MATLAB: filter noisy EKG signal - matlab

Question

What is the best filter to use to remove noise from an ECG signal with matlab?

Answer 1

If you have access to the Signal Processing Toolbox, then check out the Savitzky-Golay filter, namely the function sgolay. There's an accompanying demo, just run sgolaydemo.
The following is an example to show the various ways you can apply filtering and de-noising to a signal. Note some of these functions requires certain toolboxes to be present:
% load ecg: simulate noisy ECG
Fs=500;
x = repmat(ecg(Fs), 1, 8);
x = x + randn(1,length(x)).*0.18;
% plot noisy signal
figure
subplot(911), plot(x), set(gca, 'YLim', [-1 1], 'xtick',[])
title('noisy')
% sgolay filter
frame = 15;
degree = 0;
y = sgolayfilt(x, degree, frame);
subplot(912), plot(y), set(gca, 'YLim', [-1 1], 'xtick',[])
title('sgolayfilt')
% smooth
window = 30;
%y = smooth(x, window, 'moving');
%y = smooth(x, window/length(x), 'sgolay', 2);
y = smooth(x, window/length(x), 'rloess');
subplot(913), plot(y), set(gca, 'YLim', [-1 1], 'xtick',[])
title('smooth')
% moving average filter
window = 15;
h = ones(window,1)/window;
y = filter(h, 1, x);
subplot(914), plot(y), set(gca, 'YLim', [-1 1], 'xtick',[])
title('moving average')
% moving weighted window
window = 7;
h = gausswin(2*window+1)./window;
y = zeros(size(x));
for i=1:length(x)
for j=-window:window;
if j>-i && j<(length(x)-i+1)
%y(i) = y(i) + x(i+j) * (1-(j/window)^2)/window;
y(i) = y(i) + x(i+j) * h(j+window+1);
end
end
end
subplot(915), plot( y ), set(gca, 'YLim', [-1 1], 'xtick',[])
title('weighted window')
% gaussian
window = 7;
h = normpdf( -window:window, 0, fix((2*window+1)/6) );
y = filter(h, 1, x);
subplot(916), plot( y ), set(gca, 'YLim', [-1 1], 'xtick',[])
title('gaussian')
% median filter
window = 15;
y = medfilt1(x, window);
subplot(917), plot(y), set(gca, 'YLim', [-1 1], 'xtick',[])
title('median')
% filter
order = 15;
h = fir1(order, 0.1, rectwin(order+1));
y = filter(h, 1, x);
subplot(918), plot( y ), set(gca, 'YLim', [-1 1], 'xtick',[])
title('fir1')
% lowpass Butterworth filter
fNorm = 25 / (Fs/2); % normalized cutoff frequency
[b,a] = butter(10, fNorm, 'low'); % 10th order filter
y = filtfilt(b, a, x);
subplot(919), plot(y), set(gca, 'YLim', [-1 1])
title('butterworth')

Answer 2

Two filter design tools/demos that you may want to check out:
FDATool in the Signal Processing Toolbox (if you have access to it).
Analog Filter Design Toolbox from James Squire on the MathWorks File Exchange. There appear to be simulations for fitering EKG data included in the toolbox.
These should give you a chance to try out different filters and filter parameters to see how they perform with EKG/ECG data.