Hello AFNIers,
Perhaps you can help me with the following question: How to do FDR so that the AFNI threshold slicer works with the following?
I have a volume of T-values (two-sided) which was not computed with AFNI and it is not marked as AFNI-statistical volume. It is a 4D volume of T-stats such that each subbrick includes different statistics. Besides, each voxel has different number of degrees of freedom which is saved in another volume.
In order to compute the p-values I did
;; Create mask of positive and negative T-values
3dcalc -a Tstats+orig -expr '1*ispositive(a)-1*isnegative(a)' -prefix signTstats
;; Absolute value of T-values
3dcalc -a Tstats+orig -expr 'abs(a)' -prefix absTvalues
;; compute p-values
3dcalc -a absTstats+orig -b df+orig -expr 'fitt_t2p(a,b)' -prefix pvaluesUNC
Now that I have the p-values I can compute FDR z-scores with 3dFDR (I want to consider the p-values=1 so I use -old option)
3dFDR -old -input pvaluesUNC+orig -mask mask+orig -force -prefix zvaluesFDRabs
and finally multiply by the mask of signed values to get the signed statistic
3dcalc -a zvaluesFDRabs+orig -b signTstats -expr 'a*b' -prefix zvaluesFDR
However, I would like to see the q-values in the AFNI GUI as I move the slicer.
This might not be possible, but in case it were possible, any suggestion will be great. Thanks very much in advance!!
Cesar
Pd. I also have F-values but i guess the reasoning will be the same...