OK, I think I misunderstood your question. I thought you wanted to
filter out small clusters by specifying the cluster extent, as done in
SPM's "Results" step. Sorry for the confusion...
I guess you didn't get the "spatial extent (yes|no)" option because you
probably have already selected the cluster defining threshold (aka
primary threshold) during the "setup" step. In that case, SnPM assumes
that you want to perform a cluster size test during the "results" step
later on, so it doesn't even ask if you want to assess the spatial
extent then.
As for your question 1), ST_Ut corresponds to the cluster defining
threshold I described above. All the clusters are defined at this
threshold, but it doesn't directly control cluster p-values. For
question 2), there is no p-value for ST_Ut, because it's not a cluster
size. If you are interested, you can find the critical cluster size at
the bottom of the SnPM output.
-Satoru
> Dear Hayasaka san,
> I copied the help information from snpm_pp.m as follows. According
to
> your answers, "SpatEx" is obsolete or no longer existed. My further
> questions are: 1) Is alpha control the significance for voxel based
> inference and ST_UT control the significance for cluster size
inference?
> 2)
> Also how to estimate a P value for the ST_UT value, or what is the
cluster
> size I'm now inferring using the associated threshold"
>
> Thank you very much for your help!
>
>
> % Next come parameters for the assessment of the statistic image...
> %
> % (4) alpha: (p-value for filtering)
> % First, you are asked 'Use corrected threshold?' You can choose
to
> set
> % the threshold as FWE-corrected('FWE'), FDR-corrected('FDR') or
> % uncorrected('None'). A FWE threshold controls the chance of one
or
> % more false positives; a FDR threshold controls the expected
> % fraction of false positives among the voxels detected.
> %
> % Then you specify the threshold, the statistical significance
> % level at which you wish to assess the evidence against the null
> % hypothesis. In SPM this is called "filtering by corrected
> % p-value". SnPM will only show you voxels (& suprathreshold
> % regions if you choose) that are significant based on the method
you
> % choose at level \alpha. I.e. only voxels (& regions) with
> % corrected (or uncorrected) p-value less than \alpha are shown to
> % you.
> %
> % Setting \alpha to 1 will show you all voxels with a positive
> statistic.
> %
> %
> % (5) SpatEx: If you collected supra-threshold cluster statistics
during
> % the SnPM computation phase, you are offered the option to assess
> % the statistic image by supra-threshold cluster size (spatial
> % extent).
> % 5a) ST_Ut: If you chose to asses spatial extent, you are now
prompted
> % for the primary threshold. This is the threshold applied to the
> % statistic image for the identification of supra-threshold
> % clusters.
> %
> % The acceptable range is limited. SnPM has to collect
> % suprathreshold information for every relabelling. Rather that
> % pre-specify the primary threshold, information is recorded for
> % each voxel exceeding a low threshold (set in snpm_cp) for every
> % permutation. From this, suprathreshold cluster statistics can be
> % generated for any threshold higher than the low recording
> % threshold. This presents a lower limit on the possible primary
> % threshold.
> ---------------------------------------------
> Ze Wang, Ph. D
> Center for functional neuroimaging,
> Dept of Neurology,
> Medicine school,
> University of Pennsylvania
> 3400 Spruce Stree,
> 3W Gates building,
> Philadelphia, PA 19104
> Tel: 215-662-7341
>
>
> -----Original Message-----
> From: Satoru Hayasaka [mailto:[log in to unmask]]
> Sent: Tuesday, October 24, 2006 10:15 AM
> To: Ze Wang
> Cc: Satoru Hayasaka
> Subject: RE: [SPM] A question about snPM3 (for spm2) cluster size
> inference
>
>
> > I'm using the latest version of snPM3 for spm2. But I met a
problem
> of
> > getting the cluster size inference. I chose the "collect
> supra-threshold
> > stats" option and found the SNPM_ST.mat file, but never got a chance
> to
> > set
> > the threshold for cluster size inference, i.e., I did not get the
> "SpaEx"
> > option popped out. Could you help me about this?
>
> In SnPM, there is no option for "spatial extent" during the "results"
> process. All the cluster p-values are corrected for whatever the
p-value
> you enter. For example, if you select FWE with p<0.05, then all the
> clusters you see in the glass brain have p-values less than 0.05 FWE
> corrected. So there isn't a need for the "spatial extent" option.
>
> -Satoru
|