How to make brain wave amplitudes comparable?

musemonitor
excel
csv

#1

Hi,

I’m currently trying to use Muse to make an audio representations of the 5 brainwaves (alpha, beta, delta, theta, gamma) from a muse reading of a conversation. My issue, which I have seen others discuss here as well, is how to handle the relative differences in power between the waves. Let me use this chart from a session as example:

This chart is made by:

  1. Recording a session with muse monitor as a csv file
  2. Opening in excel
  3. Running the script provide here: http://musemonitor.com/Macro.php

I then convert the resulting data for each brain wave into audio volumes and use this in a audio software to represent which brainwave is most active during the different parts of the conversation.

However, as can be seen in this particular example, the amplitude of the brainwaves are really hard to compare, meaning that even though the relative fluctuation in one specific brainwave is accurate, they don’t seem to be using the same scale and the output is - in this example - dominated by the delta wave. Given that delta waves are most prominent during deep sleep, and this example is from a normal conversation, this seems odd and I’m guessing due to partly the difference in relative power and partly how delta is more sensitive to disturbances such as open eyes. But let’s ignore the disturbances for now and focus on the math.

So here comes the question: does anyone know a mathematical method to make the values use the same scale i.e. so that the amplitudes become comparable? So that a delta wave value higher than a theta wave value actually means that the brain is currently predominated by delta waves than theta waves? In the example above, dividing all delta values with 2 is tempting…:slight_smile:

Thanks in advance.


#2

You said this recording is from a “normal conversation”, so your Delta is likely high due to eye muscle movement interference rather than brain activity.

You can see this in action very clearly with Muse Monitor. Switch to the brainwave graph and stare at a single point on the screen for a few seconds, now make your eyes look left and right a few times and you will see all the brainwave values jump up, especially Delta. I believe Delta suffers from eye interference the most as it is the lowest frequency band and when you move your eyes it makes big slow (relatively speaking) arcs in the raw data from the eye muscle electrical signals, which you can see in the Muse Monitor raw data view.

To get the cleanest brainwave data, try not to move your eyes, or move them slowly to minimise the effect.


#3

Thanks for the brisk reply!

Yea, but do you think that interference (eye movement) would be this high? Maybe so, and if that’s the case I’m happy, then the data makes sense.

So do you think the data can be used in a comparative way (e.g. same amplitude for theta and gamma means same brain activity for these two brainwaves)?


#4

Yes, I’m pretty sure the high delta is eye movement since you said you recorded this while talking to someone rather than relaxing.

Regarding comparing the data for audio, that’s up to you! If it’s just for an art project, then I’d just normalize the data until you get some sounds that you like.


#5

Yes, totally get that. I’m trying to make a representation to some individuals of what waves are dominant during various parts of the conversation. So, eye movement disturbance is not a problem, since that actually happens. However if the brain waves have different powers and the resulting amplitudes are not really comparable it gets trickier, since the reaction from them - and me - would be asking:

“So, how come that brain wave is that prominent all the time?” and the answer is because I don’t know my science…

What I 've done is to normalize the output where the highest value for any of the waves (in this case delta) is 1 and the lowest for any of the waves is zero, and the volumes vary within this scale. Hence, delta waves sound much louder, but again - not a problem if this is really the case, and not me mixing up relative/absolute scales or different output powers or something like that.


#6

Just to clarify; eye movement doesn’t cause Delta waves. During eye movement there are electrical (non brain) signals that are sent to the eye muscles. This electrical data interferes with the brain data and causes Delta to rise. So effectively the Muse becomes an EMG rather than an EEG.

Once you eliminate the eye muscle interference, either with sophisticated signal processing, or by re-recording without eye movement, then you’ll actually have comparable signals.


#7

Yes I get that. The physiology part is no problem, it’s the signal processing I’m unsure of. But not anymore. Thanks!


#8

Hi erikradbo,

many natural systems, like the brain, are said to be scale-free. An important characteristic of scale-freeness is that power and frequency are connected by a power-law: the lower the frequency the higher the power. EEG is no different: even in the absence of artifacts delta has a higher power than higher frequencies. To make band-powers more comparable you could do something like pre-whitening: fit a power-law and subtract it. Apart from that there are many other things you could do to, like: find the long-term mean of these time series you are comparing and subtract it.