Show me your Ultracortex signals!

algmuralgmur UK
edited May 2021 in Headware

Hi all,

My PhD has a large EEG component to it and I'm used to working with the research-grade systems like those from Brain Products, wet-electrode setups etc. I'm definitely not an expert, but I have a few years' experience analysing signals and working with them. What I'm doing right now is I'm playing around with my OpenBCI Ultracortex IV model with Cyton+Daisy and before I start really getting into the analyses I want to do, my first goal has been to optimise the filtering such that it best matches what I see in the OpenBCI_GUI. I've managed to implement the filters to a reasonable degree but there's still a lot of noise present and it doesn't look good to me.

Now, my baseline level is what I'm used to working with, and I know this dry-electrode hobby-level setup is never going to be equivalent to what I work with at my research institute, but I have no sense of how close I can get. Therefore, I wanted to ask if people wouldn't mind posting some pictures of their (filtered) results look like, to see if I've reached a similar level or if I can optimise further. If my results look like yours, then I'm probably not going to get any better and I can begin with my next steps.

However, if I see many of you with much nicer, cleaner signals (i.e. ERPs, filtered continuous signals) then I know I have more tweaking to do. If anyone wouldn't mind posting the best you've achieved, perhaps with any tips you have found that helps increase the SNR during recording, that'd be greatly appreciated.

Comments

  • wjcroftwjcroft Mount Shasta, CA

    ...my first goal has been to optimize the filtering such that it best matches what I see in the OpenBCI_GUI. I've managed to implement the filters to a reasonable degree but there's still a lot of noise present and it doesn't look good to me.

    Hi Algmur, can you clarify? The "noise present" is visible on both the GUI and your own (Brainflow?) filtering? Or just with your own?

    Wet electrodes with gel, paste, or saline, will always give stronger signals than passive dry electrodes. You can use any of these wet systems with your Cyton.

    Here is Jeremy Frey's comparison of a medical grade g.tec amplifier, with Cyton. Using wet electrodes.

    https://arxiv.org/abs/1606.02438

    Regards, William

  • algmuralgmur UK

    Hello wjcroft!

    Thanks for your reply! I should have been more clear in what I was looking for - sorry about that.
    I meant that my experience of what "good signals" looks like is based on wet-electrode large-scale systems and judging a dry-electrode system like the Ultracortex IV that I have, will never be a fair comparison (the wet electrode setup will always be better). What I am looking for is an answer to just how close one can get, visually, to the best signals with a dry-electrode setup. I think many of us (judging from the posts I've read on here when searching) initially suffer from the fact that we see much cleaner results in the GUI than when we apply the filters ourselves. That seems to come up all the time on here and it's nobody's fault, just a bunch of people trying to figure out what best works for them.

    To clarify, I am only interested (for now) in optimising the best results for the dry-electrode setup I have. I am not specifically interested in minimising a discrepancy between my own filtering and what appears in the GUI. I'm interested in seeing what other people are able to achieve in their dry-electrode setups. That's where I didn't explain myself fully earlier (sorry about that).

    I set up a small experiment for myself where I recorded myself listening to a beeping sound or looking at a picture of a cat (randomised order). There were 50 examples of each. When I did my filtering and interpolated 1-2 bad electrodes I then went to plot the ERP and topography and I expected to see occipital topography vs temporal topographies for the visual vs audio stimuli, respectively. However, what I got out was just a mess:

    Then I started wondering, "Am I asking too much of this technology? Is it possible to do this with a dry-electrode system?" That's what I was getting at with my question. I wanted to see if people were able to achieve nice clear basic responses and if so, then I know I have to work harder on my setup and preprocessing to get there.

    I hope I've clarified my original question enough :)

  • BillhBillh

    I don't have pictures atm, but try setting filters to remove everything above about 14 Hz and see if that looks better. If that looks better and is not just flat, slowly increase the filter on the high end and see at what point your noise returns. If the plot is flat when filtered at 14 Hz you may have a problem with lead adhesion. bad connections or some such..

  • algmuralgmur UK

    @Billh said:
    I don't have pictures atm, but try setting filters to remove everything above about 14 Hz and see if that looks better. If that looks better and is not just flat, slowly increase the filter on the high end and see at what point your noise returns. If the plot is flat when filtered at 14 Hz you may have a problem with lead adhesion. bad connections or some such..

    Thanks Billh! I will give that a go.

Sign In or Register to comment.

© Vanilla Keystone Theme 2024