[Eeglablist] Enquiry regarding power values calculated using spectopo

[Cedric Cannard]

Hi Gladys,

I have never summed channels' power values before so I am not sure what type of analysis you are doing.

For continuous data, I would generally get the mean power value of each frequency band of each channel, on raw power values (with prior extraction of non-overlapping 1-sec windows with 1-sec gaps between them to limit auto-correlation).
Then do the log10, and then average over electrodes if that's what you want to do.
I am assuming you are comparing it to another condition. If so, I would do the comparison for each electrode individually. Then, I would do a permutation stat test, and correction for multiple comparisons (FDR for example).

You can do all this easily with the EEGLAB STUDY.
You can do the window extraction by adding the power spectrum parameters: 'epochlim', [0 1], 'epochrecur', 2

Cedric

--
Cédric Cannard, MSc, PhD candidate in Neuroscience
Université Toulouse III, CerCo Lab, CNRS, France
ccannard at protonmail.com

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Sunday, May 3, 2020 6:51 AM, #HENG JIAMIN GLADYS# <JHENG007 at e.ntu.edu.sg> wrote:

> Hi Cedric,
>
> Thanks for the reply. I have also selected various channels for the frontal region (e.g. F1, F3, F2, F4).
>
> In that case, would you do log10 for each channel first before summing them together? i.e., log10(F1) + log10(F2) + log10(F3) + log10(F4)
> or
> Sum the power values of the channels first then do log10 on the summed value? i.e., log10(F1 + F2 + F3 + F4)
>
> Regards,
> Gladys
> ---------------------------------------------------------------
>
> From: Cedric Cannard <ccannard at protonmail.com>
> Sent: Sunday, May 3, 2020 1:36 AM
> To: #HENG JIAMIN GLADYS# <JHENG007 at e.ntu.edu.sg>
> Cc: EEGLAB List <eeglablist at sccn.ucsd.edu>
> Subject: Re: [Eeglablist] Enquiry regarding power values calculated using spectopo
>
> Hi,
>
> There are probably diferent ways to do it.
> I personally do it like this for example for continuous data over channel 1:
>
> 1) spectra = spectopo(EEG.data(1,:), EEG.pnts, EEG.srate, 'plot','off');
> 2) spectra = 10.^spectra; %brings log dB signal to power (amplitude = power squared)
> 3) compute mean, SD or whatever you want here on power values
> 4) spectra = log10(spectra)  %bring back to log for more normalized signal across subjects
>
> Best,
> Cedric
>
> --
> Cédric Cannard, MSc, PhD candidate in Neuroscience
> Université Toulouse III, CerCo Lab, CNRS, France
> ccannard at protonmail.com
>
> Original Message
> On Saturday, May 2, 2020 6:12 AM, #HENG JIAMIN GLADYS# <JHENG007 at e.ntu.edu.sg> wrote:
>
>> Dear EEGlab list,
>>
>> I have calculated power values using the following formula, as found in a previous post (https://sccn.ucsd.edu/pipermail/eeglablist/2015/009249.html):
>>
>> % for your epoched data, channel 2
>> [spectra,freqs] = spectopo(EEG.data(2,:,:), 0, EEG.srate);
>>
>> % delta=1-4, theta=4-8, alpha=8-13, beta=13-30, gamma=30-80
>> deltaIdx = find(freqs>1 & freqs<4);
>> thetaIdx = find(freqs>4 & freqs<8);
>> alphaIdx = find(freqs>8 & freqs<13);
>> betaIdx = find(freqs>13 & freqs<30);
>> gammaIdx = find(freqs>30 & freqs<80);
>>
>> % compute absolute power
>> deltaPower = mean(10.^(spectra(deltaIdx)/10));
>> thetaPower = mean(10.^(spectra(thetaIdx)/10));
>> alphaPower = mean(10.^(spectra(alphaIdx)/10));
>> betaPower = mean(10.^(spectra(betaIdx)/10));
>> gammaPower = mean(10.^(spectra(gammaIdx)/10));
>>
>> From my understanding, spectopo "plots the mean log spectrum of a set of data epochs at all channels as a bundle of traces".
>> Therefore, the spectra output is mean log values.
>> With the formula to compute absolute by taking 10.^(spectra(freqIdx)/10), does that mean that the power value is transformed back to actual raw values?
>>
>> I'm asking this as my resulting data appear to be mostly positively skewed, and I am planning to do a log10 transformation, but am not sure if I am applying log transformation twice on the data.
>>
>> Thanks in advance,
>> Gladys
>>
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
>> For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu