Andy,
You can use the rest_AlphaSim matlab code available at
www.restfmri.net. I run matlab from the directory containing the code,
and use lines such as those below to set essential variables and run
the simulation.
Note that your Monte Carlo simulation can be applied to a mask that
covers the whole brain, or can focus on a small volume as in the
example below, which looks only at the bilateral posterior cingulate.
Jim
addpath('/Users/jim/spm5_1782');
maskfile = '/Users/jim/projects/merit/analyze6/mask2/BILAT_post_cing.nii';
outdir = '/Users/jim/projects/merit/analyze6/mask2/';
outname = 'sm97_10000';
rmm = 2.83; % sqrt (4+4) for 2x2x2 voxels and edge connectivity
s = 9.7; % intrinsic smoothness in mm
pthr = .01; % uncorrected voxel threshold
iter = 10000; % number of iterations
rest_AlphaSim(maskfile,outdir,outname,rmm,s,pthr,iter);
On 7/16/14, Andy Yeung <[log in to unmask]> wrote:
> Dear all,
>
> I've read some papers saying '... might not be of large effect size, a
> Monte Carlo simulation was used to determine a cluster size-based
> significance threshold... more sensitive to small effects than... FWE
> correction'
>
> Could anyone teach me how to do so? I'm using SPM 8. Thanks.
>
> Best,
> Andy
>
|