I'm new to Matlab. I'm doing ECG signal analysis. I have ECG data in .mat format and I have used pan_tompkin algorithm to calculate R peaks of the ECG signal, then I have calculated the RR interval from the output index of R peaks, so I get RR_interval vector (1x7831 double). Once all R-peaks were detected, the full RRI signal is obtained. I need to create a sliding window with one heartbeat overlapping width of N=158 samples corresponding to 3 minutes to extract some features from RRI, For example, for each part of the windowed signal 9 features need to be calculated(6 time-domain and 3 frequency domain features). PLEASE HELP, I don't know how to start doing this in Matlab.
Dear Huda,
the easiest way (not the nicest) would be to program a loop. Something like this:
displacement=1;%here you can user larger values to reduce the overlap between windows; if you chose 1 the window is shifted by one beat only
currentCalculation=1;
for i=1:displacement:numel(RR)-N
currentRRs=RR(i:i+N-1);
feature(currentCalculation)=std(currentRRs);%now you do some Proecesing with your currentRRs and store the calcualted features in a matrix/vector; here as an example the standard deviation is calculated
currentCalculation=currentCalculation+1;
end
However, you should be careful, as a fixed nuimber of beats (158) won't result in equal lengths of considered RR intervals. Moreover, to calculate HRV features it might be necessary to filter the RR intervals to remove artifacts and non-sinus events.
Greetings, Sebastian
Dear Huda,
the easiest way (not the nicest) would be to program a loop. Something like this:
displacement=1;%here you can user larger values to reduce the overlap between windows; if you chose 1 the window is shifted by one beat only
currentCalculation=1;
for i=1:displacement:numel(RR)-N
currentRRs=RR(i:i+N-1);
feature(currentCalculation)=std(currentRRs);%now you do some Proecesing with your currentRRs and store the calcualted features in a matrix/vector; here as an example the standard deviation is calculated
currentCalculation=currentCalculation+1;
end
However, you should be careful, as a fixed nuimber of beats (158) won't result in equal lengths of considered RR intervals. Moreover, to calculate HRV features it might be necessary to filter the RR intervals to remove artifacts and non-sinus events.
Greetings, Sebastian
I agree with Sebastian - if you are new to Matlab, using a FOR-loop is the most straight forward (and also most flexible) thing to do - especially if you are using several features.
If you use only few features, vectorized code might be faster at the cost of readability (and you have to understand how to do that - which is not straight forward for a beginner). Since R2016a there are also some built-in moving functions (see e.g. MOVMEAN, MOVSTD). I also have written a vectorized function to compute (nearly) arbitrary functions in a moving window (see last link).
Greetings, David
http://ch.mathworks.com/help/matlab/ref/movstd.html
http://ch.mathworks.com/help/matlab/ref/movmean.html
http://ch.mathworks.com/matlabcentral/fileexchange/58949
Write a for loop, its easy.
For the vectorized code, you can use filter() and pass your own function handle.
Thank you all for help. I'm trying to calculate some features over RR interval (time intervals between consecutive heart beats) using sliding window, for example, I'm using the following code for calculating the Standard deviation of RRI (SDNN):
h = 1; % window size of 3 overlapping 1
load('RRInterval');
x = R_R; % input signal is RRinterval
N = length(x); % length of the signal
SDNN = ones(1, N); % array for output for SDNN: Standard deviation of RRI.
for i = 1 : N
% calculate standard deviation using the built-in std command
% for the current window
SDNN(i) = std(x(max(1, i - h) : min(N, i + h)));
end
Do you think this code is correct? I'm confused between window size and overlapping, I need to use sliding window with one heartbeat overlapping and size or width of 3 minutes in which I can calculate the feature that I need from each part of the windowed.
Dear Huda,
You can often refer to Matlab's help. Please feel free to look at the following link to understand the Sliding Window Method over a signal. This Matlab example illustrates the principle of this process.
http://www.mathworks.com/help/dsp/ug/sliding-window-method-and-exponential-weighting-method.html?searchHighlight=slide%20window&s_tid=doc_srchtitle
Dear Sebastian Zaunseder
Why increasing the value of "displacement" decreases the overlap between the window?
Could you please explain the below line that you put in your answer. Thanks
"displacement=1;%here you can use larger values to reduce the overlap between windows; if you chose 1 the window is shifted by one beat only "
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R