Greetings from HacDC and another EEG Designer

Comments

• chipaudette

  February 2014

  You guys are badass! Nice work!
  
  Thank you for the compliment. Y'all appear to be using the amplifier-per-channel approach (please correct me), which is definitely the optimum route....but gets expensive really fast, if one were looking to go to higher channel counts. And the number of channels is important.
  
  Four months ago, I believed that one only needed a few channels of EEG to do everything that a hobbyist could want to do. In the last couple of months, however, as I explored by visual and auditory entrainment explroations (http://eeghacker.blogspot.com/2014/01/blinky-lights-visual-entrainment.html), I discovered that having lots of channels is critical when you're in the exploratory phase of hypothesis testing. In my visual entrainment explorations, I don't know what EEG channels will give interesting results. When you don't know exactly what you're looking for, it's best to record lots of channels. During those studies, I happened to be using the 8-channel version of OpenBCI V1 (and V2). In retrospect, I wish that I had been using the 16 channel version. If I had more channels, my open-ended explorations might have found more interesting results.
  
  So, while you're approach of one-amp-per-channel is very solid, and while you're results are very cool, I would ask you about your opinion about how to scale your solution (within a reasonable cost) up to a higher number of channels.
  
  I myself am slowly moving to the position that, yes, indeed, high channel counts is important for EEG. But that's just my opinion...and there is space is this world for many other opinions...
  
  Either way, I am greatly appreciative of anyone getting up off the couch and putting in the effort to make something cool. And you guys (and/or gals) are totally doing it. You guys are awesome!
  
  Chip
  
  
• mirage335

  June 2014

  Thank you for the reply. While a multi-channel design is indeed useful, at the time, there seemed to be a wide gap between the theoretical noise floor and commercial systems.
  
  Our device is also intended for generic use, not just EEG. The wide supply voltage (+/-12V) provides the dynamic range for stronger signals like EoG and ECG without sacrificing EEG sensitivity.
  
  The ADS1299 chip used by OpenBCI was actually released in the midst of our development process. As that chip may have approximately half the performance of our system, we now have two ways to scale up channel count. One is to use multiple amplifier boards, at a cost of ~$200/channel. Another is to simply use our circuitry and the ADS1299 chip together, giving multiple channels at reasonable performance, and one high-performance channel. Ideally, active electrodes would still be used on both systems.
  
  In the interest of integrating these to systems, I would like to collaborate on OpenBCI's design. Specifically, we should consider sharing connector types (planning on shielded RJ45 with CAT7/8 cable), medically isolated power supplies, headsets, etc.
  
  By the way, the blinkly light demo is intriguing. We attempted to implement something similar using GNURadio, where the EEG signal was upconverted into bianural beats. Ideally, we hope to duplicate similar academic research that has noted sleep quality improvements through such a feedback loop.