typically the spectral power density (uV2/Hz) estimates for alpha (8-13Hz) at each of the homologous channels are computed using overlapping (50-75%) hanning or hamming windowed chunks and log_e transformed before subtraction. different results will likely arise as a function of the reference montage (avg, ears, cz, mastoids).
<br><br>
see also the articles @ <br>
<a href="http://psyphz.psych.wisc.edu/%7Eshackman/EEG_ERP_Methodology.htm#Anterior_Cerebral_Asymmetry">http://psyphz.psych.wisc.edu/%7Eshackman/EEG_ERP_Methodology.htm#Anterior_Cerebral_Asymmetry</a><br>
and<br>
<a href="http://psyphz.psych.wisc.edu/%7Eshackman/EEG_ERP_Methodology.htm#Average_Reference">http://psyphz.psych.wisc.edu/%7Eshackman/EEG_ERP_Methodology.htm#Average_Reference</a><br>
and<br>
<a href="http://psyphz.psych.wisc.edu/web/pubs/2004/Electrode_density.pdf">http://psyphz.psych.wisc.edu/web/pubs/2004/Electrode_density.pdf</a><br><br>i took a look at the spectopo function it allows overlapping and logging...i'm not certain how one would do the windowing, although it looks (see below) that the default behavior of the pwelch function is to use a hamming window.
<br><br>% spectopo() - Plot the mean log spectrum of a set of data epochs at all channels <br>% as a bundle of traces. At specified frequencies, plot the relative <br>% topographic distribution of power. If available, uses pwelch() from
<br>% the Matlab signal processing toolbox, else the EEGLAB spec() function.<br>% Plots the mean spectrum for all of the supplied data, not just<br>% the pre-stimulus baseline.<br>% Usage:
<br>% >> spectopo(data, frames, srate);<br>% >> [spectra,freqs,speccomp,contrib,specstd] = ...<br>% spectopo(data, frames, srate, 'key1','val1', 'key2','val2' ...);
<br>% Inputs:<br>% data = If 2-D (nchans,time_points); % may be continuous = a single epoch,<br>% else a set of concatenated data epochs. Else, a 3-D set of data <br>% epochs (nchans,frames,epochs)
<br>% frames = frames per epoch {default|0 -> data length}<br>% srate = sampling rate per channel (Hz)<br>%<br>% Optional inputs:<br>% 'freq' = [float vector (Hz)] vector of frequencies at which to plot power
<br>% scalp maps, or else a single frequency at which to plot component <br>% contributions at a single channel (see also 'plotchan').<br>% 'chanlocs' = electrode locations file or
EEG.chanlocs structure. For format, see<br>% >> topoplot example<br>% 'limits' = axis limits [xmin xmax ymin ymax cmin cmax] set x y and color axis<br>% limits. May omit final values or use NaN's
<br>% Ex: [0 60 NaN NaN -10 10], [0 60], ...<br>% Default color limits are symmetric around 0 and are different <br>% for each scalp map {default|all NaN's: from data}<br>
% 'title' = [quoted string] plot title {default: none}<br>% 'freqfac' = [integer] ntimes to oversample -> frequency resolution {default: 2}<br>% 'nfft' = [integer] length to zero-pad data to. Overwrites 'freqfac' above.
<br>% 'winsize' = [integer] window size in data points {default: from data}<br>% 'overlap' = [integer] window overlap in data points {default: 0}<br>% 'percent' = [float 0 to 100] percent of the data to sample for computing the
<br>% spectra. Values < 100 speed up the computation. {default: 100}.<br>% 'freqrange' = [min max] frequency range to plot. Changes x-axis limits {default: <br>% 1 Hz for the min and Nyquist (srate/2) for the max. If specified
<br>% power distribution maps are plotted, the highest mapped frequency <br>% determines the max freq}.<br>% 'reref' = ['averef'|'off'] convert input data to average reference
<br>% {default: 'off'}<br>% 'mapnorm' = [float vector] If 'data' contain the activity of an independant <br>% component, this parameter should contain its scalp map. In this case
<br>% the spectrum amplitude will be scaled to component RMS scalp power.<br>% Useful for comparing component strengths {default: none}<br>% 'boundaries' = data point indices of discontinuities in the signal {default: none}
<br>% 'plot' = ['on'|'off'] 'off' -> disable plotting {default: 'on'}<br>% 'rmdc' = ['on'|'off'] 'on' -> remove DC {default: 'off'}
<br>%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%<br><br><br>but<br><br><br>>> help pwelch<br> PWELCH Power Spectral Density estimate via Welch's method.<br> Pxx = PWELCH(X) returns the Power Spectral Density (PSD) estimate,
<br> Pxx, of a discrete-time signal vector X using Welch's averaged, <br> modified periodogram method. By default, X is divided into eight<br> sections with 50% overlap, each section is windowed with a Hamming
<br> window and eight modified periodograms are computed and averaged.<br><br><br>hth, alex shackman<br><br><div><span class="gmail_quote">On 8/8/07, <b class="gmail_sendername">Arnaud Delorme</b> <<a href="mailto:arno@salk.edu">
arno@salk.edu</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Dear Gregor,<br><br>first you need to compute alpha power. To do so may use the function
<br>spectopo or newtimef to compute the alpha power. For example for a<br>single channel on the left and on the right.<br><br>x = 10; % example of channel index on the left<br>y = 11; % example of channel index on the right
<br>[ spec_x freqs ] = spectopo(EEG.data(x,:), EEG.pnts, EEG.frames, 'plot',<br>'off');<br>[ spec_y freqs ] = spectopo(EEG.data(y,:), EEG.pnts, EEG.frames, 'plot',<br>'off');<br>[tmp alpha_ind] = min( abs(freqs-10)); % find closest frequency to alpha
<br>at 10 Hz<br>asymmetry = spec_x(alpha_ind) - spec_y(alpha_ind) % I think asymmetry is<br>about subtracting power but I might be wrong.<br><br>Hope this helps.<br>Best regards,<br><br>Arno<br><br><a href="mailto:gkohls@ukaachen.de">
gkohls@ukaachen.de</a> wrote:<br>> Dear all,<br>><br>> I collected resting EEG to analyze frontal Alpha asymmetry. Can anyone<br>> help me how to analyze alpha band and asymmetries with EEGlab? I tried<br>> but did not succeed.
<br>><br>> Many thanks in advance,<br>> Gregor<br><br>_______________________________________________<br>eeglablist mailing list <a href="mailto:eeglablist@sccn.ucsd.edu">eeglablist@sccn.ucsd.edu</a><br>Eeglablist page:
<a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu">eeglablist-unsubscribe@sccn.ucsd.edu
</a><br></blockquote></div><br><br clear="all"><br>-- <br>Alexander J. Shackman<br>Laboratory for Affective Neuroscience<br>Waisman Laboratory for Brain Imaging & Behavior<br>University of Wisconsin-Madison<br>1202 West Johnson Street
<br>Madison, Wisconsin 53706<br><br>Telephone: +1 (608) 358-5025<br>FAX: +1 (608) 265-2875<br>EMAIL: <a href="mailto:shackman@wisc.edu">shackman@wisc.edu</a><br><a href="http://psyphz.psych.wisc.edu/~shackman">http://psyphz.psych.wisc.edu/~shackman
</a>