Dear Margaret,
> In our event-related fMRI study, we have the behavioral response, let's
> say accuracy(continuous variable) corresponding to each trial, and we want
> to see the correlation between BOLD response in fMRI and this accuracy.
> How can we perform this kind of analysis in SPM97?
In general it is possible, but not trivial. For efMRI SPM generates a delta
functions for every event that are then convolved with the basis function you
choose (eg. HRF). What you want to do is to model HRFs with different
amplitudes according to your behavioural variable. There are 3 possibilities
to do that:
1.) You create the covariates as usual and posthoc multiply the modelled
events by your behavioural value. This is a bit tricky and I would say
impossible if events are closely spaced in time.
2.) Instead of creating delta functions of size one, you can change the
height of the delta functions according to your behavioural parameter before
convolution. You need to change the appropriate code in spm_fmri_spm_ui.m.
% create stimulus functions
%-----------------------------------------------------------
D = [];
SF = full(sparse(ons*RT/dt,1,1,k*RT/dt,1));
This line of code generates the delta functions. Remember this vector has a
temporal resolution of 100 ms.
j = round([1:k]*RT/dt);
for i = 1:h
d = conv(SF,DER(:,i));
D = [D d(j)];
end
3.) You wait for SPM99 where this will be a feature.
-Christian
--
Dr Christian B|chel
Wellcome Department of Cognitive Neurology
Functional Imaging Laboratory
12 Queen Square, London WC1N 3BG, UK
Tel.: +44-171-833-7483
Fax.: +44-171-813-1420
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|