Thanks Steve,
I am not sure if i expressed myself correctly: my concern is not that much the
implementation of GLM. Or maybe it's a dumb question. I admitt I havent done
any reading about glm. Just an intuition.
I have a simple block design task (finger-tapping) in the first level analysis,
where I included the motion correction as additional regressors. So I end up
alltogether with 8 regressors and corresponding parameter estimates. From FEAT
analysis for this subject I have extracted the time series for a the most
significant voxel and ran glmfit in matlab. The result was obviously almost the
same as in FEAT and for each parameter estimate matlab gives me a t/p value.
So my question is how do I calculate the overall t value, which would
correspond to the uncorrected t value reported by FEAT. Or is it only the t
value for the first regressor - corresponding to the task itself - which I
should be interested in in this situation?
Thanks,
Martin
On Sunday 25 April 2010 19:33:45 you wrote:
> Hi - first-level FEAT is the same as any other implementation of the
> GLM, except that it has autocorrelation correction (see the old
> Woolrich NeuroImage paper for details) which whitens the data and the
> model - that's why you won't get quite the same results as an OLS
> implementation like the one you're probably using in matlab.
>
> Cheers.
>
> On 24 Apr 2010, at 20:32, Martin Kavec wrote:
> > Hi,
> >
> > I wonder, how FEAT calculates t statistic for a given coordinate in
> > case of
> > multiple regressors, e.g. if I include a motion parameters in the
> > model.
> >
> > I am playing around a bit in Matlab with glmfit function and I feed
> > it with the
> > full design matrix and preprocessed vector of fMRI data. The
> > coefficient
> > estimates I get pretty close to what I get from FEAT, and I get a t
> > value for
> > each regressor. But I am not sure how to obtain a total t value.
> >
> > Thanks in advance,
> >
> > Martin
>
> ---------------------------------------------------------------------------
> Stephen M. Smith, Professor of Biomedical Engineering
> Associate Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ---------------------------------------------------------------------------
--
**********************************
Senior Clinical Research Associate in Neuroradiology
MRI Unit of the Department of Radiology
Erasme Hospital
Lennik Street 808
B-1070 Brussels
BELGIUM
tel: +32-2-555-4325
fax: +32-2-555-3994
email: [log in to unmask]
**********************************
-----------------------------------------------
Find a way, or make one!
|