search

andersonwinkler / PALM

PALM: Permutation Analysis of Linear Models
69 stars 28 forks source link

PALM output file #62

Open lyhoo23618-csu opened 21 hours ago

lyhoo23618-csu commented 21 hours ago

Dear Anderson,

I'm running palm to test for group differences in cortical thickness (from Freesurfer). When I check the results of the output file (fsp_tfce_tstat_fwep.mgz). If i understand correctly, this file saves the FWE corrected p values for each vertex. However, I find a large regions presenting values at 0; and some regions presenting values between 0-0.45 (as shown below). If i want to identify the regions with significant (p=0.05) group differences, shall I take a threshold at 0.05 for this file?

image

Here is my code for PALM:

hemi = 'lh'; input = 'lh.213.thickness.10.mgh'; maskfile = 'lh.cognition_compare.glmdir/mask.mgh'; surffile = '/opt/freesurfer/7.3.2/subjects/fsaverage/surf/lh.white'; areafile = '/opt/freesurfer/7.3.2/subjects/fsaverage/surf/lh.white.avg.area.mgh';

cft = 1.3; pthresh = 10^-cft; zthresh = fast_p2z(pthresh); iters = 5000;

zthreshstr = sprintf('%f',zthresh); itersstr = sprintf('%d',iters);

palm('-i',input,'-m',maskfile,'-d','design.mat','-t','design.con','-logp',... '-n',itersstr,'-C',zthreshstr,'-o','fsp','-s',surffile, areafile, '-T');

Can you help me understand these output files?

Thank you! Best, Hao Li

lyhoo23618-csu commented 18 hours ago

Hi, i just relaized there is a '-logp' flag, so the threshold would be 1.3 for me, that make sense.