Using a specific wavelet transform on new ECG signals

I'm working with ECG signals and am trying to use a wavelet technique to reduce some of the noise in various data sets. I was able to use a 'Continuous Wavelet 1-D using FFT' technique in the Wavelet Analyzer toolkit to achieve a good response (at least on one sample). I have exported the CWTFT struct to a variable called CWTS and want to apply the same wavelet function on a number of other pieces of data.

My overall goal would be to have a command or function that applies the same wavelet transform I generated to new input data.

I am somewhat of a novice in the wavelet / signal processing realm, so if there are smarter or better ways to approach this, please let me know. I tried several random wavelets and this option seemed to do the best for me.

How I generated the wavelet transformed data: 1-D/FFT -> 'dog', 6 parameter, linear analysis, linear synthesis, scale limited to first 16 options

Here are some of the details: CWTS Contents: CWTS.cfs, CWTS.scales, CWTS.frequencies, CWTS.omega, CWTS.meanSIG, CWTS.dt, CWTS.wav

Here are the pre / post transform plots of the data:

Pre-Transform/Raw Data

Post-Transform

Thanks for any help you guys can offer!

[farzana iqbal]

Hi, I am also working on a similar project. Would you mind sharing your model here or of you can email at [e-mail]

[Helen Nonyelu]

Hi can anyone suggest on how to denoise ecg signal using waveletneural network and also the code if possible

Answer 1

[Prateek Khandelwal]

Hi !

Wavelet Analyzer app is merely a front-end for easy accessibility to various algorithms under wavelet toolbox.

I would suggest you go through the documentation page on Generation of MATLAB code for 1-D decimated Wavelet Denoisising & Compression for basic steps that might be useful for your goal .

[Thomas Fogarty III]

thanks - that seems to be what I was looking for! I'll give it a shot and see if I can make it work!

Answer 2

[ziani]

Here is the complete error-free code for Wavelet Analysis of Physiologic Signals.

