The document provides information about electroencephalography (EEG), including a description of EEG, the international 10-20 system for electrode placement, different EEG rhythms and waveforms, techniques for recording EEG signals, and methods for preprocessing EEG data such as removing baseline, re-referencing, resampling, filtering power line interference, and detecting bad electrodes. It also includes example code for performing some of these preprocessing steps.
The document provides information about electroencephalography (EEG), including a description of EEG, the international 10-20 system for electrode placement, different EEG rhythms and waveforms, techniques for recording EEG signals, and methods for preprocessing EEG data such as removing baseline, re-referencing, resampling, filtering power line interference, and detecting bad electrodes. It also includes example code for performing some of these preprocessing steps.
The document provides information about electroencephalography (EEG), including a description of EEG, the international 10-20 system for electrode placement, different EEG rhythms and waveforms, techniques for recording EEG signals, and methods for preprocessing EEG data such as removing baseline, re-referencing, resampling, filtering power line interference, and detecting bad electrodes. It also includes example code for performing some of these preprocessing steps.
The document provides information about electroencephalography (EEG), including a description of EEG, the international 10-20 system for electrode placement, different EEG rhythms and waveforms, techniques for recording EEG signals, and methods for preprocessing EEG data such as removing baseline, re-referencing, resampling, filtering power line interference, and detecting bad electrodes. It also includes example code for performing some of these preprocessing steps.
Download as PPT, PDF, TXT or read online from Scribd
Download as ppt, pdf, or txt
You are on page 1/ 35
Sudan University Of Science And Technology
Biomedical Engineering Department
Presentation About :
Electroencephalography (EEG) Prepared by: Hamida Tharwat Doaa Mohamed Reham Alhadi Mihad Modather Saraa Ali Afraa Eltyeb Leena Gamal Introduction :
The Electroencephalography(EEG) :
non-invasive technique in which the electrical activity of the brain is
recorded from the scalp by means of electrodes. The number of electrodes used can range from 32to 256. The resulting encephalogram presents the changes in electrical activity as a function of time. EEG Rhythms and Waveforms:
Delta rhythm, <4 Hz. The delta rhythm is typically encountered during deep sleep and has a large amplitude. It is usually not observed in the awake, normal adult, but is indicative of, e.g., cerebral damage or brain disease (encephalopathy).
Theta rhythm, 4 - 7 Hz. The theta rhythm occurs during drowsiness
and in certain stages of sleep. Alpha rhythm, 8-13 Hz. This rhythm is most prominent in normal subjects who are relaxed and awake with eyes closed; the activity is suppressed when the eyes are open. The amplitude of the alpha rhythm is largest in the occipital regions.
Beta rhythm, 14-30 Hz. This is a fast rhythm with low
amplitude, associated with an activated cortex and which can be observed, e.g., during certain sleep stages. The beta rhythm is mainly observed in the frontal and central regions of the scalp.
Gamma rhythm, >30 Hz. The gamma rhythm is related to a
