Hello, Can you update to the latest version of FSL ( 5.0.11 ) and let me know if this fixes the randomise crash? Kind Regards Matthew -------------------------------- Dr Matthew Webster FMRIB Centre John Radcliffe Hospital University of Oxford -------------------------------- Dr Matthew Webster FMRIB Centre John Radcliffe Hospital University of Oxford > On 19 Mar 2018, at 17:12, Niall Bourke <[log in to unmask]> wrote: > > Hello neuroimaging experts, > > Problem: running randomise the unpermutated statistic is 0, generating a blank output file > > Details: > I have seen posts relating to setup_masks in the past. I can assume this was under 5.0.8/ as there seems to be differences in the script. > > My data n=68, get nice results with the same design and contrasts before including masking. > > The following is run: > > setup_masks ${root}/designMatrix/demo.mat ${root}/designMatrix/contrast.con ${root}/designMatrix/Lesion `ls ${root}/lesionMasks/*.nii.gz` > > Example randomise call: > randomise -i <input> -o <output> -d /work/nbourke/NODDI/HPC_export/designMatrix/Lesion.mat -t /work/nbourke/NODDI/HPC_export/designMatrix/Lesion.con --vxl=-4 --vxf=/work/nbourke/NODDI/HPC_export/designMatrix/Lesion [other options] > > > Questions: > > 1. Could someone clarify the dash before the 4 in --vxl=-4 > 2. The unpermutated statistic is less than 0 -> The contrasts are ok, is this a degrees of freedom? I basically cant get a result this way or is there a work around. > 3. Is randomise still expecting an separate input rather than the 4D.nii such as the previous version of setup_masks where you would specify --vxl=4,5,..72 vxf=Lesion001, Lesion002, .. > > > Any guidance on why running randomise with the output of setup_masks reduces the results to blank images would be extremely helpful. Many thanks.