Re: [Marsbar-users] Percent signal change question

[Thomas S. <tst...@ne...>]

Hi,

just to give a more intuitive explanation of the problem:
The Marsbar-FAQ tells me that the percent signal change is expressed in 
relation to the mean signal of the ROI.
But: When I look at all 4 conditions of the experiment, the mean of 
those 4 conditions signal changes should be the same
as the mean signal of the ROI, because there are no other periods in 
this experiment.
So how can they all be negative?
I think I still missed a detail with that?

Best regards
Thomas

Dr. Thomas Stephan
Abteilung fuer funktionelle, strukturelle und molekulare Bildgebung
Zentrum fuer Sensomotorik
Klinikum Grosshadern der Universitaet Muenchen 
Marchioninistr. 23
81377 Muenchen

tst...@ne...
+49 089 / 7095-4819 (Fon)
+49 089 / 7095-4801 (Fax)



Thomas Stephan schrieb:
> Hi,
>
> I used marsbar to extract percent signal change values for all 4 
> conditions of a block design experiment.
> The experiment included 4 conditions, and all of them have been 
> modelled, because there is no real
> baseline condition in this experiment. ROIS have been defined as 
> spheres around the maxima of the
> RFX-activation clusters. Percent signal change values have been 
> extracted for each subject and each condition.
> These values have been averaged to give a group-percent-signal-change 
> estimate for each experimental condition.
> For some of the ROIs, all of these four values are negative, for other 
> ROIs all of them are positive.
> Now the question:
> As I read in the marsbar documentation, all the values are scaled to 
> the mean of the roi.
> How can it be that the values do not sum to zero? The 4 conditions 
> model the whole period of the experiment,
> so I thought that they should sum to zero.
>
> Best regards
> Thomas Stephan
>