Melina Uncapher wrote:
> Dear SPMers~
>
> I understand the utility of using dispersion derivatives in an event-
> related fMRI design; however, would this technique be useful and/or
> feasible in a single-trial fMRI design in order to investigate the duration
> of a sustained response? If so, how is this accomplished in SPM?
>
If you set up a t contrast having
entries that correspond to the
dispersion derivative(s) then you will
be looking for voxels which have a
significantly longer (or shorter) response.
This approach is only valid however if
you assume that the other aspects of the
HRF (ie. height of the canonical,
latency of the canonical) are the same
over all voxels.
Understandably, you may not want to make
this assumption.
A workaround would be to compute F
contrasts looking for *any* differences
in the height/latency/length of the HRF
(between say two conditions). Having
established significant differences in
this way you could then look at voxels of
interest and see just what it was that
was different eg. responses may be
smaller but longer in condition 1 versus
condition 2.
For a discussion of these issues (but in
the context of estimating delays rather
than lengths) see Henson et al
(doi:10.1006/nimg.2001.0940) or
Liao et al (doi: 10.1006/nimg.2002.1096).
A final comment to make is that,
depending very much on your experiment,
it may be possible to model
extended/shortened responses at the
neuronal level rather than at the
hemodynamic level eg. by subdividing
your events into early and late
responses and then modelling each with a
simple canonical HRF.
Best wishes,
Will.
> Thanks in advance for your response.
>
> ~Melina Uncapher
>
> ================
> Melina R. Uncapher
> Dept of Neurobiology and Behavior
> University of California at Irvine
> [log in to unmask]
>
>
>
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7478
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|