It's basically something I've modelled after randomise (FSL's permutation program) and AlphaSim (AFNI's monte carlo simulator).
I've built a simulator into the glm analysis program. It will allow you to perform either monte carlo or permutation simulations. On each iteration it will perform clustering to determine the size of the largest cluster, in order to determine the distribution of the max cluster size. You can then pass this distribution to mri_surfcluster and it will determine the pvalue for each of the clusters. I may try to add a random field theory pvalue as well, but we'll have to see. FDR is already built into tksurfer.
note: We can't use randomsise or AlphaSim because they do not operate on our surface geometry.
doug
Glenn Lawyer wrote:
Hi, Doug,
You mentioned on the list that you are preparing a tool that uses clustering to reduce the multiple comparison penalty. Do you have time to post some more details, or perhaps just pass along some of your references? This area is very interesting to me.
+glenn
--
I want to go farther, farther!
A. E. L.Glenn Lawyer < +352 061 967 244 < Instituttgruppe for psykiatri < Postboks 1130 Blindern < 0318 Oslo <
< http://folk.uio.no/davidgl >H <><><><><><><><><><><><><><><><><:)
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer