Amplow_amphigh

Help ft_freqsimulation

% method 'amplow_amphigh
%   signal is build up of 5 components; s1, s2, s3, s4 and noise.
%     s1: first frequency
%     s2: second frequency
%     s3: DC shift of s1 and s2, should have frequency of 0
%     s4: amplitude modulation (AM), frequency of this signal should be lower than s1 and s2
% and the output consists of the following channel
%     1st channel: mixed signal = (s4 + s3)*s1 + (s4 + s3)*s2 + noise,
%     2nd channel: s1
%     3rd channel: s2
%     4th channel: s3
%     5th channel: noise
%     6th channel: s4
%     7th channel: mixed part 1: (s4 + s3)*s1
%     8th channel: mixed part 2: (s4 + s3)*s2

Simulating the data

cfg = [];
cfg.method   = 'amplow_amphigh';
cfg.fsample  = 1000;
cfg.trllen   = 10;
cfg.numtrl   = 10;
cfg.output   = 'all';
% first frequency
cfg.s1.freq  = 6;
cfg.s1.phase = 0;
cfg.s1.ampl  = 1;
% second frequency
cfg.s2.freq  = 20;
cfg.s2.phase = 0;
cfg.s2.ampl  = 1;
% DC shift of s1 and s2
cfg.s3.freq  = 0;
cfg.s3.phase = 0;
cfg.s3.ampl  = 1; %determines amount of modulation, should be at least s4.ampl
% amplitude modulation (AM)
cfg.s4.freq  = 1; %frequency of this signal should be lower than s1 and s2
cfg.s4.phase = -1*pi;
cfg.s4.ampl  = 1;
% noise
cfg.noise    = 0.1;

data = ft_freqsimulation(cfg);

What does the signal look like?

figure;
sel = 1:2000;
subplot(3,3,1); plot(data.trial{1}(1,sel)); title(data.label{1})
subplot(3,3,2); plot(data.trial{1}(2,sel)); title(data.label{2})
subplot(3,3,3); plot(data.trial{1}(3,sel)); title(data.label{3})
subplot(3,3,4); plot(data.trial{1}(4,sel)); title(data.label{4})
subplot(3,3,5); plot(data.trial{1}(5,sel)); title(data.label{5})
subplot(3,3,6); plot(data.trial{1}(6,sel)); title(data.label{6})
subplot(3,3,7); plot(data.trial{1}(7,sel)); title(data.label{7})
subplot(3,3,8); plot(data.trial{1}(8,sel)); title(data.label{8})
print -dpng amplow_amphigh_fig1.png

% show powerspectrum simulated data
cfg = [];
cfg.method    = 'mtmfft';
cfg.channel   = 'mix';
cfg.output    = 'pow';
cfg.taper     = 'hanning';
cfg.foilim    = [2 60];

fft_data = ft_freqanalysis(cfg,data);
figure; ft_singleplotER([],fft_data);
print -dpng amplow_amphigh_fig2.png

Analysis Methods

Calculate correlation between power of power

% mtmconvol
cfg = [];
cfg.method    = 'mtmconvol';
cfg.channel   = 'mix';
cfg.output    = 'pow';
cfg.taper     = 'hanning';
cfg.foi       = 2:2:60;
cfg.toi       = data.time{1}(3001:7000); %power is calculated at every sample
cfg.t_ftimwin = 4./cfg.foi; %timewindow used to calculated power is 4 cycles long and therefore differs over frequencies
cfg.keeptrials = 'yes';

freq1 = ft_freqanalysis(cfg,data);
figure;
imagesc(freq1.time, freq1.freq, squeeze(freq1.powspctrm(1,1,:,:))); axis xy
print -dpng amplow_amphigh_fig3.png

figure; plot(freq1.time,squeeze(freq1.powspctrm(1,1,3,:)))   % at 6 Hz
hold on
plot(freq1.time,squeeze(freq1.powspctrm(1,1,10,:)),'r')      % at 20 Hz
hold on
plot(data.time{1}(3001:7000),data.trial{1}(6,3001:7000),'g') % lowest -> original amplitude modulation
legend('power at 6 Hz','power at 20 Hz','s4 (AM)','location','Best')
print -dpng amplow_amphigh_fig4.png

% calculate covariance with ft_timelockanalysis
cfg = [];
cfg.covariance         = 'yes';
cfg.keeptrials         = 'no';
cfg.removemean         = 'yes';
timelock = ft_timelockanalysis(cfg,freq1);
freqlabel = [2:2:60];
figure; imagesc(freqlabel,freqlabel,timelock.cov)
title('covariance')
colorbar
axis xy
print -dpng amplow_amphigh_fig5.png

% calculate correlation
cov = timelock.cov; % all trials
d = sqrt(diag(cov)); % SD, diagonal is variance per channel
r = cov ./ (d*d');
figure; imagesc(freqlabel,freqlabel,r)
title('correlation')
colorbar
axis xy
print -dpng amplow_amphigh_fig6.png

Coherence between amplitude envelopes (by Hilbert transform)

cfg = [];
cfg.hilbert = 'abs';
cfg.channel = 'mix';
cfg.bpfilter = 'yes';
cfg.bpfreq   = [4 8]; %do not bandpass to tight, then amplitude modulation is lost
data_bp6 = ft_preprocessing(cfg, data);
data_bp6.label = {'mix@6Hz'};

cfg.bpfreq   = [18 22];
data_bp20 = ft_preprocessing(cfg, data);
data_bp20.label = {'mix@20Hz'};

data_bp = ft_appenddata([], data_bp6, data_bp20);
figure
plot(data_bp.time{1}, data_bp.trial{1}); legend(data_bp.label)
print -dpng amplow_amphigh_fig7.png

cfg = [];
cfg.method    = 'mtmfft';
cfg.output    = 'powandcsd';
cfg.taper     = 'hanning';
cfg.foilim    = [0 60];
cfg.keeptrials = 'no';
cfg.channelcmb = {'all', 'all'};

freq3 = ft_freqanalysis(cfg, data_bp);
freq3_coh = ft_freqdescriptives([], freq3);

figure
plot(freq3_coh.freq, freq3_coh.cohspctrm);
title('coherence between amplitude envelopes at 6 and 20 Hz');
xlabel('frequency');
ylabel('coherence');
print -dpng amplow_amphigh_fig8.png

% zoom in
xlim([0 10]);
print -dpng amplow_amphigh_fig9.png