Hi Anderson,
Thank you a lot for these explanation, I will try these designs !*
Otherwise if I try this one (4 subjects 2 runs),
EB EV1 EV2
1 1 0
1 1 0
2 1 0
2 1 0
3 0 1
3 0 1
4 0 1
4 0 1
with contrast
1 -1 (g1>g2)
-1 1 (g2>g1)
with the -e option and -permuteBlocks, it's the same as I've already run, i.e with each input the mean of the 2 runs per subject (as below), right ?
EB EV1 EV2
1 1 0
1 1 0
1 0 1
1 0 1
Last question : I read that FDR correction is less conservative. So I wanted to know if my process was good.
first, I use the uncorrected p tstat from randomise (e.g rando_tfce_p_tstat1)
I launch this command :
fdr -i rando_tfce_p_tstat1 -q 0.05 --oneminusp
and if for example FDR display 'threshold = 0.006'
In fslview I watch 'rando_tfce_p_tstat1' between 0.994,1. finally can I say the voxels I can see are significant at the FDR corrected q-value threshold at 0.05 ?
best regards,
Antoine
|