Hi there,
I've performed a trial run of FSL-VBM on 4 subjects (i.e., 2 groups; each group has 2 subjects). There didn't appear to be any major problems along the way. However, I'm uncertain about a few points:
1. At the end of Step C (creating the template), the manual advises that I "CHECK the "template_GM_4D" image in struc with the movie loop in fslview." Similar advice is given at the end of Step D (processing the native GM images): "Please do not forget the final CHECK of the 4D image of modulated registered GM images "GM_mod_merg" using the movie loop in fslview." I performed these checks, but wasn't sure what I was checking for. There didn't appear to be any noticeable problems with the images, so I moved on.
2. Is the _tfce_corrp_tstat1 image (i.e., the randomise output) expected to look like the image provided in Step E of the manual? I ask because mine doesn't. I should first mention that because I didn't find any significant gray matter volume differences between the 2 groups using a corrected p threshold of .05, I used a corrected p threshold of .20, just to see what significance looks like. I executed the following command:
fslview /usr/local/fsl/data/standard/MNI152_T1_2mm fslvbm_tfce_corrp_tstat1 -l Red-Yellow -b 0.80,1
Compared to the image provided in Step E, the _tfce_corrp_tstat1 image is not brain-extracted and the significant voxels are not graded from yellow to red, they're just red. I can't explain the discrepancy.
I'd really appreciate some guidance on these points.
Michael