recorded (https://openbci.com/index.php/forum/#/discussion/comment/9867).

 I was wondering if this is because of the
reversed polarity due to attaching the electrodes in the negative pins (see
attached) and we should simply flip the signal

Thanks!

image

imageimage

Comments

  • wjcroftwjcroft Mount Shasta, CA
    J,

    Yes, if you connect using the default pins, (IN1N, IN2N, ... IN8N) and SRB2, all closest to the board. Then your ERP will 'appear' inverted because the input channels are the negative lead of the differential amp. The SRB2 reference bus gangs together all the positive pins, IN1P, ... IN8P.

    William

  • marles77marles77 Poland

    @wjcroft
    Hi,
    What would be a preferable way to reverse the reversed polarity in this case? Connecting electrodes to the top pins or maybe recalculating brainflow data? If the latter one is better, what would be the best method and time for it?
    Marcin

  • wjcroftwjcroft Mount Shasta, CA

    Marcin, hi.

    Inverting the signal polarity in software would seem to be much easier than changing the pin connections. The reason the pin option is not recommended is that the two SRB pins, SRB1 and SRB2 (the one closest to board and the default used) -- these SRB pins are NOT interchangeable. The ADS1299 allows more flexibility when using the SRB2 bus as a reference connection.

    William

  • marles77marles77 Poland

    OK, that makes sense. So, do I understand it correctly, that each chanel/signal oscilates around 0 and all that is needed is to multiply values (microvolts) by -1? Or is there any additional constant value that needs to be taken into account? Sorry if this is a silly question, but I just want to be absolutely sure.

  • wjcroftwjcroft Mount Shasta, CA

    Yes the sign inversion should work. Do note that the raw channel data from from Cyton, has not filtered off the DC offset. So you may first want to do a high pass at .5 Hz, or bandpass from .5 Hz to say 45 Hz, before your -1 multiply.

    https://openbci.com/forum/index.php?p=/discussion/201/large-millivolt-data-values-fbeeg-full-band-eeg

    Regards, William

  • marles77marles77 Poland

    Understood. Thanks very much.
    Regards
    Marcin

Sign In or Register to comment.