Dear Ping Gong and SPMers
At 11:11 PM 6/27/2005 +0800, P.Gong wrote:
>Sorry to bother you! I have two questions about SPM.
>
>I use SPM2 to analysis PET data. My tracer is 18F-FDG. I read many papers
>and found two results : correct and uncorrected P. Does the
>diffence between correct P and uncrrected P the choice of FWE and
>None? But if the result is P=0.001 with a corrected P of 0.05, .how can
>i operate the procedure of P=0.001? I select FWE and P value =0.05 and
>get corrected P of 0.05 , right?
The "FWE" button thresholds voxels based on their corrected p-values, to be
displayed in the glass brain later on, whereas the "none" button thresholds
voxels based on their uncorrected p-values. Note that, p-values in both
cases (corrected or uncorrected) I refer to voxel-level p-values.
Now, in the case of "p=0.001 with a corrected P of 0.05", it usually refers
to corrected p-values for cluster extent, rather than voxel intensity. To
perform such a cluster size test, you need to apply an uncorrected
threshold (i.e., "none") to form clusters, then assess your activation
based on corrected cluster-extent p-values. In your example, you first
apply a threshold with "none" at 0.001 to form clusters, then look for
clusters with p<0.05 corrected in terms of cluster extent.
>
>The second question: PET data for 12 subjects, group 1 has 6 patients and
>group 2 has 6 healthy men. I want to compare the change brain areas
>between patients and healthy men by SPM2. Process is as following:
>
>1. normalize
>
>2. smooth (8¡Á8 ¡Á8)
>
>3. (parameter estimation)
>
>¡ú two sample t-test
>
> ¡úproportional scaling
>
> 50 (enter)
>
> (Threshold masking) absolute
>
>Does my operation right? absolute and relative, which I should select?
>
>For my comprehension, absolute obtain corrected P value and relative
>obtain uncorrected P value. I don¡¯t know if my comprehension is right.
Actually the "threshold masking" has nothing to do with p-values. It has
something to with masking, to specify which voxels to be included in the
analysis based on intensity. In "absolute", you explicitly specify an
intensity threshold value, whereas in "relative", you specify the threshold
value in term of the proportion to the global value.
Hope this helps!
-Satoru
Satoru Hayasaka ==============================================
Post-Doctoral Fellow, MR Unit, UCSF / VA Medical Center
Email: shayasak_at_itsa_dot_ucsf_dot_edu Phone:(415) 221-4810 x4237
Homepage: http://www.umich.edu/~hayasaka
==============================================================
|