How do I calculate the duration of a peak event for a cell activity signal?

I have a set of cell activity data plotted over time. I used the findpeaks function to detect peak "events" (in orange circles in screenshot below) which is measured as "dFF." I want to calculate how long each detected peak lasts for. Specifically, I want the code to take the time point of the lowest valley/point of the signal 2 seconds before and after the peak is detected and to subtract those time points to calculate the duration of the peak event. How would I be able to accomplish this?

Example code for the image below:

for i = 1:ncells % loops through each cell 
    fig_title = strcat('DFF, Cell # ', num2str(i));
    [pks, locs] = findpeaks(z_dff(i,:), MinPeakProminence=2, MinPeakDistance=10); % detects peaks in cell activity
    plot(timeSec, z_dff(i,:), timeSec(locs), pks, "o")
    title(fig_title)
                                                                                                                                                                                                                                                                                                                                                                                          
    xlim([1, maxTime])
    ylim ([(min(z_dff, [], "all")-2) (max(z_dff,[],"all"))])

    xlabel('Time');
    ylabel('dFF');

    pause;

end

[Mathieu NOE]

hello

if you still need assistance, please share some data

Answer 1

[Vinay]

Hii Esther,

As per my understanding you want to calculate the duration of the peak, which is defined as the time difference between the lowest valley before and after 2 seconds of the peak time.

The approach involves identifying the indices of the data points within the intervals [peakTime - 2, peakTime] and [peakTime, peakTime + 2], and then calculating the minimum values of "dff" within these intervals.

time = 0:0.1:10; % Example time vector
dFF = sin(time) + 1*randn(size(time)); % Example dFF signal with noise

% Find peaks
[peakValues, peakLocs] = findpeaks(dFF, time, MinPeakProminence=0.5, MinPeakDistance=2);

% Initialize array to store durations
peakDurations = zeros(length(peakLocs), 1);

% Loop through each peak to calculate duration
for i = 1:length(peakLocs)
    peakTime = peakLocs(i);
    
    % Find indices within the intervals
    preIntervalIdx = find(time >= peakTime - 2 & time <= peakTime);
    postIntervalIdx = find(time >= peakTime & time <= peakTime + 2);
    
    % Ensure intervals are valid
    if ~isempty(preIntervalIdx) && ~isempty(postIntervalIdx)
        % Find minimum valley values in each interval
        preValleyValue = min(dFF(preIntervalIdx));
        postValleyValue = min(dFF(postIntervalIdx));
        
        % Find corresponding times for these minimum values
        preValleyTime = time(preIntervalIdx(dFF(preIntervalIdx) == preValleyValue));
        postValleyTime = time(postIntervalIdx(dFF(postIntervalIdx) == postValleyValue));
        
        % Calculate duration
        peakDurations(i) = postValleyTime - preValleyTime;
    else
        % If no valid valleys are found, set duration to NaN
        peakDurations(i) = NaN;
    end
end

% Display results
disp('Peak Durations:');
disp(peakDurations);

% Plot results
figure;
plot(time, dFF);
hold on;
plot(peakLocs, peakValues, 'ro'); % Peaks
for i = 1:length(peakLocs)
    if ~isnan(peakDurations(i))
        plot([preValleyTime, postValleyTime], [preValleyValue, postValleyValue], 'go'); % Valleys
    end
end
title('Peaks and Minimum Valleys');
xlabel('Time');
ylabel('dFF');
legend('Signal', 'Peaks', 'Valleys');

Kindly refer to the following documentation for “findpeaks”:

https://www.mathworks.com/help/signal/ref/findpeaks.html

I hope this resolves the issue!