Correlation about two neural signals in specific time

new_fs = 1000;

window = 1*new_fs; % 1 second of window

start_stop_plot=[872 1016];

start_stop_plot_points_bin = [start_stop_plot(1)*new_fs, start_stop_plot(2)*new_fs];

correlation_bin = zeros(start_stop_plot_points(2)-start_stop_plot_points(1),1);

for t_bin = start_stop_plot_points_bin(1):start_stop_plot_points_bin(2)-window;

spectrum1_t = mean_value_data_smooth_1(t_bin:t_bin+window-1);

spectrum2_t = mean_value_data_smooth_2(t_bin:t_bin+window-1);

correlation(t)=corr(spectrum1_t,spectrum2_t);

end

I attached mean_value_data_smooth_1 e 2.

The problem is the index exceedes the number of arrays. Can you help me??

Answer 1

[Saarthak Gupta]

Hi,

I understand you are trying to find the rolling correlation between the two signals (‘mean_value_data_smooth_1’ and ‘mean_value_data_smooth_1’).

Both signals have 272 data points. However due to a window size of 1000, you are accessing sub signals of length 1000 which are larger than the signal itself. Moreover, the start and stop points have been incorrectly specified.

Please refer to the following code to obtain the rolling correlation between two signals:

function corrs_list = rollingCorrelation(data1, data2, startIdx, stopIdx, windowSize)
% data1 and data2 must be row vectors
n = min(size(data1,1), size(data2,1));
corrs_list = zeros([stopIdx-startIdx+1 1]);

for i=startIdx:stopIdx
    if i+windowSize-1>n
        break;
    end
    
    sample1 = data1(i:i+windowSize-1);
    sample2 = data2(i:i+windowSize-1);
    
    corrs_list(i-startIdx+1) = corr(sample1, sample2);
end
end

The function returns a list of correlations between all sub signals of size ‘windowSize’ in the given data, between start and stop indices (‘startIdx’ and ‘stopIdx’, respectively).

Please refer to the following MATLAB documentation for more details:

• https://in.mathworks.com/help/stats/corr.html