Dear Adam,
the randomise flag "--uncorrp" outputs uncorrected p-value images. The flag "-x" outputs voxelwise corrected p-value images. I am not sure but you may need the voxelwise uncorrected images for the FDR correction. Using both flags mentioned above would provide you with the voxelwise uncorrected image.
Kind regards,
Clara
----- Ursprüngliche Mail -----
Von: "Adam Raikes" <[log in to unmask]>
An: [log in to unmask]
Gesendet: Mittwoch, 30. August 2017 00:15:57
Betreff: [FSL] FSLVBM and Randomise
I'm working on some VBM analyses and have been following the user guide to do so. I've gotten through all of the steps and run randomise identically to what is shown (randomise -i GM_mod_merg_s3 -m GM_mask -o fslvbm -d design.mat -t design.con -T -n 5000).
I would like to look at my results and correct for false discovery, not family-wise error. I've looked at the FDR guide on the FSL website that states I need 1-p images. However, the only images I have after randomise are the *tfce_corrp* images, which are 1-p FWE-corrected images. Is there a way to get TFCE images that are not already FWE corrected?
Here is the file structure of my /stats directory:
design.con fslvbm_tfce_corrp_tstat3.nii.gz GM_mod_merg_s2.nii.gz GM_mod_merg_s3_tstat4.nii.gz
design.mat fslvbm_tfce_corrp_tstat4.nii.gz GM_mod_merg_s2_tstat1.nii.gz GM_mod_merg_s4.nii.gz
design.txt fslvbm_tstat1.nii.gz GM_mod_merg_s2_tstat2.nii.gz GM_mod_merg_s4_tstat1.nii.gz
fslvbm3b fslvbm_tstat2.nii.gz GM_mod_merg_s2_tstat3.nii.gz GM_mod_merg_s4_tstat2.nii.gz
fslvbm.defragment fslvbm_tstat3.nii.gz GM_mod_merg_s2_tstat4.nii.gz GM_mod_merg_s4_tstat3.nii.gz
fslvbm.generate fslvbm_tstat4.nii.gz GM_mod_merg_s3.nii.gz GM_mod_merg_s4_tstat4.nii.gz
fslvbm_logs GM_mask.nii.gz GM_mod_merg_s3_tstat1.nii.gz template_GM.nii.gz
fslvbm_tfce_corrp_tstat1.nii.gz GM_merg.nii.gz GM_mod_merg_s3_tstat2.nii.gz
fslvbm_tfce_corrp_tstat2.nii.gz GM_mod_merg.nii.gz GM_mod_merg_s3_tstat3.nii.gz
Thanks in advance for anyone who can help
|