Dear Sylvain,
If you want to know how strong the activation is, you want to look at the
parameter estimates (beta) of the model. Essentially, these are the estimated
"amplitudes" of the response. The t scores, p- values, ...etc are a measure of
how well those estimates fit the data. My suggestion would be to determine
which areas fit the model by looking at the thresholded t-maps, and then examine
the betas in those regions. Those values are saved in the images called
"con_xxxxx.img", and they are generated automatically when you do your analysis
(somebody please correct me if that's not the case).
I hope that helps
-Luis
Sylvain Clément wrote:
> dear Stefan,
> I followed this thread with interest and have some questions...
>
> At 20:42 21/11/2000 +0000, you wrote:
>
> >I wouldn't say that you're not allowed to say that... :-) but the
> >corrected p-values just give you the 'improbability' that your
> >observation was acquired under the null-hypothesis (as specified by your
> >contrast). This means you can only infer from a t-map that there was
> >some evidence such that you can reject the null-hypothesis for voxel,
> >clusters or sets. No information about the strength I'm afraid.
>
> That's what I've always thought. but, is there a good measurement of
> strength of activation?
> May be one can use the amplitude of the fitted response ? What's your
> opinion on that point?
> I'm just a beginner in using SPM and I'm a little disappointed in front of
> the different ways used to analyse fMRI data.
>
> Thanks in advance
>
> Sylvain Clément
>
> >email: [log in to unmask]
>
> ------------------
> Sylvain Clément
> Equipe "Fonction
> auditive" http://www.scico.u-bordeaux2.fr/~psyac
> Laboratoire de Neurophysiologie
> B.P. 63, UMR CNRS 5543
> 146 rue Léo Saignat
> 33076 Bordeaux Cedex, FRANCE
>
> Tel: 05 57 57 16 51
> Fax: 05 56 90 14 21
--
http://www.bme.umich.edu/~hernan
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|