On 18 Feb 2004, at 19:38, Michael Lipton wrote:
> Am I correct that FEAT will report voxels with signal timecourse that
> correlates with the model regardless of the phase of that correlation.
> That is, could decline in signal that is time-locked to the stimulus
> look the same as signal increase when viewed on the Z-score maps?
FEAT analyses FMRI data within the General(ised) Linear Modelling
framework. With a 1 contrast for the EV of interest, FEAT Z-maps will
show voxels where the GLM fit (the associated parameter estimate beta)
is significantly positive. If you're interested in looking at the
negative fit ('de-activation') you will need to specify a -1 contrast.
For more details on the theory behind FEAT you might want to read
through the fslcourse pages available at
> I realize that I can look for this at a given location or within a
> cluster using tsplot to assess the actual signal timecourse. However,
> there a way to generate statistical images that disclose such areas of
> negative corelation?
yep, use a separate -1 contrast for the EV of interest
> Michael Lipton
Christian F. Beckmann
Oxford University Centre for Functional
Magnetic Resonance Imaging of the Brain,
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
Email: [log in to unmask] - http://www.fmrib.ox.ac.uk/~beckmann/
Phone: +44(0)1865 222782 Fax: +44(0)1865 222717