 |
 |
Hi Michelle, I don't quite understand what you are trying to model with this design and contrasts. Could you describe what the experiment and hypotheses are? Thanks. All the best, Anderson On 21 October 2016 at 15:54, Michelle VanTieghem < [log in to unmask]> wrote: > Hi FSL experts, > > I have a 2 (between sub) x3(within sub) mixed effects model that I ran > using FLAMEO from the command line. > > flameo --cope=all_copes_angry_happy_neutral --varcope=all_varcopes_angry_happy_neutral > --mask=mask.nii.gz --dm=design.mat --tc=design.con --cs=design.grp > --runmode=flame1 > > I am trying to perform standard cluster correction (whole-brain) on the > resulting zstat images. To my best knowledge, I should use smoothest to > calculate the smoothness of the images, and then use cluster to perform the > cluster correction. > > smoothest -d 140 -r res4d.nii.gz -m mask.nii.gz >> > cluster_correction/smoothness_output > > cluster -i cluster_correction/zstat5.nii.gz -t 2.3 -o > cluster_correction/cluster_zstat5_PI_COMP_H_A_clust -d 0.43451 > --volume=234519 -p .05 --othresh=cluster_zstat5_PI_COMP_H_A.nii.gz > > > However, when I ran that cluster correction for my model, i am getting > HUGE clusters that cover half of the brain, so they are absolutely not > correctly thresholded. (Whereas when I open the raw z-stats image and > threshold manually, I do not get such large clusters). I suspect something > is wrong either with the original model (see design files attached) or the > smoothest/cluster correction. > > For smoothest, the input requires DF. degrees of freedom are unknown in a > mixed effects model, but since flame 1 uses number of copes - all of the > predictors including conditions, subjects, confounds, I calculated the DF > from the number of rows - number of columns in the design matrix. Is the > appropriate degrees of freedom for this analysis? > > Thanks! > Michelle > > > > >