state of active information processing of the cortex. Using an electrode located over the sensorimotor area and connected to a high-sensitivity recording technique, the gamma rhythm can be observed during finger Movements. RECORDING TECHNIQUE INTERNATIONAL 10-20 SYSTEM EEG SIGNAL FROM PATIENT 1.Baseline of EEG signal pop_rmbase( ) function [EEG, com] = pop_rmbase( EEG, timerange, pointrange); com =''; if isempty(EEG(1).data) disp('pop_rmbase(): cannot remove baseline of an empty dataset'); return; end; if nargin < 1 help pop_rmbase; return; end; if nargin < 2 & EEG(1).trials > 1 timerange = [num2str(EEG(1).xmin*1000) num2str(EEG(1).xmax*1000)]; % whole epoch latency range fprintf('pop_rmbase(): using whole epoch as baseline.\n'); elseif nargin < 2 & EEG(1).trials == 1 if isempty(result), return; end; if ~isempty(usrdat) ,return; end; % ----------------- if numel(result) < 2 | ((isempty(result{1}) | strcmp(result{1},'[]') ) ... & (isempty(result{2}) | strcmp(result{2},'[]'))) timerange = [num2str(EEG(1).xmin*1000) num2str(EEG(1).xmax*1000)]; % whole epoch latency range fprintf('pop_rmbase(): using whole epoch as baseline.\n');
if ~isempty(timerange) com = sprintf('%s = pop_rmbase( %s, [%s]);', inputname(1), inputname(1), ... num2str(timerange)); else com = sprintf('%s = pop_rmbase( %s, [], %s);', inputname(1), inputname(1), ... vararg2str({pointrange})); end; return; RE-REFERENCING EEG SIGNAL Re-referencing only applies to EEG data, MEG is reference-free. In EEG, voltages recorded at each electrode are relative to voltages recorded at other electrodes. Theoretically, a reference could be anywhere but the reference needs to be carefully chosen because any activity in the reference electrode will reflected in the activity at other electrodes. Often, the mastoids are chosen as reference electrodes, because while being close in distance to the electrodes, they record less signal from the brain RE-REFERENCE CODE data_out = ( data_in - ones( n_channels, 1 ) * ref_est' ) ; if exist( 'nm' ) figure(1);clf; subplot(3,2,1),plot( taxis, ref_est, '-k' ); title( 'reference signal estimate' ); subplot(3,2,2),plot( taxis, err, '-k+' ); , title( [{'robust estimation error (squared & normalized)'},{''}] ); xlabel( 'time' ), ylabel( 'error' ) subplot(3,2,4),imagesc( data_out ); c = caxis; colorbar, title( 'rCARIII data' ) subplot(3,2,3),imagesc( taxis, 1:n_channels, data_in ); colorbar, title( 'data' ); caxis( c ); subplot(3,2,5), plot( taxis, nn_ref_est, 'k+' ), colorbar, title( 'Number of Used Channels' ), set(gca,'ylim',[-1 20] ) subplot(3,2,6), plot( taxis, used_iter, 'k+' ), colorbar, title( 'Used Iteration' ), %set(gca,'xlim',[1 120] ) %print -depsc2 ../out/rCARII.eps, close(1); print( nm, '-dpsc2', '-append' ), close(1); end if( 0 ) keyboard figure(2);clf; subplot(1,2,1), plot( taxis, median( data_in ), '-k' );title('median datain') subplot(1,2,2), imagesc( data_in - ( ones( 19, 1 ) * median( data_in ) ) ), title( 'median re-ref data' ) print( nm, '-dpsc2', '-append' ), close(2); end % if( 0 )
data_out = data_out'; end Graphic: RE-SAMPLING
It is to decrease the size of your datafile and
facilitate quicker processing of your data, you should resample your data according to the Nyquist rule, the sampling frequency must be at least twice the highest frequency that we wish to analyse. If we are carrying out signal with 30Hz. That means, a resampling frequency would be 60Hz. RESAMPLING CODE pop_resample() - resample data (pop up window). Usage: >> [OUTEEG] = pop_resample( INEEG ); % pop up interactive window >> [OUTEEG] = pop_resample( INEEG, freq);
>>. Inputs: INEEG
input data freq - frequency to resample (Hz
Outputs: OUTEEG - output data FILTERING
A band-pass filter (also spelled band pass) is a device that passes
frequencies within a certain range and rejects (attenuates) frequencies outside that range .
