Hello,
Your first image ( tbss_tfce_corrp_tstat2 ) has been analysed with TFCE and corrected for multiple-comparisons, your second image has been analysed in a purely voxelwise fashion and without correction. Given the differences between the two methods, I'm not sure exactly what uncorrected voxelwise p-value I would expect a voxel to have if it has a corrected TFCE p-value of 0.046.
When randomise generates the uncorrected voxelwise p-values it does not matter which ( if any ) cluster options are selected, the p-value calculation will be the same.
> Hello FSL Experts,
>
> I have an FA data set in which I found a significant region of interest when overlaying one of my corrected tstat files (tbss_tfce_corrp_tstat2). However, when I load the uncorrected file (tbss_vox_p_tstat2) I find no significant regions when setting a threshold at p < 0.001 (I do see more regions at 0.003). That said, I am a little confused as to how I can have a significant region using the corrected tstat (~0.046) but no significant regions at 0.001 for the uncorrected tstat. Does cluster size impact the uncorrected output?
>
> Cheers
>
|