Hi Liang,
since tfce statistic cannot be modeled by a parametric distribution,
you need a permutation testing, like randomise to assess its significance.
So, if you want to have a p-value map for your tfce map,
you need the summary of all the permutations in your randomise (i.e., max_statistic of each perm.).
As you've guessed correctly, what you have now, is the tfce statistic image of your FA (thats what your flsmaths did),
which is not convertible to a p-value map.
So, you should rerun the randomise with a TFCE (-T) option.
cheers
Hi FSL experts,
I have two questions on TFCE method when processing the FA data using TBSS.
1) In the randomise command, I initially didn't add -tfce but use -x term and got some uncorrected stat images (also voxelwise p images) and FWE corrected images. However, now I want to see the TFCE results, just don't rerun the data in terms of taking a long time. Considering the information in the TBSS website, I can apply the fslmaths command to existing stat images as follows:
fslmaths *_tstat1.nii.gz -tfce 2 0.5 12 newfile (the set of -tfce maybe be changed depend on individual investigations).
I find that there is only one generated file (newfile.nii.gz). I guess this file is orignial tfce image and basiclly it includes very large number (e.g. > 100).
My question is how to threshold this file or convert the file into p images to get whether there is significant voxels.
Any suggestions? thanks.
Liang Wang
--
Liang Wang, Postdoc
Woodward Lab
Department of Psychiatry
University of British Columbia
BC Mental Health & Addiction Services
938 West 28th Avenue
Vancouver BC V5Z 4H4
Telephone: 1-604-875-2000 (ext. 4735)
Fax: 1-604-875-3871
Email: [log in to unmask]