fmri_xcorr

Description

FMRI_XCORR Voxelwise lagged cross-correlation between regressor and fMRI data. Correlates a regressor with every brain voxel across a range of time lags, returning a (228453 x 2*maxlag+1) correlation matrix. Column k corresponds to a temporal lag of (-maxlag + k - 1) scans. Useful for detecting haemodynamic delays or testing the HRF shape voxelwise.

Usage

c = fmri_xcorr(regr, data, maxlag)

Inputs

REGR: Regressor (T x 1) or (T x R) for R regressors.
DATA: fMRI time series, vectorized (228453 x T).
MAXLAG: Maximum lag in scans (integer). The function computes correlations for lags in [-maxlag, +maxlag].

Outputs

C: Cross-correlation matrix (228453 x (2*maxlag+1)). Column (maxlag+1) is the zero-lag correlation (= fmri_corregressor).

Examples

c = fmri_xcorr(stim_ts, data, 5);
[peak_r, peak_lag] = max(c, [], 2);
peak_lag = peak_lag - (maxlag+1);     % convert to signed lag
fmri_showslices(peak_lag .* (peak_r > 0.3), 1, 2, [-5 5])