Future Plans

Short Term

  • Stay with the biofeedback approach using the swelling or contracting circles as a focus in the User Interface (UI) for the moment
  • Stay with mu suppression in the opposite hemisphere to the side imagined moving as the best marker of success in the trials
  • Run a larger number of sessions to see if there is detectable bilateral mu suppression in my current set up using my current motor images, hands squeezing or pumping up circles.
  • Try varying the imagery but concentrate on the hands because they are said to have the clearest response when monitoring C3 and C4
  • Continually improve the User Interface to show the mu and beta responses at C3 and C4 as the focus of my feedback.  
  • Review processing methods of the band powers produced by the BrainFlow software.  
  • Build the variables I have used, e.g. tolerance levels to avoid noise, prompt durations, etc., into a settings file so that they can be varied easily for trial sessions
  • Run enough sessions to decide whether mu and/or beta suppression has been identified. Update the graphs of results to focus on trends.
  • If mu suppression becomes evident and reproduceable, yippee, move to Longer Term – Success So Far.
  • If not, consider more significant changes.  Go to Medium Term – Think Again

Medium Term – Think Again

Copilot has already suggested that perhaps C3 C4 Cz and CPz would be better electrode positions based on other Motor Imagery work, but I’d like to hear from actual people who are more expert than me and perhaps have achieved success. So the Questions page and please mail your feedback on electrode positions to me.  I would then have to decide on the options suggested and whether to change.  CPz is currently used as Ref on the OpenBCI cap but maybe another position could be used for that.

I am also hoping to have some feedback on whether or not I would benefit from upgrades in kit. Possible upgrades include:

  • Updated Ganglion or maybe Cyton if this would improve my results.  My Ganglion is over 10 years old which is a long time in Tech.
  • Changing to the newer dry electrodes.  My experience with the first iteration of these wasn’t great but these may be much better
  • If the dry electrodes would be a good move, design and print a frame for holding them in place in my chosen new electrode positions
  • Consider the feedback I receive and perhaps redesign the UI and change some processing algorithms.

Then back to experiments to get mu/beta suppression or any other suggested markers of motor imagery EEG signal changes.

Longer Term – Success So Far!

Ok, we can now produce a detectable change in EEG mu and beta band powers based on motor imagery.  The next step is to turn the whole emphasis around and, rather than the computer telling the subject which side to imagine moving, the subject will decide and the computer will be trained to recognise which side the readings are suggesting.  In the history of my project, you may have read that I was intending to use a V1KU for this purpose, if not, please read the Project History page to find out what that is.  However, today AIs are much more capable at pattern recognition than that gadget – or so a totally unbiassed assessment from Copilot would suggest!

Unity has the software features available to use an AI to do just that, so my intention would be to learn enough about the technique to try it out.  I may be able to use shorter sessions too if recognition is reliable.  I would be aiming to recognise just the three patterns, Left, Right and Centre.

Let’s assume for a moment that that works too 😊.  What could we do with that information.  The sky’s the limit and I’m sure there will be suggestions.

  • Move a cursor
  • Steer a 2D car around corners
  • Anything that you can code with just 3 variables

In a more frivolous moment, I imagined using the Left and Right positions as a dot and a dash and the centre as an end marker.  You could then write words in Morse Code.  I would be in for the prize for the slowest “Hello World” ever.

Sharing the Software

This version is highly specific to my kit, Ganglion, EEG cap, Windows PC.  I am happy to share the Unity project with others working in the same area but they would need to know enough about Unity and C# to be able to adapt it to their situation.

Unity can produce executable apps for Windows, Mac, iPads, iPhones and Android devices.  The Project History page mentioned Noah the Boa, the Unity app I wrote. (It is available from the Google Play store if you have young children)  So another possibility is to make the project more configurable and make an executable version so that others could use it out of the box.  It would probably be a Windows app in the first place.  I am more familiar with that environment and do not know at present whether a Ganglion or Cyton would link to an Android device. Since mine links to the PC via Bluetooth, it seems a plausible method for a tablet too.  As far as making it available for Cyton too, and Mac platforms, I would need additional kit (like a Mac or a Cyton!) to do that.  But there are lots of things we could do.

But, hey, feet on the ground! Let’s get the basic stuff to work first and dream later.  At this stage I’m more confident than not, but it may be a bit of a long road.  Thanks for any input in advance.  I will try to add a blog every now and then to keep people posted with changes and results.  Watch the Results So Far page for updates in that area too.