Hi - I suspect that the masking is interacting with the cluster sizes to
give what you're seeing.
I suspect that when you use the large mask the "amygdalae cluster", whilst
existing (before p-thresholding) within this mask, has had it's full
extent curtailed by the masking. In other words, the original cluster was
partially in and partially out of the large mask, so that after masking
it's estimated size was decreased, so that it failed the p-threshold-test.
Then, when you reduced the mask still further, the reduced stringency of
the multiple-comparison-correction (due to small masking) allowed this
cluster to then pass the test.
Does this sounds plausible? Cheers, Steve.
On Mon, 6 Dec 2004, Richard Albistegui-DuBois wrote:
> This may reflect my abysmal ignorance of statistics, but I am stuck and
> confused.
> When I run a group analysis on a certain dataset with no prethreshold
> masking, and cluster correction at p<0.05, I observe widespread
> clusters of positive Z's, including in the amygdalae. If I use a
> prethreshold mask on several cortical areas and the amydalae (and I
> have confirmed the the amygdalar portion of the mask includes the +Z
> clusters in the unmasked results), the amygdalar clusters do not
> appear. If, however, I use a mask which ONLY includes the amygdalar
> portions of the previous mask, the clusters do appear.
>
> I thought that using a mask would make the cluster correction less
> stringent, and so I am confused as to why the amygdalar clusters would
> show up in the unmasked image, not in the large masked image, but would
> show up in the amydalar masked image. Have I completely misunderstood
> cluster correction?
>
> I have checked, and all other aspects of the analysis were identical.
> The mask images are 0 outside the mask and 255 within the mask.
>
> Thanks in advance.
>
> Richard Albistegui-DuBois, Ph.D.
> UCLA Brain Mapping/Neuropsychiatric Institute
> phone: (310) 825-3814
> email: [log in to unmask]
> AIM: dubistegui
>
Stephen M. Smith DPhil
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|