POP_EEGFILTNEW: If nargin < 1
help pop_eegfiltnew
return
end
if isempty (EEG. data)
error (cannot filter empty dataset.);
End if nargin < 2 geometry = {[ 3, 1}, [3, 1], [3, 1], 1, 1, 1, 1}; geomvert = [ 1 1 1 2 1 1 1]; result = inputgui ( geometry , geometry, geomvert , geomvert , title , filter the data pop_eegfiltnew, helpcom , pophelp ) if isempty (result) , return; end locutoff = str2num (result {1}) ; hicutoff = str2num (result {2}); filttorder = str2num (result {3}); revfilt = result {4}; minphase = result {5}; plotfreq = result {6}; usefft = []; Else if nargin < 3 hicutoff = []; if nargin < 4 filtorder = []; end if nargin < 5 || isempty(revfilt) revfilt = 0 ; end if nargin < 6 usefft == 1 end if nargin < 7 ||isempty (plotfreqz) plotfreqz = 0 end if nargin < 8 ||isempty (minphase) minphase = 0; end if locutoff == 0,loucytoff = []; end if if locutoff == 0,loucytoff = []; end if hicutoff == 0, hicutoff = [] ; end if isempty(hicutoff) hicutoff = locutoff ; locutoff = []; revfilt = ~revfilt ; end edgArray = sort([locutoff hicutoff )]; if isempty (eddgArray) error (Not enough input regrument .); end if any (edgArray < 0 |edgeArray >= fNquist ) error (cutoff freqency out of range); end if ~isempty (filtorder) && (filtorder < 2|| mod (filtorder, 2) ~=0) end maxTBWArray = edgArray ; if revfilt == 0 maxTBWArray (end) = fNYquist edgeArray (end); else if length (edgeArray) l 2 ; end maXDF = min (maxTBWArray); if isempty (filtorder) if ervfilt == 1 df = min ([maxDF * TRANSWIDTHRATIO 2]) maxDF ]; else df = min ([max([edgeArray(1)* TRANSWIDTHRATIO 2 ])MAXdf]); END POWER INTERFERENCE The standard frequency in North American is 60 HZ , while 50 HZ in rest of the world. So here in Sudan it is 50Hz. The power line interface (50/60 Hz) is the main source of noise in most of the bio-electric signal. We will use a Notch filter. A notch filter is a band-stop filter with a narrow stopband (high Q factor). The quality factor of a notch filter is, Q= (f2 - f1)/fcenter. The center frequency is the center frequency of the stopband for a notch filter. It is the also referred to as the null frequency or the notch frequency. FILTERING 50HZ USING A NOTCH FILTER IN MATLAB : fs = 2000; %#sampling rate f0 = 50; %#notch frequency fn = fs/2; %#Nyquist frequency freqRatio = f0/fn; %#ratio of notch freq. to Nyquist freq.
%#filter signal x y = filter(b,a,x); DETECTION several methods can be used to detect bad electrodes : 1- Visual detection : To get a general idea of the quality of your data it is always good to look at it. To visualise your data : >> Plot >> Channel data (scroll) If you are visualising your data before filtering, you may want to remove the 0Hz offset. In the eegplot window >> Display >> Remove DC offset To visualise the data as a butterfly, select the Stack option. This is a good way of checking for bad electrodes and for periods of noisy data. - 2-Statistical measures :
Statistical measures of the behavior of signals for each
electrode and analysis of their spectral properties . EEGLAB provides a function for the automatic detection of bad electrodes in data using : - Kurtosis : A signal with normal distribution has a kurtosis value near to 0. The kurtosis value is expressed as z-score. The default z-score limit is 5 - Probability : This quantifies the time course of the electrode signals . noisy electrodes unusual time courses. - Spectral measures : Reveal electrodes contaminated by noise that has a characteristic frequency range. The function PSD_calc(), calculates the Power Spectral Density (PSD) for each electrode. The user can identify potential bad electrodes from PSD plot; bad electrodes are characterised by : - Greater activity in the low frequency range or - A lower amplitude at all frequencies than the other electrodes . MATLAB CODE function [k] = kurt(data) [r,c]=size(data); if r==1, kdata = data'; % if a row vector, make into a column vector r = c; else kdata = data; end % fprintf ('size of k data = [% d, % d] \ n', size(kdata,1),size(kdata,2)); mn = mean( k data); % find the column means diff = kdata -ones(r,1)* mn; % remove the column means dsq = diff.*diff; % square the data
k = (sum(dsq.*dsq)./std(kdata).^4)./r - 3; pop_rejchan function :
pop_rejchan() - reject artifacts channels in an EEG dataset using joint
Inputs: 'elec' - [n1 n2 ...] electrode number(s) to take into consideration for rejection 'threshold' - [max] absolute thresold or activity probability limit(s) (in std. dev.) if norm is 'on'. 'measure' - ['prob'|'kurt'|'spec'] compute probability 'prob', kurtosis 'kurt or spectrum 'spec' for each channel. Default is 'kurt'. 'norm' - ['on'|'off'] normalize measure above (using trimmed normalization as described in the function jointprob() and rejkurt(). Default is 'off'. 'precomp' - [float array] use this array instead of computing the 'prob' or 'kurt' measures. 'freqrange' - [min max] frequency range for spectrum computation. Default is 1 to sampling rate. divided by 2.
The average of the log spectral power is computed over the frequency range of interest. CONCLUSION
We applied different mat lab functions on
(EEG) signal in order to DE noise the signal to be used in right diagnosis.
