How to up-sample both signals and labels together?

I have ECG-signals and corresponding labels with the sampling rate of 250 Hz (attached).

Attached "sig.mat" is the ECG-signal and "labl.mat" is the labels, where five signals and five labels (5x1 cell).

I want to up-sample these signals and labels coherently (synchronized in time after upsampling) with 360 Hz.

I heard I can use "resample" function for the signal, but I don't know it for the labels.

I may use labeler app, but this is too slow.

I would appreciate it if anyone could help me on how to label automatically in accordance with the up-sampled signals.

Thanks,

Answer 1

[Chunru]

%websave("upsamp.mat", "https://www.mathworks.com/matlabcentral/answers/uploaded_files/1642771/upsamp.mat");
websave("upsamp_R.mat", "https://www.mathworks.com/matlabcentral/answers/uploaded_files/1642906/upsamp_R.mat");

load upsamp_R
fs = 250;
fsnew = 360;
for i =1:length(sig)
    % signal and label
    s = sig{i};
    l = labl{i};
    
    % categories
    c = categories(l);
    
    % original and new time
    t = (0:length(s)-1)/fs;
    tnew = t(1):(1/fsnew):t(end);
    
    % interpolation
    snew = interp1(t, s, tnew, 'linear');
    lnew = interp1(t, double(l), tnew, 'nearest');
    
    % convert back to categorical data
    lnew = categorical(lnew, 1:length(c), c);
    
    signew{i, 1} = snew;
    lablnew{i, 1} = lnew;
end
sig

sig = 5×1 cell array
    {[    -0.4456 -0.3906 -0.4456 -0.5006 -0.4731 -0.5555 -0.5830 -0.6380 -0.5280 -0.5280 -0.6105 -0.6380 -0.5280 -0.6105 -0.5555 -0.5830 -0.5280 -0.5280 -0.5830 -0.5555 -0.4731 … ] (1×5000 double)}
    {[-0.0973 -0.0211 0.0297 -0.0211 -0.0211 0.0805 0.1567 0.1821 0.1821 0.2329 0.3345 0.3345 0.3599 0.4107 0.4615 0.5123 0.5123 0.5631 0.6139 0.5885 0.5377 0.4869 0.5885 0.6139 … ] (1×5000 double)}
    {[    0.2383 0.3253 0.4124 0.4124 0.3834 0.3543 0.4414 0.4704 0.3834 0.4124 0.4704 0.5284 0.4994 0.4124 0.4124 0.4704 0.5284 0.4124 0.4124 0.4704 0.4704 0.3543 0.3253 0.3834 … ] (1×5000 double)}
    {[    -0.6125 -0.5854 -0.5854 -0.5311 -0.5311 -0.4768 -0.5854 -0.6397 -0.5311 -0.6125 -0.6397 -0.6397 -0.5583 -0.5583 -0.6397 -0.7211 -0.6397 -0.5854 -0.6668 -0.7754 -0.7211 … ] (1×5000 double)}
    {[    -0.3252 -0.2707 -0.2707 -0.3797 -0.3797 -0.3252 -0.2980 -0.3252 -0.3524 -0.2707 -0.2435 -0.2980 -0.4069 -0.3797 -0.2707 -0.3252 -0.3797 -0.3252 -0.2707 -0.2162 -0.2435 … ] (1×5000 double)}

signew

signew = 5×1 cell array
    {[      -0.4456 -0.4074 -0.4120 -0.4502 -0.4883 -0.4876 -0.4868 -0.5441 -0.5708 -0.5968 -0.6349 -0.5677 -0.5280 -0.5303 -0.5876 -0.6220 -0.6258 -0.5494 -0.5693 -0.5998 -0.5616 … ] (1×7199 double)}
    {[-0.0973 -0.0444 -0.0014 0.0255 -0.0098 -0.0211 -0.0042 0.0664 0.1228 0.1630 0.1807 0.1821 0.1990 0.2357 0.3063 0.3345 0.3373 0.3549 0.3853 0.4206 0.4558 0.4911 0.5123 0.5123 … ] (1×7199 double)}
    {[      0.2383 0.2987 0.3592 0.4124 0.4124 0.3987 0.3785 0.3584 0.4027 0.4486 0.4688 0.4148 0.3930 0.4140 0.4543 0.4946 0.5252 0.5051 0.4559 0.4124 0.4124 0.4462 0.4865 0.5268 … ] (1×7199 double)}
    {[      -0.6125 -0.5937 -0.5854 -0.5809 -0.5432 -0.5311 -0.5221 -0.4844 -0.5372 -0.5990 -0.6367 -0.5703 -0.5583 -0.6133 -0.6321 -0.6397 -0.6306 -0.5741 -0.5583 -0.5741 -0.6306 … ] (1×7199 double)}
    {[      -0.3252 -0.2874 -0.2707 -0.2798 -0.3555 -0.3797 -0.3706 -0.3328 -0.3101 -0.3048 -0.3237 -0.3426 -0.3252 -0.2700 -0.2510 -0.2662 -0.3101 -0.3858 -0.3933 -0.3585 -0.2828 … ] (1×7199 double)}

labl

labl = 5×1 cell array
    {[n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    …    ] (1×5000 categorical)    }
    {[n/a    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    …    ] (1×5000 categorical)   }
    {[T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    n/a    n/a    n/a    …    ] (1×5000 categorical)    }
    {[n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    …    ] (1×5000 categorical)    }
    {[T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    …    ] (1×5000 categorical)}

lablnew

lablnew = 5×1 cell array
    {[n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    …    ] (1×7199 categorical)    }
    {[n/a    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    P    …    ] (1×7199 categorical)   }
    {[T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    …    ] (1×7199 categorical)}
    {[n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    n/a    …    ] (1×7199 categorical)    }
    {[T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    T    …    ] (1×7199 categorical)}

whos

  Name         Size               Bytes  Class          Attributes

  ans          1x37                  74  char                     
  c            4x1                  432  cell                     
  cmdout       1x33                  66  char                     
  fs           1x1                    8  double                   
  fsnew        1x1                    8  double                   
  i            1x1                    8  double                   
  l            1x5000              5434  categorical              
  labl         5x1                27690  cell                     
  lablnew      5x1                38685  cell                     
  lnew         1x7199              7633  categorical              
  s            1x5000             40000  double                   
  sig          5x1               200520  cell                     
  signew       5x1               288480  cell                     
  snew         1x7199             57592  double                   
  t            1x5000             40000  double                   
  tnew         1x7199             57592  double                   

figure
subplot(211); plot(s)
subplot(212); plot(snew)

figure;
subplot(211); plot(double(l))
subplot(212); plot(double(lnew))

[Mibang]

Thank you very much, great!

I realized that I uploaded wrong label data.

I would appreciate it if you would be able to make sure with the new label data although it is no dbout on your work.

Thanks!

[Chunru]

It works with new data as updated above.

[Mibang]

Great, thanks, I also did it with your code, but deleted because yours is better.

May I ask one? Why is the new data has the length of 7199 instead of 7200?

360/250=1.44

7200/5000 = 1.44, but 7199/5000 = 1.4398.

Are you able to make the length 7200 rather than 7199? Or, 7199 is correct?

Thanks!

Besides, why t(end) ~= tnew(end)?; t(end)==19.9960 and tnew(end)==19.9944.

[Chunru]

It is due to the way we produce new time:

t = (0:length(s)-1)/fs;
tnew = t(1):(1/fsnew):t(end);

We don't extend the time after the original end of the time. You can extend it if you like by add a small time after t(end).

[Mibang]

I see, that is why...I will think about it more while I accept your answer.

Many thanks!