Good afternoon, FSL Experts.
Pursuant to the previous posts on this topic, I am currently attempting to de-mean my TBSS data using randomise's -D option, after manually de-meaning covariates of no interest (e.g. age).
[Syntax used: randomise -i all_FA_skeletonised.nii.gz -o tbss -m mean_FA_skeleton_mask -d SubtractionWithAgeCovaried.mat -t SubtractionWithAgeCovaried.con -n 10000 --T2 -V -D].
However, FSL warns me that "tfce has detected a large number of integral steps. This operation may require a great deal of time to complete." I have left the randomise open and running for several days (using a nearly brand new quad-core machine with an i7 processor), and the window appears to be stuck on permutation 2 after giving the above quoted warning. I was just wondering whether anyone else has successfully completed an analysis of this nature -- I worry that the machine might just be frozen, mandating a restart and a re-run of the randomise step. I do not want to stop it, however, if it is working just fine and just needs more time to handle the data de-meaning and subsequent permutations.
I would also be curious to hear an expert's opinion regarding the use of subtracted images for within-subjects TBSS analyses. Specifically, I have two groups (clinical intervention, control) with two time points each. I am considering using FSLMATHS to subtract each subject's second time point FA map (after all registration and processing steps) from their first time point FA map, and then using these "difference images" in TBSS to determine whether the post-intervention differences (if any) are statistically non-zero and whether there is a group-by-time interaction.
Thanks kindly,
Greg
Gregory Lieberman, PhD Candidate
Neuroscience Graduate Program
University of Vermont
[log in to unmask]
|