mscohere for each frequency band_EEG data

Hi all,

I'am working with the following code to get coherence values. I get some values but how can I get the measure for each frequency band? I tried to define F as F=[8.05.13] for alpha and F= [1.0.5.4] for delta. However, I get exactly the same values for alpha and delta. Any help would be greatly appreciated.

ch1= 7 % left electrode
ch2= 48 % right electrode
window = 2*EEG.srate;
nooverlap = 0;
nfft =window*2;
fs= EEG.srate;
F= [8.05.13]% change this for different bands

[Cxy, F] = mscohere (EEG.data(ch1,;),(EEG.data(ch2,;)window, nooverlap,nftt,fs);

[Star Strider]

First, this:

F= [8.05.13] % change this for different bands

does not make sense, and should throw an error. (It did when I tested it to see what running it would do.)

If you want to create it as a vector beginning at 8 with a step of 0.05 and ending at 13, do this instead:

F= 8:0.05:13; % change this for different bands

however that is also likely not appropriate.

This is sufficient:

F = [8 13]; % change this for different bands

Also note that you are not passing ‘F’ as an argument to mscohere, so the function does not even know that it exists!

[belen calvo]

Hi, sorry I'm new to matlab and wanted to try this analysis with some dataset I found. I'm used to analyzing in R but I figure I could learn a bit of matlab now that I have a lot of time due to the pandemic. So, basically I'm trying a new analysis with old data to learn matlab and also to see if something interesting comes up.

You mean the following:

f= [8 13];

[Cxy, F] = mscohere (EEG.data(ch1,:), (EEG.data(ch2,:), window, nooverlap,nfft,f,fs)

It says error the sampling frequency must be a scalar...

[Star Strider]

No worries!

MATLAB function arguments must be entered in the order described in the documentation. If you do not want to specify an argument (that is, use the default value in the function), enter that argument as empty square brackets [], although only some arguments may be defined as appearing together. That is the situation with mscohere. So to use ‘F’, it will be necessary to use this set of arguments, again in the order required (quoting from the documentation):

[cxy,f] = mscohere(x,y,window,noverlap,f,fs) returns the magnitude-squared coherence estimate at the frequencies specified in f.

Note that here ‘nfft’ is not an accepted argument in this context. The only option is to use those listed, if you want to use ‘F’.

(I have rarely used mscohere, so my experience with it is limited.)

[belen calvo]

THANKS! yes, now I get different coherence values for each band. The only problem now is that I only get as output two values-I guess one for 8 and another one for 13 in F= [8 13]. I would like to get all the coherence values in the range going from 8 to 13 in that band (e.g. alpha). I tried F= [8:13] and it returns 6 values..not sure I'm picking what I want though. I'm gonna keep on reading about mscohere and papers that have used this function.

[Star Strider]

My pleasure!