Dear Anne-Catherin,
In SPM, there are several different ways
to model the hemodynamic response
function (HRF). These include,
Fourier Basis (FB) sets, Finite Impulse
Responses (FIR), and the Canonical HRF.
All of these methods consist of
setting up regressors that go into the
General Linear Model.
For FB, the response is broken down into
different frequency components and for
FIR it is broken down into different
post-stimulus time windows. The most
popular method, however, is the
Canonical HRF. This is a single
regressor which captures the
stereotypical BOLD response to an
impulse of neural activity - ie.
a peak response after about 6 seconds,
followed by a smaller undershoot at
about 15 secs, returning to baseline at
about 25 secs. Mathematically, it is
composed of two Gamma functions.
See spm_hrf.m for the code.
Additionally, it is commonplace to allow
for voxel-to-voxel and
subject-to-subject variability in the
HRF by including two additional
regressors - the derivative of the
canonical with resect to time and the
derivative with respect to dispersion.
These allow you to cope with different
peak onsets and spreads. The code for
this is in spm_get_bf.m - but it's in
there with a lot of other stuff - this
function is called by spm_fMRI_design.m
which sets up the design matrix with
whatever HRF model you choose.
With the above 3 regressors (canonical
plus two derivatives) there are 3
associated parameters denoting the size
of each. These are computed when the GLM
is fit.
For more on this topic
see
http://www.fil.ion.ucl.ac.uk/spm/course/notes01/event/welcome.htm
Also, this topic (with others) is
covered well in
http://www.fil.ion.ucl.ac.uk/spm/papers/SPM-Chapter.pdf.
Hope this helps.
Best wishes,
Will.
Anne-Catherin Schuch wrote:
> Dear everybody,
>
> I am working with efMRI, examining the variability of the BOLD response.
>
> Now, I wonder which model does SPM use to model the hemodynamic response?
> And what parameters are variable when adjusting the model onto the raw
> data?
>
> I am glad for references and guidance to the code in spm_xxx.m
>
> Thanks in advance,
>
> best regards
--
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/
|