Questions for channels and maximum frequency


#1
  1. There are four channels : FP1 (left,front) FP2(right,front) TP9(left,back) TP10(right,back) , correct?
  2. What’s the maximum frequency that can be recorded? With emotiv insight and epoc+ it’s 43Hz. I see sampling rates such as 220Hz and 500Hz (two sampling rates?) in the specifications so the maximum frequency that could be recorded would be half of whichever of those two sampling rates is the correct one.

#2

The sensor locations depend on your head size. Interaxon updated their documents to state that the front two sensors are closer to AF7 and AF8. So the four locations are AF7, AF8, TP9 and TP10.

The “sampling rate” and the “maximum recorded frequency” are two separate things.

Sampling rate, is the rate at which data is sent out from the system. For the Emotive headset this is 128Hz.

For the Muse, the sampling rate is set via the preset value when you connect to the device, the documentation for the presets is in the developer docs here.

  • The MU-01 (2014 Muse with two USB ports) has RAW EEG rates of 220Hz and 500Hz available.
  • The MU-02 (2016 Bluetooth LE Muse with one USB port) has a RAW EEG rate of 256Hz.

Regarding the maximum frequency that can be recorded, this is talking about the frequency data which is calculated from the RAW EEG data. The name of this calculation is a Fast Fourier Transform, so you will often see this referred to as just “FFT data”.

The Emotive SDK outputs pre-calculated FFT data, which has been filtered for background noise and electrical noise. They output these FFTs up to a maximum of 43Hz.

The Muse SDK does not output pre-calculated FFT data, instead it outputs individual brainwaves (Alpha, Delta etc), and more importantly RAW EEG. From the RAW EEG data, you can do you own FFT calculations without any of the limits of the Emotive SDK, but on the downside, you have to do your own noise filtering.

FFTs are very easy to calculate and if you want to see the real time FFT output from the Muse, this is available in my Muse Monitor app and looks like this:


In this screenshot the colored bands represent the areas of the five brainwaves.

As you can see here, Gamma (represented by the greek letter γ) is from 30Hz to 44Hz, so this is likely why Emotive cut off their FFT data at 43Hz.


#3
  1. That answers my question
  2. Considering I get the 2016 model (256Hz sampling rate) and I am using the muse monitor app: does the app only show the FFT data (with max frequency being 44Hz) or can it also show the RAW EEG data (with max frequency being 128Hz - 256 divided by two) - and I would probably have to do my own noise filtering after exporting the data to my computer.

#4

There is no maximum frequency to the Muse FFT data, because the Muse does not output FFT data.

If you decided to calculate FFT data from the Muse RAW EEG data, then you can calculate it to any frequency maximum you like depending on your FFT windowing parameters.

Muse Monitor has many graphs which show you all the data available from the Muse.
You can see screenshots on the website: http://www.MuseMonitor.com

Regarding filtering, the Muse has a built in power frequency notch filter, which can be switched between 50Hz to filter out European power noise and 60Hz to filter US power noise. This is applied to the pre-calculated brainwave data (Alpha, Delta etc).


#5

So wouldn’t the technical maximum then be 128Hz? (256 divided by two) or would it be 256Hz?

Or would it be what I mentioned above -50Hz (european power noise)


#6

If you are calculating your FFT over a single 1 second window of 256 samples then yes.

This page has a good article on FFT calcs:
http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/fft_funda/

I would recommend that to increase frequency resolution (not max) you use a sliding sample window of multiple seconds of data, you can see this in action with Muse Monitor, if you press the settings (cog) icon and select a large “FFT Window” of say 2048.


#7

From what I can see the 2016 model is better (and the 2014 model is hard to find and probably used) but you can choose between 220Hz and 500hz sampling rate on the 2014 model? If so how? Just curious


#8

You can only set the 500Hz mode if you’re connecting to the 2014 Muse with a desktop Mac/PC, using Muse IO.

In the Muse IO Command line options, you can specify the preset as AB (6 channel) or AD (4 channel) to get 500Hz data.


#9

Hi,

Your app Muse Monitor is great!
In the csv export, are the blinks removed? Or should I remove it from the export file?

Thank you!

Roger


#10

Both Blink and JawClench markers are included in the Muse Monitor CSV recording, but I don’t display them in the Online Charts, as with eyes open, there are TONS of blink markers. I guess I could add a checkbox to show them on there if you think they’re important?


#11

I am going to use than only from the export file. Because I am going to calculate the blink rate for each 3-minute period. That is great!

Therefore, for the calculation of the the absolute brain waves values, this “noise” already was removed, right?
Can you provide how this “noise” (blink and jaw clench) are detected and removed from the export data. It is because I am interested in carry out research using Muse and your app. So, I need to describe the methods in detail.

Thanks

Roger


#12

The absolute brain wave values have the noise filtered, but I’m not sure exactly how. You’d need to ask Interaxon directly.


#13

Thank you!
I have a doubt about how the brain waves *(alpha, beta, gamma, etc) are calculated and exported to csv.

  1. I know that the absolute values are transmitted at 10Hz. However, which epoch does the app use to calculate it (FTT window)? It is 256 ~ 1 second?

  2. If I change the “Recording Interval”, Does the FTT window change? I mean, regardless if I choose “constant” or “1 second” or “1 minute”, Is the value that appears in the export file for absolute waves a FFT calculation from 256 points? I am interested to know how changes in the recording interval can change the results for the brain waves.

Thank you in advance.

Roger


#14
  1. The absolute values are calculated by the Interaxon SDK, not by my own FFT. Their documentation is here.
  2. The recording interval is how often a data snapshot is written to file, it does not change any of the calculations.

#15

Excellent. thank you so much