Find absolute, relative, and mean power of theta, beta and TBR (theta/beta ratio)

I have raw eeg data that I need to find the absolute, relative, and mean power of the theta and beta bands and theta/beta ratio. I have attached the data file. The data was taken from OpenBCI and imported into MatLab. Please advise how I can do this as I have over 40 file I have to analyze the same. Thank you in advance.

Answer 1

[Star Strider]

You can filter them using the approach in Need to break down EEG signals into 4 frequency bands.

Once you have the filtered signals, square them (using the element-wise exponentiation operator .^, so vector ‘x’ would be ‘x.^2’ to convert amplitude units to power units).

To process your files, see Import or Export a Sequence of Files, and Read Collection or Sequence of Text Files.

I must leave to you how best to read those files, since I have no idea what the format is.

[CJMTEX]

Thank you for your response. I followed the direction from the link you provided to process theta (see below). I changed the Fs number to 250 and changed the bandstop frequencies but go an error on the bandstop. I changed the bandstop frequencies back and got some results but how do I know if the bandstop frequencies are correct...are they generic numbers typically used? Also, the data returned was an "a & b table of numbers" and a number for "g_theta". How do I know which one is the filtered signal so I know which one to square? I'm sorry if I am asking remedial questions, I am new to this and fill like a fish out of water. Thank you again.

Fs = 250;                                               % Sampling Frequency
Fn = Fs/2;                                              % Nyquist Frequency
Wp = [4  7.9]/Fn;                                       % Theta Passband
Ws = Wp .* [0.6  1.05];                                 % Bandstop Frequencies
Rp = 10;                                                % Passband Ripple
Rs = 40;                                                % Stopband Ripple
[n,Ws] = cheb2ord(Wp,Ws,Rp,Rs);                         % Determine Optimal Order
[b,a] = cheby2(n,Rs,Ws);                                % Transfer Function Coefficients
[sos_theta,g_theta] = tf2sos(b,a);                      % Second-Order-Section For Stability

[Star Strider]

My pleasure!

Since I posted that, my filter design approach has changed. I would now design it as:

Fs = 250;                                               % Sampling Frequency
Fn = Fs/2;                                              % Nyquist Frequency
Wp = [4  7.9]/Fn;                                       % Theta Passband
Ws = Wp .* [0.8  1.05];                                 % Bandstop Frequencies
Rp =  1;                                                % Passband Ripple
Rs = 60;                                                % Stopband Ripple
[n,Ws] = ellipord(Wp,Ws,Rp,Rs);                         % Determine Optimal Order
[z,p,k] = ellip(n,Rp,Rs,Wp);                            % Transfer Function Coefficients
[sos_theta,g_theta] = zp2sos(z,p,k);                    % Second-Order-Section For Stability

figure
freqz(sos_theta, 2^16, Fs)
set(subplot(2,1,1), 'XLim',[0 20])                      % Optional
set(subplot(2,1,2), 'XLim',[0 20])                      % Optional

This designs a stable, efficient filter with narrow transition regions. Experiment with the ‘Ws’ multiplication limits to get the result you want. To use this for the other filters as well, simply change the ‘Wp’ limits. Everything should scale appropriately.

Use filtfilt to do the actual filtering.

[CJMTEX]

I see that the numbers changed once I reran it but I still am unsure which number to use to square. Please advise.

[Star Strider]

I am not certain what you are doing. Square the output of the filter and then sum it to get the sum--squared result, or take the square root of the mean of the squared vector to get the RMS (root mean square) amplitude.

Note also that I have no idea how to read those files, so I cannot provide an example.

[CJMTEX]

Thank you for you help. I was wondering if you knew of a person or company I can pay to process my eeg data? I have been trying to figure this out for months and it seems that the further I dig the more confused I become. I am running out of time so at this point I just need to get the data so I can write up my findings. Thank you in advance.

[Star Strider]

There is no need to do that.

There are several posts on MATLAB Answers that filter EEG appropriately it (I wrote at least some of them). After you filter them, do whatever processing you need to do on the results.