FDR - False Discovery RateFDR User Guide | ![]() |
fdr is a simple program which takes in a p-value image and uses FDR theory to carry out multiple comparison correction. For detail on FDR theory see Tom Nichols' FDR page.
The p image fed into fdr can be produced in a number of ways, for example:
ttologp -logpout logp1 varcope1 cope1 `cat dof` (this outputs the log(p) image) fslmaths logp1 -exp p1 (this creates the p-value image) fdr -i p1 -m ../mask -q 0.05 Probability Threshold is: 0.00339819 (this tells you the FDR threshold - p-values below 0.00339819 are significant) fslmaths p1 -mul -1 -add 1 -thr .99660181 -mas ../mask thresh_1_minus_p1 (this creates a 1-p image for ease of display, thresholds that at 1-thresh and then remasks)
Note that if you are looking at a cope created by FLAME fixed-effects, the correct dof is not in the "dof" text file, but in the tdof_t1 image.
*_vox_p_tstat*
image. This is a p-value image from
0:1 where 1 is most significant (arranged this way to make display and
thresholding simple). Therefore this needs to be inverted before
feeding into fdr. Thus an example would be:
fslmaths grot_vox_p_tstat1 -mul -1 -add 1 1_minus_grot_vox_p_tstat1 fdr -i 1_minus_grot_vox_p_tstat1 -m mask -q 0.05This gives you the threshold that should be applied to the p-values. To view this, the easiest thing to do is take 1-threshold and threshold the original _vox_p_tstat1 image:
fslmaths grot_vox_p_tstat1 -thr <1 - FDR threshold> thresh_grot_vox_p_tstat1