Hello,
Basic info: I am using PALM to run a group level analysis on the HCP data. As a check that I am running palm correctly and understand how to load and view the results, I wanted to get a group average for right hand activation in a motor task. I used ~700 subjects. The design matrix was a column of ones and the contrasts were C = [1 0; -1 0]. I ran the subcortical and cortical surfaces separately through PALM, then combined the outputs using wb_command.
My problem: It seems as if p-values are not being created, and when they are, there are no p-values for the surface, only for the subcortical structures. When the open the tstat output, I see tstat results, but some of the corresponding p-value outputs do not show anything (and I have not tried to threshold the p-values). Does anyone have any ideas as to why this might be? Below I have included more details on what I see when I view the results, as well as the actual commands I used to run palm and to combine my outputs.
I found a similar post with randomise, but the response just had the user post his files for someone to review.
Here are some more specifics about what I am seeing: When I looked at the results (the p-values, saved as -logp) in wb_view here is what I get:
Volume view tfce: I see regions of activation in the cerebellum and near the ventricles with -logp values ranging from 0 to 150,000
Volume view with cfwep: No activation visible and no p values in the chart view.
Cortical surface tcfe & cfwep: no activation visible anywhere on the surface, when I would expect to see activation in the motor cortex.
When I look at the t-stat directly:
tcfe: I still see activation in the volume view in the cerebellum, but I also see activation on the cortical surface. There is a t-stat of 24.3 (the highest value in the range created) for the motor cortex.
cfwep: A range of zero with no activation areas anywhere on the volume or cortical surface view.
I also tried running palm in a single step, instead of separating the cortical and subcortical. When I did this, I got the same results for the tstat in both volume and surface view, but for the p-values (tfce) while both showed nothing on the surface view, I lost the activation I saw previously in the volume view (I know this is not the correct way to run it, I just figured maybe it could help me figure out what was going wrong).
Does anyone have any ideas as to why I am not seeing significant activation in the motor cortex-- i.e. why some files do not show any p-values at all, and when they do, why none of those pvalues appear on the cortical surface?
Thanks,
Kelly
Here is the palm command I am running:
palm -i Cope5combined_motor_subcortical.nii -noniiclass -d designmatrix_av.mat -t C_av.con -o results_RH_av_subcortical -n 500 -corrcon -logp -accel gamma -T -nouncorrected
palm -i Cope5combined_motor_left.func.gii -i Cope5combined_motor_right.func.gii -d designmatrix_av.mat -t C_av.con -o results_RH_av_cortical -n 500 -corrcon -corrmod -logp -accel gamma -T -tfce2D -s L_midthickness.surf.gii -s R_midthickness.surf.gii -nouncorrected
And this is how I am combining the outputs:
results=results_RH_av
Cope=Cope5combined_motor
for c in 1 2; do
wb_command -cifti-create-dense-scalar $results\_ordn_tstat_c${c}.dscalar.nii -volume $results\_subcortical_vox_tstat_c${c}.nii $Cope\_subcortical_label.nii -left-metric $results\_cortical_dpv_tstat_m1_c${c}.gii -right-metric $results\_cortical_dpv_tstat_m2_c${c}.gii
wb_command -cifti-create-dense-scalar $results\_tfce_tstat_c${c}.dscalar.nii -volume $results\_subcortical_tfce_tstat_c${c}.nii $Cope\_subcortical_label.nii -left-metric $results\_cortical_tfce_tstat_m1_c${c}.gii -right-metric $results\_cortical_tfce_tstat_m2_c${c}.gii
wb_command -cifti-create-dense-scalar $results\_tfce_tstat_cfwep_c${c}.dscalar.nii -volume $results\_subcortical_tfce_tstat_cfwep_c${c}.nii $Cope\_subcortical_label.nii -left-metric $results\_cortical_tfce_tstat_mcfwep_m1_c${c}.gii -right-metric $results\_cortical_tfce_tstat_mcfwep_m2_c${c}.gii
done
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
|