How to discard beginning and end of the signal when extracting elements of the signal identified via findpeaks?

Hi all,

I have a 'good looking'signal like this:

I am extracting parts of the signal between first red and first green, second red and second green and so on, using the following code:

load 'badsignal'
% identify all troughs in data
[all_trough_value,all_trough_location]= findpeaks(-thigh_orient_y);
all_trough_value = -all_trough_value;
% Identify main peaks (i.e. ones lower than 50; blue stars)
[main_trough_value,main_trough_location]= findpeaks(thigh_orient_y,'minpeakheight',-50);
% Identify start of the movement cycle (first minimum before main peak; red circle)  
for i = 1:length(main_trough_location)
    peak_location = find(islocalmax(thigh_orient_y,'MinProminence',40));
    backward = arrayfun(@(x) diff(thigh_orient_y(x:-1:1)),peak_location,'UniformOutput',false);
    num_back = cellfun(@(x) find(x>0,1,'first'),backward);
    start_location = peak_location-num_back+1;
    start_value = thigh_orient_y(start_location);
end    
% Identify end of the movement cycle (first minimum after the main peak; grean circle)
counter = 0;
for i = 1:length(main_trough_location) 
    %finds end of movement cycle
    index = all_trough_location > main_trough_location(i);  
    if sum(index)>0 % this will ignore if the main trough is too far right so there are no peaks afterwards    
        counter = counter + 1;       
        end_location(counter) = all_trough_location(find(index,1));
        end_value(counter) = all_trough_value(find(index, 1));      
    end   
end
% Plot
figure
plot(thigh_orient_y,'k')
hold on
plot(all_trough_location,all_trough_value,'ko')
plot(main_trough_location,main_trough_value,'b*')
plot(start_location,start_value,'r*')
plot(end_location,end_value,'g*')
h = legend('signal',' all troughs','midcycle','start', 'end');
set(h,'location','northeast','box','off')
box off   
% Extract parts of the signal between red and green cycles 
%Settings  
a = start_location ;
b = end_location;
n = length(a);
T_or_y_cycle = cell(n,1) ;
%Fill in the above cell with data per movement cycle 
for i = 1:n
    T_or_y_cycle{i} = thigh_orient_y(a(i):b(i)) ;
end 

However, I get the following error

Index exceeds the number of array elements (8).

when I have a 'bad-lookng' signal (attached) like this

which I suppose is due to lack of first minimum after last main peak as circled in red on the right side. I might have signals that will lack both minimas at the begginig and end of the signal (as indicated by large red cirlces).

Thus, if this is the case, when I extract elements of the signals into the cells I want to omit first and last parts i.e. first red to first green and last red to last green. Or using any more optimal way to make this work.

Can you help please?

Answer 1

[Mathieu NOE]

hello

I wanted to challenge your code and came with a different solution . This new code will retain only valid peaks with one local mimimum before and after each peak . Your cell array extraction is now inside the single main loop.

load 'badsignal'
A = thigh_orient_y;
x = 1:numel(A);

% find all peaks above -50
ind_peaks = islocalmax(A);
ind_peaks = find(ind_peaks & (A > -50));
x_peaks = x(ind_peaks);
A_peaks = A(ind_peaks);

% find all troughs
ind_trough = find(islocalmin(A));
x_ind_trough = x(ind_trough);
A_trough = A(ind_trough);

% find local minimum before and after peak 
ind_failed = [];
for ci = 1:numel(ind_peaks)
    % local minimum before peak
    tmp1 = find(x_ind_trough <= x_peaks(ci),1,'last');
     % local minimum after peak
    tmp2  = find(x_ind_trough >= x_peaks(ci),1,'first');
   if ~isempty(tmp1) & ~isempty(tmp2)
        i_lower(ci) = ind_trough(tmp1);
        x_lower(ci) = x(i_lower(ci));
        A_lower(ci) = A(i_lower(ci));
        i_higher(ci) = ind_trough(tmp2);
        x_higher(ci) = x(i_higher(ci));
        A_higher(ci) = A(i_higher(ci));
        x_peaks_final(ci) = x_peaks(ci);
        A_peaks_final(ci) = A_peaks(ci);
        
        % Extract parts of the signal between red and green cycles 
        T_or_y_cycle{ci} = A(i_lower(ci):i_higher(ci)) ;
        %figure(1+ci),plot(T_or_y_cycle{ci})  % for debug / vizualisation only
    
        
   else
        ind_failed = [ind_failed ci]; % for information mostly
   end
    
    
end

figure(1)
plot(x,A,x_peaks_final,A_peaks_final,'db',x_lower,A_lower,'or',x_higher,A_higher,'og');

[Tomaszzz]

many thanks!

[Mathieu NOE]

as always, my pleasure !