load mit200
figure
plot(tm,ecgsig)
hold on
plot(tm(ann),ecgsig(ann),'ro')
xlabel('Seconds')
ylabel('Amplitude')
title('Subject - MIT-BIH 200')
qrsEx = ecgsig(4560:4810);
fb = dwtfilterbank('Wavelet','sym4','SignalLength',numel(qrsEx),'Level',3);
psi = wavelets(fb);
figure
plot(qrsEx)
hold on
plot(-2*circshift(psi(3,:),[0 -38]),'r')
axis tight
legend('QRS Complex','Sym4 Wavelet')
title('Comparison of Sym4 Wavelet and QRS Complex')
hold off
wt = modwt(ecgsig,5);
wtrec = zeros(size(wt));
wtrec(4:5,:) = wt(4:5,:);
y = imodwt(wtrec,'sym4');
y = abs(y).^2;
[qrspeaks,locs] = findpeaks(y,tm,'MinPeakHeight',0.35,...
    'MinPeakDistance',0.150);
figure
plot(tm,y)
hold on
plot(locs,qrspeaks,'ro')
xlabel('Seconds')
title('R Peaks Localized by Wavelet Transform with Automatic Annotations')
plot(tm(ann),y(ann),'k*')
title('R peaks Localized by Wavelet Transform with Expert Annotations')
figure
plot(tm,ecgsig,'k--')
hold on
plot(tm,y,'r','linewidth',1.5)
plot(tm,abs(ecgsig).^2,'b')
plot(tm(ann),ecgsig(ann),'ro','markerfacecolor',[1 0 0])
set(gca,'xlim',[10.2 12])
legend('Raw Data','Wavelet Reconstruction','Raw Data Squared', ...
    'Location','SouthEast')
xlabel('Seconds')
[qrspeaks,locs] = findpeaks(ecgsig.^2,tm,'MinPeakHeight',0.35,...
    'MinPeakDistance',0.150);
% Load the ECG signal data
load('mit203.mat')
figure
plot(tm,ecgsig)
hold on
plot(tm(ann),ecgsig(ann),'ro')
xlabel('Seconds')
ylabel('Amplitude')
title('Subject - MIT-BIH 203 with Expert Annotations')
wt = modwt(ecgsig,5);
wtrec = zeros(size(wt));
wtrec(4:5,:) = wt(4:5,:);
y = imodwt(wtrec,'sym4');
y = abs(y).^2;
[qrspeaks,locs] = findpeaks(y,tm,'MinPeakHeight',0.1,...
    'MinPeakDistance',0.150);
figure
plot(tm,y)
title('R-Waves Localized by Wavelet Transform')
hold on
hwav = plot(locs,qrspeaks,'ro');
hexp = plot(tm(ann),y(ann),'k*');
xlabel('Seconds')
legend([hwav hexp],'Automatic','Expert','Location','NorthEast')
ecgmra = modwtmra(wt);
figure
plot(tm,ecgmra(5,:).^2,'b')
hold on
plot(tm,ecgmra(4,:).^2+0.6,'b')
set(gca,'xlim',[14.3 25.5])
timemarks = repelem(tm(ann),2);
N = numel(timemarks);
markerlines = reshape(repmat([0;1],1,N/2),N,1);
h = stem(timemarks,markerlines,'k--');
h.Marker = 'none';
set(gca,'ytick',[0.1 0.6]);
set(gca,'yticklabels',{'D5','D4'})
xlabel('Seconds')
title('Magnitude-Squared Level 4 and 5 Details')
load NIRSData
figure
plot(tm,[NIRSData(:,1) NIRSData(:,2)])
legend('Subject 1','Subject 2','Location','NorthWest')
xlabel('Seconds')
title('NIRS Data')
grid on
cwt(NIRSData(:,1),10,'bump')
figure
cwt(NIRSData(:,2),10,'bump')
[wcoh,~,F] = wcoherence(NIRSData(:,1),NIRSData(:,2),10);
figure
surf(tm,F,abs(wcoh).^2); view(0,90)
shading interp
axis tight
hc = colorbar;
hc.Label.String = 'Coherence';
title('Wavelet Coherence')
xlabel('Seconds')
ylabel('Hz')
ylim([0 2.5])
set(gca,'ytick',[0.15 1.2 2])
taskbd = [245 1702 2065 3474];
tvec = repelem(tm(taskbd),2);
yvec = [0 max(F)]';
yvec = reshape(repmat(yvec,1,4),8,1);
hold on
stemPlot = stem(tvec,yvec,'w--','linewidth',2);
stemPlot.Marker = 'none';
load dpoae
figure
plot(t.*1000,dpoaets)
xlabel('Milliseconds')
ylabel('Amplitude')
[dpoaeCWT,f] = cwt(dpoaets,2e4,'VoicesPerOctave',16);
helperCWTTimeFreqPlot(dpoaeCWT,t.*1000,f,...
    'surf','CWT of OAE','milliseconds','Hz')
[~,idx1230] = min(abs(f-1230));
cfsOAE = dpoaeCWT(idx1230,:);
plot(t.*1000,abs(cfsOAE))
hold on
AX = gca;
plot([25 25],[AX.YLim(1) AX.YLim(2)],'r')
plot([175 175],[AX.YLim(1) AX.YLim(2)],'r')
xlabel('msec')
title('CWT Coefficient Magnitudes')
frange = [1150 1350];
xrec = icwt(dpoaeCWT,[],f,frange);
figure
plot(t.*1000,dpoaets)
hold on
plot(t.*1000,xrec,'r')
AX = gca;
ylimits = AX.YLim;
plot([25 25],ylimits,'k')
plot([175 175],ylimits,'k')
grid on
xlabel('Milliseconds')
ylabel('Amplitude')
title('Frequency-Localized Reconstruction of Emission')
xdft = fft(xrec);
freq = 0:2e4/numel(xrec):1e4;
xdft = xdft(1:numel(xrec)/2+1);
figure
plot(freq,abs(xdft))
xlabel('Hz')
ylabel('Magnitude')
title('Fourier Transform of CWT-Based Signal Approximation')
[~,maxidx] = max(abs(xdft));
fprintf('The frequency is %4.2f Hz\n',freq(maxidx))