fmri_detrend

Description

FMRI_DETREND Removes low-frequency drift from fMRI data using spline fitting. Fits a cubic spline to a set of equally-spaced anchor points along the temporal dimension and subtracts it from the data. This removes slow drifts (scanner drift, respiratory, movement artefacts) while preserving higher-frequency BOLD signal. Operates voxelwise.

Usage

newdata = fmri_detrend(data)
newdata = fmri_detrend(data, no_of_anchor_points)

Inputs

  • DATA: fMRI time series, vectorized (228453 x T) or volume (91 x 109 x 91 x T).

  • NO_OF_ANCHOR_POINTS: Number of spline knots. More knots = more aggressive detrending. Default: 6.

Outputs

  • NEWDATA: Detrended data in the same format as DATA.

Examples

data = fmri_detrend(data);          % default 6 knots
data = fmri_detrend(data, 4);       % gentler detrending
data = fmri_detrend(data, 10);      % more aggressive

Notes

  • For high-pass filtering in the frequency domain, see fmri_bandpassfilter.

See Also

  • fmri_tempsmooth

  • fmri_bandpassfilter

  • fmri_zscorevoldata