Hello FSL people,
I am doing a TBSS analysis on FA and MD data using the previous version of
FSL (Release 4.0). I am going to use TFCE in randomise to correct for
multiple comparisons, but I need some assistance:
How has TFCE been "optimalized" for the TBSS skeleton data? I've seen these
parameters posted, but am not sure if they are the way to go with this kind
of data: --tfce_H=2 --tfce_E=1, --tfce_C=26.
Could someone explain what the height parameter refers to? Is it similar to
the cluster threshold in the -c corrections?
Also, I only want to include in the TFCE calculations voxels that have a p
>.95 (i.e., I don't want non-significant (at that alpha) voxels to be
considered in the "cluster formation".) Does TFCE already do this for me?
Is it legit to do the following:
1. Threshold a previous, uncorrected, randomise output at 0.95 and create a
mask.
2. Use this as the mask in a subsequent randomise analysis using the -T
(optimization?) option.
|