I think that PALM may output simple p-values and not -log10(p), you can convert it with
fscalc p.mgz log10 mul -1 -o sig.mgz
On 05/10/2017 01:27 PM, Sahil Bajaj wrote:
Hello FreeSurfer experts,
I used PALM toolbox to identify significant clusters after multiple comparison correction using following command:
palm -i lh.WASI_LGI.5.mgh -s fsaverage/surf/lh.white fsaverage/surf/lh.white.avg.area.mgh -d Xg_WASI_LGI.csv -t Contrast_WASI.csv -m WASI_S5/lh.WASI_LGI.glmdir/mask.mgh -o WASI_S5/ResultsLH_WASI_LGI -C 1.95 -twotail -n 10000 -nouncorrected
My output ResultsLH_WASI_LGI_clustere_tstat_fwep.mgz gives me attached screen shot cluster when I open with FreeView.
However, I noticed that the colorbar shows values around 0.7 at the red cluster and value 1 elsewhere i.e. values lie between 0 and 1 for the whole brain.
(1). From the command and according to previous posts on FreeSurfer forum, I am assuming that significant clusters should be showing value of 1.3 (-logp, where p = 0.05) so I am not sure why the whole brain has value of 1 which is higher than at the red cluster.
(2). If significant cluster should show value of 1.3 then isn't it true that the rest of the brain should show values much lower than 1.3 but in actual rest of the brain is showing values higher than 0.7.
After displaying partial correlation maps (pcc.mgh), I made sure that the red cluster really has higher correlation than the rest of the brain.
I would really appreciate any help if someone could correct me if I am interpreting the results incorrectly?
Thanks.
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer