Generate K randomly rearranged ECG signals from the extracted beats, where K can be any large integer.

ECG Signal Information from "data_116_N_V_R.mat":

load data_116_N_V_R.mat

openfig(illustration.fig)

whos -file data_116_N_V_R.mat

  Name                 Size              Bytes  Class          Attributes

  ECGsignal       649783x1             5198264  double                   
  RRinterval        2410x1               19280  double                   
  Rpeak             2411x1               19288  double                   
  annotation            -                 2779  categorical              

- ECGsignal: MITDB ECG signal sampled at 360 Hz.

- Rpeak: Locations of R-peaks in the ECG signal, provided as sample indices.

- RRinterval: Intervals in samples between consecutive R-peaks based on the Rpeak array.

- annotation: Beat-type annotation centered around each R-peak. In this test data, each beat is labeled as either "N" or "V," but it can represent various other types.

Beat Definition:

Please, refer to the attached figure (illustration.fig).

openfig('illustration.fig');

Each beat is defined by two portions extending around a central R-peak:

- Left portion: 40% of the preceding RR interval, to the left of the R-peak.

- Right portion: 60% of the succeeding RR interval, to the right of the R-peak.

I want to segment every beat and then randomly rearrange segmented beats to form new ECG signals.

Discontinuity Adjustment for Rearranged Beats:

Due to the random reordering, there will be discontinuities between consecutive beats. To minimize these:

- Adjust the first sample of each right-side beat by +1 or -1 relative to the last sample of the preceding left-side beat.

- +1 adjustment if the second sample of the right-side beat is greater than the first, then shift the rest smaples of the beat accordingly.

- -1 adjustment if the second sample of the right-side beat is smaller than the first, then shift the rest smaples of the beat accordingly.

Goal:

- Generate K randomly rearranged ECG signals from the extracted beats, where K can be any large integer (let's say e.g., 10).

- I want to have updated RR-intervals (RRinterval), R-peaks (Rpeak), and annotations in accordance with newly generated ECG signals.

[Image Analyst]

Is this your homework? Have you segmented the individual beats yet? Do you not know how to use randperm to randomly order things (once they've been segmented/isolated/extracted)?

[Mibang]

Please, do not mention any unnecessary. Thank you.

[Voss]

@Mibang: Please answer @Image Analyst's questions. In general, please share any progress you have made so far on this project.

[Mibang]

Of course, "randperm" is a necessary tool, thank you, but it is none of question becuase it is obvious to use it for shuffling the beats. I am wokring on it and I will keep you posted if I can solve it.

[Mibang]

I revised variables name because they were not updated at the initial uploading.

[Voss]

Have you segmented the individual beats yet?

[Mibang]

No, I am working on it and got only sudo code yet.

Answer 1

[Image Analyst]

Because each segment may be a different number of samples (elements) you will have to first go through the signal and store the individual segments in a cell array, say it's called allBeats.

See the FAQ: What is a cell array

Then you can shuffle the order

randomOrder = randperm(numel(allBeats))
allBeats = allBeats(randomOrder);

Then you can concatenate all the cells together into a single double array.

allBeats = {[1,222,3], [40,55,66,77]} % Sample data with y-values of 2 beats

allBeats = 1x2 cell array
    {[1 222 3]}    {[40 55 66 77]}

signal = cell2mat(allBeats) % Create single vector signal.

signal = 1×7

1 222 3 40 55 66 77

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

[Mibang]

You also need to find updated RR-interval and R-peaks of new signals. And "annotation" updated as well.