processing fMRI files

Hello,

I am trying to generate ROI-to-ROI correlation matrices for subjects with an in-house data-set.  I have noticed that the correlation matrices differ from run to run. Typically this may mean that the random number generator is not being initialized (seeded). In digging through the processing I did find some initialization (with rand('state',0)).

Is this typical behavior or am I doing something wrong

Thanks,

Doug

BTW: Here is my batch pipe:

batch.Setup.preprocessing.steps = {'functional_realign&unwarp','functional_center','structural_center','structural_segment&normalize','functional_normalize','functional_art','functional_smooth'};

batch.Setup.preprocessing.art_thresholds=[art_thres1,art_thres2];
batch.Setup.preprocessing.sliceorder='interleaved (GE)';
batch.Setup.done=1;
batch.Setup.overwrite='Yes';

%% DENOISING step
% CONN Denoising % Default options (uses White Matter+CSF+realignment+scrubbing+conditions as confound regressors); see conn_batch for additional options
batch.Denoising.filter=[0.01, 0.1]; % frequency filter (band-pass values, in Hz)
batch.Denoising.done=1;
batch.Denoising.overwrite='Yes';

and the correlation matrix is generated in python with nilean

nilearn.input_data import NiftiLabelsMasker
masker = NiftiLabelsMasker(labels_img=atlas_filename, standardize=True,memory='nilearn_cache', verbose=5)
time_series = masker.fit_transform(fmri_filenames)