> Dear SPMers,
I just realized that I originally sent this message out with the wrong
subject heading.......Here it is again......
>
> If I threshold a T-map at p=.01 and k=14 contiguous voxels, then I get
a
> Table that reveals the following information.
>
>
> Expected voxels per cluster, <k> = 5.798
> Expected number of clusters, <c> = 5.32
> Search volume: S = 3036994 mm^3 = 30852 voxels
=763.3 resels
> Voxel size: [3.8, 3.8, 7.0] mm (1 resel = 35.60
voxels)
>
>
> I'd like to determine the effective false positive rate for individual
> voxels in this T map. I assume that it's going to be less than .01
because
> of the 14 contiguous voxel spatial extent threshold. Is the formula below
> correct?
>
> Voxelwise False Positive Rate = (5.798 voxels per cluster)*(5.32
expected clusters)/(763.3 resels*35.6 voxels per resel)
>
> =
(30.8454 expected voxels)/(2.7173e+004 total voxels)
>
> =.0011
>
>
> Finally, if I performed a conjunction analysis using two T maps
> thresholded in exactly the same way as above, then would the false
positive
> rate be .0011^2?
>
>
> Many Thanks,
>
> > Daniel Weissman, PhD
> > Center for Cognitive Neuroscience
> > Duke University
> > Durham, NC 27705
> > phone: (919)-681-1029
> > fax: (919)-681-0815
> > e-mail: [log in to unmask]
> >
> >
>
|