It is more straightforward if you use -log10(p). If you want to do an unsigned test, then run it twice, once with pos and once with neg, then multiply the resulting cluster-wise p-values by 2. doug

On 10/21/2014 02:11 AM, Joo Tan wrote:

Hi

I have a few questions regarding the use of “mri_volcluster” for clusterwise thresholding. Essentially, we have a volume of pvalues obtained from a two tailed t-test comparing beta values from condition A vs condition B.

When I ran this command:

mri_volcluster --intransformedP.nii --sum cwthres_sum.txt --sign abs --fwhm 6--mask avgmas.nii --cwpvalthresh 0.05--thmin 2.3 --cwsig cwthresclusters.cwsig.nii

I got this error message:

ERROR: you must specify a pos or neg sign with --fwhm

Input was –log10(p-value)

Thmin was –log10(primary p threshold)

Is there a way to perform clusterwise thresholding on results from two tailed tests? Is it essential to use –log10(p) values as input? Can I use raw t-values or any other values as input and thmin instead?

Finally, are there any papers/references which this command is based on?

Many thanks!

---

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