Questions

Context

To put my questions into context, my current project aim is to create a BCI using motor imagery to evoke a recognisable and reproduceable response in EEG signals from the chosen electrode locations.  The responses I am looking for are mu suppression and beta suppression. The software uses a visual feedback system to assist the subject in improving the imagery and hence the response.  Essentially, this involves training the subject to evoke this response from prompts on the screen directing them to imagine movement on one or other side of the body or no movement at all.  In due course, the aim is to build in an AI to recognise the pattern of signals and decide which category of movement is being imagined. See the Future Plans page.

I want to use this page to ask questions from equipment suppliers and people a lot more expert in the field than I am in the hope that they would be kind enough to offer advice and guidance.  My interest is as a hobby rather than a profession.  With that in mind, some of my questions are set out below. 

Hardware

This question is aimed at OpenBCI and/or other suppliers of relevant equipment.  I currently use a Ganglion, 4 channel board from OpenBCI which I bought over 10 years ago.  Combined with this, I use an EEG cap with electrodes using electrode gel.  Would I get any benefit from updating these to newer models? Is the current Ganglion the same or is it faster, for example.  And would the newer range of dry electrodes, which are more convenient to set up, give me a similar signal strength with similar impedance?

Electrode Placement

I have 4 channels available in the Ganglion. I have chosen C3/C4 and P3/P4 as the positions being monitored.  Are these the best locations for motor imagery.  An alternative has been suggested as C3, C4, Cz and CPz.  For the moment I am sticking with my original choice.  The OpenBCI EEG cap has a Reference electrode at the CPz location so I don’t want to rock the boat, but I’d be interested in any feedback from those who have used motor imagery.

Frequency Band Choices

From what I have read, the most reliable method of detection is mu frequency suppression in the side of the brain opposite to side on which there is imagined movement.  I collect both mu frequency ( 7Hz-14Hz) and beta frequency (14Hz to 30Hz).  These ranges are wider than those which some use but should be fine for my purposes.  I collect beta because there may be some beta suppression too, which can be used as an additional factor in categorisation. Feedback on my frequency choices would be welcome.

Mu Suppression Detection

I outline the methods used on the Results So Far page and would very much appreciate advice on whether the approach is reasonable, or is there a better approach.