Dear Zhiqiang Zhang,
SPM performs a voxel-wise statistical test. The number of degrees of
freedom depends on your design, not on the volume covered by your mask.
The shape and volume of your mask is important only to correct the
significance levels for multiple comparisons.
Therefore, one would normally not observe changes in T- or uncorrected
p-values with and without a mask. Only corrected p-values and
cluster-level statistics are likely to change, if the mask covers much
less than whole brain volume.
Am Montag, den 14.07.2008, 10:49 +0800 schrieb Zhang Zhiqiang:
> Dear SPMers,
> I want to compare the difference in a defined area between two groups
> of subjects. A mask was first used to all data, so as to exclude the
> data outside this defined area (was set to zero). this processing was
> supposed to facilitate the detection of difference, because the "free
> degree" will be decreased after mask.
> However, we found the result of 2-sample t-test (in the defined area)
> didn't change a little (in contrast to that the mask was not used). I
> think which might be owing that SPM is still taking account the "zero
> value (outside the mask)" into t-test calculation. Therefore, I want
> to know, whether SPM (SPM2) can perform t-test only within a mask?
> Thank you very much!
> Zhiqiang Zhang
Department of Neurology [log in to unmask]
Universitaetsklinikum Freiburg Phone 49(0)761-270-5331
Breisacher Str. 64 Fax 49(0)761-270-5416
79106 Freiburg http://fbi.uniklinik-freiburg.de/