Tobias,
Sorry for the delay.
> I have a couple of questions regarding the use and interpretation of
> the CorrClusTh.m script. When I run the script on an SPM.mat I am
> asked to define the "cluster defining th.", and it spits out a cluster
> size and an associated corrected P value. Does the cluster defining
> threshold refer to the single voxel height threshold (uncorrected)?
Yes, that's right.
> And the output refers to the size of cluster of voxels at that level
> which would correspond to a FWE corrected statistic of combined
> height/extent?
No, not quite. The output gives the size of a cluster of voxels
that would be just significant for a FWE corrected inference. Period.
It's nothing fancy, it's based on the same cluster size you see
in the SPM table output.
Combined hight/extent is another beast altogether. That was available
in SPM96 and is now also available in SnPM.
> How do I report the use of this method (e.g. does it correspond to the
> one suggested in Hayasaka & Nichols, 2004)?
Nothing to report... it's the same thing you would have gotten from
SPM if you had been lucky enough to get clusters of every possible
size. :) If you feel you should report something, just cite the me,
the script name, and the date of the SPM list email that had it.
-Tom
-- Thomas Nichols -------------------- Department of Biostatistics
http://www.sph.umich.edu/~nichols University of Michigan
[log in to unmask] 1420 Washington Heights
-------------------------------------- Ann Arbor, MI 48109-2029
|