Variability Toolbox (VarTbx) for SPM


VarTbx measures within-voxel time series variability in fMRI data. The VarTbx structure is intended to be similar to a standard SPM first-level analysis. You can then proceed to pass those first-level, variability-based images to a level-2 SPM analysis in order to model group effects of interest. However, the first-level NIFTI output files could also be used within other statistics programs of your choice (e.g., FSL, AFNI, PLS).

VarTbx currently supports modeling block designs with a boxcar model, and computes temporal variability using measures such as: detrended variance (VAR), detrended standard deviation (SD), mean squared successive difference (MSSD), and SQRT(MSSD). We plan to add additional modeling approaches and variability measures in future releases.

Using VarTbx is relatively straightforward. You first specify your first-level SPM model, as usual. You then save the model, which is used as input to specify all sessions, conditions, onsets, durations, and nuisance regressors for VarTbx. Finally, after choosing a variability metric of interest, variability-based NIFTI images are then produced. Click on the video below for a short VarTbx tutorial.

VarTbx is hosted on GitHub, and is visible within SPM Extensions and on NITRC.

We welcome any comments you may have about VarTbx. Please contact us at:

Published by:

   Lifespan Neural Dynamics Group

   (PI: Douglas Garrett; Developer: Stefan Schmidt)

   Max Planck UCL Centre for Computational Psychiatry and Ageing Research,

   Max Planck Institute for Human Development

   Lentzeallee 94

   14195 Berlin, Germany.

VarTbx is provided under the GNU General Public License; see LICENSE within the VarTbx download folder for terms and conditions.

Release History

0.1 (2015-01-15): Initial release