fmri_scramblephases

Description

FMRI_SCRAMBLEPHASES Phase-scrambles a 3D brain volume or time series. Randomises the phase spectrum of the input while preserving the amplitude spectrum exactly. The result is a surrogate with the same spatial or temporal autocorrelation structure as the original but with shuffled phase relationships — i.e. same power spectrum, destroyed signal. Used for generating null distributions in permutation tests: by correlating phase-randomised regressors with brain data you obtain pseudo-normal statistical images with the same autocorrelation as the true SI, which is the basis of the cluster-size correction in fmri_csize.

Usage

fs = fmri_scramblephases(f)

Inputs

F: Input array (any dimension: 1D time series, 2D, or 3D volume). For fMRI time series: pass a (T x 1) vector to scramble temporal phases. For a 3D volume: pass a (91 x 109 x 91) array to scramble spatial phases.

Outputs

FS: Phase-scrambled array of the same size and class as F. The output is real-valued (imaginary residuals from IFFT are discarded).

Examples

% Scramble temporal phases of a regressor
regr_null = fmri_scramblephases(regressor);
% Scramble spatial phases of a statistical volume
vol_null = fmri_scramblephases(stat_volume);
% Generate null correlations for permutation test
r_null = fmri_corregressor(fmri_scramblephases(regressor), data);

References

Ebisuzaki, W. (1997). A method to estimate the statistical significance of a correlation when the data are serially correlated. Journal of
Climate, 10(6), 2147-2153.