fmri_acfestimate

Description

FMRI_ACFESTIMATE Estimates the mean spatial autocorrelation function (ACF). Estimates the mean ACF of pseudo-normal statistical images (pn-SI) by correlating phase-randomised regressors with brain data and computing the 3D autocorrelation of the resulting z-maps. This is the first step of the cluster-size correction procedure (Ledberg et al., 1998).

Usage

acf_vol = fmri_acfestimate(datapath, regr, n_perm)

Inputs

DATAPATH: Path to directory containing vectorized fMRI data (.mat files). Each file should contain one variable: a (228453 x T) matrix.
REGR: Regressor matrix (T x R) for R regressors of interest.
N_PERM: Number of phase-randomisations per subject per regressor. Default: 10. More = better ACF estimate but slower.

Outputs

ACF_VOL: Cell array (1 x R), one ACF volume (228453 x 1) per regressor. Each volume is the mean 3D autocorrelation across subjects and permutations, to be used as convolution kernel in fmri_csdist.

References

Ledberg, A., Akerman, S., & Roland, P.E. (1998). Estimation of the probabilities of 3D clusters in functional brain images. NeuroImage, 8, 113-128.
Ebisuzaki, W. (1997). J. Climate, 10(6), 2147-2153.