Computing PLV (with Hilbert) for one frequency band

Bugs
1

Hello,

I am trying to write my script to automate the process of importing the data into brainstorm, setting the channel file, and lastly find the PLV values. I remember reading somewhere that importing the data should be done in one process and have another process for PLV. Is there a way I can implement this strategy of of separate processes into my code? I mean could anyone explain why I keep getting errors for the code shown below even though it works fine for other functional connectivity metrics such as coherence and correlation.

The portion I am talking about in my current code that is giving me an error:

%% Basic processes
    % Process: Import MEG/EEG: Existing epochs
    sFiles = bst_process('CallProcess', 'process_import_data_epoch', sFiles, [], ...
        'subjectname',   SubjectNames{1}, ...
        'condition',     '', ...
        'datafile',      {data_file_to_use,'FIF'}, ...
        'iepochs',       [], ...
        'eventtypes',    '', ...
        'createcond',    0, ...
        'channelalign',  1, ...
        'usectfcomp',    1, ...
        'usessp',        1, ...
        'freq',          [], ...
        'baseline',      [], ...
        'blsensortypes', 'EEG');
    % Process: Set channel file
    sFiles = bst_process('CallProcess', 'process_import_channel', sFiles, [], ...
        'channelfile',  {RawFiles{current_length}, 'ASCII_NXYZ'}, ...
        'usedefault',   1, ...  % 
        'channelalign', 1, ...
        'fixunits',     1, ...
        'vox2ras',      1);

    % Process: Phase locking value
    sFiles = bst_process('CallProcess', 'process_plv1n', sFiles, [], ...
    'timewindow',    [0, 15], ...
    'dest_sensors',  'EEG', ...
    'includebad',    1, ...
    'plvmethod',     'plv', ...  % 'plv' %'ciplv' %'wpli'
    'plvmeasure',    2, ...  % Magnitude
    'tfmeasure',     'hilbert', ...  % Hilbert transform
    'tfedit',        struct(...
         'Comment',         'Complex', ...
         'TimeBands',       [], ...
         'Freqs',           {{ 'beta', '13, 29', 'mean'}}, ...
         'ClusterFuncTime', 'none', ...
         'Measure',         'none', ...
         'Output',          'all', ...
         'SaveKernel',      0), ...
    'timeres',       'windowed', ...  % Full (requires epochs)
    'avgwinlength',  1, ...
    'avgwinoverlap', 50, ...
    'outputmode',    'input');  % separately for each file
    num_runs=length(sFiles);
    for run=1:num_runs
        current_folder_name={sFiles(run).FileName};
        current_matrix= in_bst_matrix(current_folder_name{1});
        TF_time_avg=mean(current_matrix.TF,3);
        current_matrix.TF=TF_time_avg;
        R = bst_memory('GetConnectMatrix', current_matrix); % get full matrix
        feature_extraction_plv(R,case_index,num_runs,run);
    end   
end
2

Yes, it must be done with two different processes. The idea is that a pipeline is comprised by process elements that can be chained.

The main reference for writing scripts is in:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting

  • What are the errors that you are getting?
  • Is there any problem at computing PLV with same parameters in the GUI?
3

Thanks for the reply...

  • What are the errors that you are getting? After I run the code from the script and go to the GUI to open the graph, I get the following error:
    image

  • Is there any problem at computing PLV with same parameters in the GUI? No, everything runs fine in the GUI, as long as I import the data first, then create another process for PLV

4

I found the mistake. I had a single rhythm (Beta) in the script, so when I replaced that with (Beta and Alpha), it ran just fine.

5

The was a bug for the specific case of one frequency band.
This is now solved at commit: 1b19f1d
Update your Brainstorm instance to get the fix.

That avoids the bug, but it is now solved.
It is possible to use only one band.

6

All good. Thank you very much for the reply.