detect troughs (minimum) of a signal to fire an action. (the signal is being acquired online)

Hi.. I am acquiring a signal from an EEG. I know approximatly the frequency of the signal.

I need to play a sound when the signal is in its minimum ( a trough ). So I need to detect the correct instant to fire the audio in real ltime.

I saw some functions as findpeaks but if I understood it correctly, it needs a store signal to work and it is not used in real time detection. Is that right?

Is there any tool boxes in signal processing that I could use for this taks.?

Tks

[Image Analyst]

How would you know any particular point is in a trough? How could you predict the future to know that it won't continue lower, deeper into the trough and that it's not early in the trough or on the side of it? Could you just set a threshold somehow and if the signal falls below that threshold, it's considered to be a trough, or at least entering a trough even if it's not at the very bottom of it?

[L]

@Image Analyst We can accomodate some lag for firing. So, I was thinking about using some kind of smoothing filter (using a small window) and then calculating the first derivative. If the derivative goes feom negative to positive, than I would fire. But I am new to signal processing, so I am not sure if this is the best approach.

Answer 1

[Pratyush Swain]

Hi L,

I understand you have a mechanism to receive signal values in real time and you want to detect troughs in it. The function 'findpeaks' is not feasible since it needs a stored signal as its input. We can incorporate window approach for handling realtime signal data.

Please refer an example implementation as follows:

% Set the sampling frequency and approximate frequency of the EEG signal
fs = 1000; % Sample rate in Hz
approximateFrequency = 10; % Set Approximate frequency of the EEG signal in Hz

% Set the window size for the smoothing filter and derivative calculation.
% Ensure this window size is more than the length of received signal from online
windowSize = round(fs / approximateFrequency); % Adjust this value as needed

% Set the threshold for trough detection
threshold = -0.5; % Adjust this value based on the characteristics of your EEG signal

% Initialize variables
eegSignal = zeros(windowSize, 1);
troughDetected = false;

% Main loop for real-time processing
while true
    % Acquire the EEG signal (replace with your specific code)
    newSamples = acquireEEGSignal(); % Replace with your code to acquire the EEG signal
    
    % Handle multiple samples returned by acquireEEGSignal
    for i = 1:numel(newSamples)
        % Update the EEG signal buffer
        eegSignal = circshift(eegSignal, -1); %takes the first element of arr to end of arr 
        eegSignal(end) = newSamples(i); %substitute value at the end of arr with new value
        
        % Apply a smoothing filter to the EEG signal for noisy fluctuating
        % data
        smoothedSignal = smoothdata(eegSignal, 'movmean', windowSize/3);
        
        % Calculate the first derivative of the smoothed signal
        derivative = diff(smoothedSignal);
        
        % Check if the derivative goes from negative to positive
        if derivative(end-1) < 0 && derivative(end) > 0 && ~troughDetected
            % Play sound or perform action at the detected trough
            playSound(); % Replace with your code to play sound or perform desired action
            
            % Set trough detected flag to prevent repeated triggering
            troughDetected = true;
        end
        
        % Reset the trough detected flag if the derivative goes below the threshold
        if derivative(end) < threshold
            troughDetected = false;
        end
    end
    
    % Add a delay to control the loop rate
    pause(0.01); % Adjust the delay time as needed
end

Please note the 'threshold' value signifies a slope/derivative value and helps prevent false triggers by requiring the derivative to fall below a certain threshold before considering the detection of a new trough.This mechanism ensures that only significant troughs are detected and helps filter out small fluctuations or noise in the EEG signal.

Hope this helps.