Dear Kalina,

I'm not sure if I understand why beta only represents the percent change
under perfect fit conditions. My understanding is that Y and beta are the
(data) fit and the fitting parameter respectively, and so should be
representative of the percent change always. Perhaps you can provide an
example?  This old message by Karl Friston appears to agree with me:

> > There is no automatic facility but the percent (of whole brain signal)
> > activation of a voxel is easily calculated from the parameter estimates
> > - the variable 'beta' in working memory following a plot. These values
> > correspond to a VOI defined by the spatial smoothing kernel, centred
> > on the selected voxel.
> >
> > I hope this helps - Karl

I do agree however that perhaps instead of calculating the mean beta for the
whole VOI fitting the model to a "collective time series" (e.g. the
eigenseries that spm_regions returns), may give a more representative
number. Perhaps someone else can weigh in...

                                                                                                                Shy



> -----Original Message-----
> From: Kalina Christoff [mailto:[log in to unmask]]
> Sent: Thursday, April 12, 2001 10:59 PM
> To: Shy Shoham
> Cc: [log in to unmask]
> Subject: Re: percent signal change questions (a simple hack)
>
>
>
> Dear Shy,
>
> > the beta values can be accessed and averaged in the matlab
> workspace. It is
> > my understanding that in a globally scaled simple box-car
> design they will
> > represent percent change relative to the mean brain signal.
>
> This is just my personal opinion - and I would very much like to hear
> other people's arguments and thoughts on this - but I would argue
> that the beta values cannot be used to compute percent signal change.
>
> My understanding is that the beta values would be representative of the
> average intensity values only under conditions of perfect statistical fit
> between the model and the data. This is hardly the case for neuroimaging
> data, however (it is hardly the case for ANY data, but for neuroimaging
> even more so). And if the fit is poor, the beta values become unreliable
> (in fact they would be no longer interpretable). I have personally seen
> instances in which the average beta values form a very different pattern
> than the average intensity values for a given voxel or an ROI.
>
> On a more pragmatic note, regarding your question on how to get the
> average values: A while back I wrote some programs for ROI analysis - that
> use the Y.mad file to extract these intensity values. I wrote them
> mostly for event-related plotting, but have also been doing
> some changes to them more recently, so that average bargraphs can also be
> plotted for blocked-design studies. Several people have found them
> useful. You can download them from:
>
> http://www-psych.stanford.edu/~kalina/SPM99/Tools/roi.html
>
> but if you do so, please use them at your own discretion and
> risk. Although they have been tested extensively over the past half year,
> there can be still problems or bugs.
>
> And of course keep in mind that data are extracted from the Y.mad file
> (see the above webpage for more info).
>
> All the best,
>
> Kalina
>
> __________________________________________________________________
> ___________
> Kalina Christoff                    Email:  [log in to unmask]
>                                     Office: Rm.455; (650) 725-0797
> Department of Psychology            Home:   (408) 245-2579
> Jordan Hall, Main Quad                      Fax:    (650) 725-5699
> Stanford, CA 94305-2130
http://www-psych.stanford.edu/~kalina/
____________________________________________________________________________
_