Hello,
In general the uncorrected p-images will appear to have more significant voxels, however the uncorrected P-values do not account for multiple comparisons and so are of questionable utility unless a separate correction is applied ( e.g. FDR ).
Many Regards
Matthew
> Dear FSL experts:
>
> I am new user in FSL and I have tried TFCE and have several questions.
>
> 1: tbss_4_prestats threshold.
> bigger threshold generates less fibre branches in the mean FA skeleton image. what is appropriate (default 0.2). When I use 0.2 In my data, it seems to generate more fibre branches than displayed in the website. does the different threshold affact the statistical result?
>
> 2: Uncorrected P and PWE-Corrected results
> In uncorrected P result tbss_tfce_P_tstat, I found many significant regions (p < 0.05), but in PWE-corrected result tbss_tfce_corrp_tstat, I didn't founf any significant region (p < 0.05), is it possible?
>
> Here I used randomise -i all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -d design.mat -t design.con -n 5000 --T2 --V
>
> Thank you very much
>
> Best
>
> Shugao
|