Overview

Below is the code for running FFT on eyes-closed data. The code used follows closely to the code present here: Running FFT on EEG data in MATLAB.

The idea is that we need to segment the EEG data and then run FFT on each segments but make some modifications to get the output we want. Additionally, when we run FFT on a segment, we are really running FFT on each channel for that segment. This process reduces the data dimensions by half, and then we add one data point in the end (this is done for all segments). Afterwards we average the output of FFT for all segments.

Part 1: Load in EEG Raw File Names3 (ready for FFT).xlsx

• To view script that created this excel file: Mega-Grant EEG Preprocessing Outcome Excel

% Set the working directory
cd('Y:\STUDY 1\All EEG Files Organized\Preprocessed_RAW')

% Specify the filename
filename = 'EEG Raw File Names3 (ready for FFT).xlsx';

% Read the data from the Excel file (Eyes-Closed)
data = readtable(filename, 'Sheet', 1);

Part 2: Set the input and save pathways

% 1. Set the folder path to record the names of all the files that are done
input_filepath = 'Y:\STUDY 1\All EEG Files Organized\Preprocessed_RAW\RAW_eyes_open_and_eyes_closed_after_component_rejection\';

% 2. Pick pathway where you want to save the FFT data (Power)
save_pathway_csv = 'Y:\STUDY 1\All EEG Files Organized\Preprocessed_RAW\CSV_eyes open and eyes closed FFT\Power\';

% 3. Pick pathway where you want to save the FFT data (LogPower)
save_pathway_csv_log = 'Y:\STUDY 1\All EEG Files Organized\Preprocessed_RAW\CSV_eyes open and eyes closed FFT\PowerLog\';

Part 3: Remove file names that have already have FFT data

% % % % % % Part 1: Reading in all the files in specified folder % % % % %
% % % % % % REMAINING CODE IS AUTOMATIC % % % % % % % % 
fileNames = data.FileName;

% Remove redundancies
% Only unprocessed files will be ran by the for loop below
Already_Processed= dir(save_pathway_csv);

% Extract .set files
Already_Processed = {Already_Processed(contains({Already_Processed.name}, ".set")).name};
Already_Processed = erase(Already_Processed,".csv")'

% Remove files that have already been preprocessed
eegFiles = fileNames(~ismember(fileNames,Already_Processed));

Part 4: Running FFT on data plus saving

% % % Part 2: Running the FFT script on each eeg file % % % %
% Run an automatic version of the code for the FIRST 100 FILES!!!!

for iii = 1:length(eegFiles) 


    Current_eegFile = eegFiles{iii} %MUST BE SQUIGGLY LINE FOR SEGMENTATION TO WORK!!!!
        
    %Import data - change the name of the ID
    EEG = pop_loadset('filename',Current_eegFile,'filepath',input_filepath);
    [ALLEEG, EEG, CURRENTSET] = eeg_store( ALLEEG, EEG, 0 );
    
    % Segmentation
    EEG = eeg_regepochs(EEG, [4], [-2 2]);
    [ALLEEG EEG CURRENTSET] = pop_newset(ALLEEG, EEG, 7,'setname','ID#_Segmented','gui','off');

    % Save data into object
    A = EEG.data;
    
    % Quality Control
    fs = EEG.srate;

    Channel_Num = size(A,1);
    N = size(A,2);
    Segment_Num = size(A,3);

    freq = 0:fs/N:fs-1/fs;

    % Run FFT then take the absolute power of each channel in a segment and
    % then for all segments. 
    Power_Output = zeros(size(A(:,1:N/2+1,:))); % An empty 3-dimensional array
    Power_OutputLog = zeros(size(A(:,1:N/2+1,:)));
    
    for iii = 1:Segment_Num

        for ii = 1:Channel_Num

            % Extract current row (channel) of current segment
            xn = A(ii,:,iii);
    
            % Calculate power
            xk = (1/N^2)*abs(fft(xn)).^2; % Two-sided power
            xk = xk(1:N/2+1); % One-sided
            xk(2:end-1) = 2*xk(2:end-1); % Double values except for DC and Nyquist
    
            % Save this 
            Power_Output(ii,:,iii) = xk;

            % Convert power to dB
            xk_dB = 10 * log10(xk);
    
            % Save this too
            Power_OutputLog(ii,:,iii) = xk_dB;

        end

    end

    
    % Average all FFT data across segments into one
    Power_Output_Avg = mean(Power_Output,3);
    Power_OutputLog_Avg = mean(Power_OutputLog,3);

    % Save the frequency associated with FFT data
    frequency = freq(1:N/2+1);
    frequency_string = "x" + string(frequency);

    % Create Tables to save this data
    Table1 = array2table(Power_Output_Avg, 'VariableNames', frequency_string);
    Table2 = array2table(Power_OutputLog_Avg, 'VariableNames', frequency_string);
         
    % Introduce Channel variable
    Table1.Channels = extractfield(EEG.chanlocs,'labels')';
    Table2.Channels = extractfield(EEG.chanlocs,'labels')';
    
    % Save the files (respectively)
    cd(save_pathway_csv);
    writetable(Table1, [Current_eegFile '.csv']);

    cd(save_pathway_csv_log);
    writetable(Table2, [Current_eegFile '.csv']);
    
       
end