hi all
I applied the dual regression script to my GICA data. investigating the correlation between a set of cognitive variables and subject's RSNs.
as previously discussed here, I looked results networks-by-networks. I recognized a RSN in the group data, and i looked in the corresponding stage_3 image for significant voxels.
1) what looks strange to me is the focality of the TFCE results. for some contrast i observed just ONE significant voxel (>0.95), in some cases with also very high p-value (0.989). but surrounded by voxels with 0.5-0.7 (very far from the significant value). is it "normal" this focality ? can this suggest some problem in my data? considering that TFCE was thought to "find clusters" , actually i very hardly get clusters !!
just voxels.
2) I could see voxels which have a very low Zscore in the melodic group results (a 0,7 in a parahyppocampal voxel in the DMN component) . when applying DR, the same voxels get significant even if the parahyppocampal region is not normally included in the DMN. whay should I think about this voxels??
3) is a previous post it was suggested to use a different thresholding
dr_stage3_ic0000_tfce_p_tstat1 -thr 0.99
instead of
dr_stage3_ic0000_tfce_corrp_tstat1 -thr 0.95
when can i use this? how can we convince the always more erratic reviewers that this correction might be acceptable
thanks in advance
Alberto
|