[Eeglablist] Calculating to determine the most dominant frequency in a specific period

Mon Sep 24 13:35:13 PDT 2018

Hello Ihshan, brief notes below.

****************************************************

1. One can make different .set files containing only one kind of sections
/conditons and not others, for each single subject. Then load them into
eeglab study, clearly defining which conditions are which in the Study gui.
Then use the study GUI to calculate channel metrics, and to visualize/show
the power for different conditions. Stats comparisons between conditions
for any of the computed EEG metrics are available via the study function.
If you are new to study, download the eeglab tutorial data (the full study
data) and explore that data via the study GUI. If you're not sure where to
find into about eeglab STUDY, try googling "eeglab study tutorial"
In other words: you need to have different files that contain one kind of
condition, and other files that contain another kind of condition, etc..,
and then make a study with those files, which will allow you to compare
spectral power across your sections/conditions.
Note also, if you want to try it, that the new updated STUDY allows loading
full files, and it can create different conditions based on event
information in the data you load in.

2. Basic theory/facts about spectral power involves the Power law (low
frequencies are much larger than higher frequencies). If you google the
topic on Google or google scholar, you will find a range of online
resources, such as, for example, Mike X. Cohen's teaching site for
time-frequency/EEG, and also, as another example, brief overviews such as
those from Voytek lab:
*http://voyteklab.com/interpreting-the-electrophysiological-power-spectrum/
<http://voyteklab.com/interpreting-the-electrophysiological-power-spectrum/>*
Basic handbooks of EEG have chapters regarding spectral computation and
characteristics. For example, the Luck handbook, and the Mike X Cohen
handbook.

Regarding the code you found on Makoto's site, it's giving you absolute
power , in the second part, as per the comments in the code from Makoto. In
the first part, what's it doing is simply reading the output of the
spectopo function. Note that when one looks at the original output of the
spectopo function it is already "logged" in the way that you see in the
spectral plots (the main eeglab spectral plots do not show absolute power,
so that's is easier to see power in higher frequency bands).  Som try not
using the second part of Makoto's code, the part which transforms the logged
spectopo output back to absolute power.

For comparing between bands, a method often used is computing relative
power,

If you have not had a chance to, remember to review high-quality EEG
articles that do analysis simialr to the ones you are trying (comparing
"strength" of bands between "conditions")

Note that when one compares between different conditions, there should be
about the same amount of data/time/epochs per participant and per condition,
and it should all be cleaned/preprocessed correctly before computing
metrics.

ps. dominant band is usually alpha - it predominates in the EEG signal,
althgouh all other bands are important too.

****************************************************

On Mon, Sep 24, 2018 at 1:14 PM Vibra Lab <labvibra at gmail.com> wrote:

> Let me describe my study. A person is given 4 sections (in total around 60
> minutes for all sections for each participant). For each section,
> participant is given various questions that they need to answer. i.e. it's
> like a conversation. Due to the nature of the experiment, the length of
> each section for each participant varies. For instance, for section *one *(for
> 4 participants) there are *4 different lengths* of time window (10
> minutes, 15 minutes, 20 minutes, and 25 minutes). Due to that, I have some
> questions
>
> I tried to look at the archives but it seems that there is no answer for
> the problem that I have.
>
> *1. *How can I find what is the most dominant frequency (eg. theta,
> alpha) for each section ?
>
> I can plot all types of band via (*Plot  > Channel spectra and map*) by
> putting *6, 10, 22, 30 in* column *Frequencies to plot as scalp map* for
> instance. But, how do I choose what is the most dominant frequency (out of
> 4 frequencies in the example above) during specific period / section ? So
> that, I can only show one brain figure (eg. brain map figure with 10 Hz)
> with statistical consideration.
>
> *2.* I am trying to use the code that is made by Makoto, found it in the
> archive , but it seems not working to determine which frequency band that
> is dominant in a certain period. I tried the code in different files, but
> it showed that low frequency bands have always higher mean amplitude than
> high frequency bands (eg. delta's mean amplitude is higher than theta's
> one). Why is it so ?
>
> Here's the code from Makoto to find mean amplitude
> %%%%%%%%%%%%%%%%%%%
> % 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))
> %%%%%%%%%%%%%%%%%
>
> Thank you for your help
>
> Best,
> Ihshan
>
> if I want to show only one band, considering that is the most dominant or
> prominent during specific period,
> _______________________________________________
> 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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20180924/f1116cea/attachment.html>