Re: [Freesurfer] Cluster Multiple Comparison Correction

5000 permutations is the generally recommended minimum, while 10000 is even better. See the FSL page on randomise http://www.fmrib.ox.ac.uk/fsl/randomise/index.html for the confidence limits for a nominal alpha of 0.05 as a function of the number of permutations.

Also, I don't know how complicated your statistical model is, but I'm not sure how the permutation engine within mri_glmfit implements its permutation testing in the case of complicated models involving covariates. It is a non-trivial issue. See the above FSL page, and the references cited therein, for a discussion of some of the technical issues involved when performing permutation testing in the presence of covariates.

cheers, -MH

On Mon, 2010-10-11 at 12:10 +0800, Jay Ives wrote:

Hi Micheal,

Thanks. Looks like I will have to dump all my results done with mc-z!

Unfortunately the Freesurfer Wiki is mute on the use of permutation, unless there is some tutorial I haven't found. How many permutations should one run? Any other helpful hints?

Cheers.......J

On Mon Oct 11 9:05 , "Michael Harms" sent:

    Hi Jay,
    Permutation testing will give the most accurate results.

    cheers,
    -MH

    > Hi Experts,
    >
    > Which option is best to use: mc-z or perm?
    > I have read that perm gives more accurate results than mc-z,
    as it uses
    > the actual data rather than fake data to estimate the null
    distribution.
    > When I use perm, I get no significant clusters.
    > When I use mc-z, almost all clusters are significant.
    >
    > Thanks......J_______________________________________________
    > Freesurfer mailing list
    > Freesurfer@nmr.mgh.harvard.edu
    > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
    >
    >
    > The information in this e-mail is intended only for the
    person to whom it
    > is
    > addressed. If you believe this e-mail was sent to you in
    error and the
    > e-mail
    > contains patient information, please contact the Partners
    Compliance
    > HelpLine at
    > http://www.partners.org/complianceline . If the e-mail was
    sent to you in
    > error
    > but does not contain patient information, please contact the
    sender and
    > properly
    > dispose of the e-mail.
    >

    _______________________________________________
    Freesurfer mailing list
    Freesurfer@nmr.mgh.harvard.edu
    https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer