SPM12

More coming soon! please email andrew or raise a PR on github if you'd like to contribute to this page

Running first-level analyses with parfor

Info

Example contributed by Arkady Konovalov

Simple parallelisation of a for-loop can be performed using parfor. This functionality is provided by MatLab and enables faster processing of for loops simply by changing the syntax at the start to say parfor rather than for.

Here is an example function which makes use of parfor whilst computing GLMs using SPM.

function glm_level1(model)
% This function takes a model structure as input and performs first-level
% estimations in a General Linear Model (GLM) analysis for a set of subjects.

subjects = model.Subj;

% FIRST LEVEL (individual) estimations
% Get the number of subjects to be processed.
N = size(subjects,2);

% Iterate over each subject in "subjects" using parallel processing
parfor i = 1:N

        % Get the current subject ID from "subjects"
        id = subjects(i);

        % Get the corresponding BIDS (Brain Imaging Data Structure) ID and
        % session information for the current subject.
        BIDS_id = model.ids{id};
        BIDS_sess = model.sess{id};

        % Construct the path to the GLM folder for the current subject.
        path = [model.glmfolder BIDS_id];

        % Construct the path to the SPM.mat file for the current subject.
        modelfile = [path '/SPM.mat'];

        % Delete the existing SPM.mat file for the current subject (clean
        % up previously done models)
        delete(modelfile);

        % Create a job structure for the current subject.
        job = analysis_job_func(BIDS_id, BIDS_sess, model);

        % Create an empty cell array to be used as inputs for the "spm_jobman" function.
        inputs = cell(0,1);

        % Set the SPM defaults to 'FMRI'.
        spm('defaults', 'FMRI');

        % Run the current job using the "spm_jobman" function.
        spm_jobman('run', job, inputs{:});

end

end

Note

Make sure you specify the appropriate number of cores when starting the MatLab GUI App, you may not notice a substantial speed-up if you run MatLab using the default of 4 cores. Do try to avoid asking for substantially more than you might need however - BlueBEAR is a shared resource.