Hi FSL experts,
Here I have some questions about the result of dual regression, after the melodic ICA process.
So assuming that I have ran group melodic using a model comparing two groups (10 PAT VS 10 CON), contrast 1 being PAT>CON and contrast 2 being CON>PAT. And I successfully recognised 8 R-S networks.
Now I wanted to know if there's any difference in those networks between these two groups, so I did dual regression using model above and 5000 permutations, and got a output file of 3 stages.
Then I used "fslstats -R" to check for 1-p value of each component and each contrast, and there was a component that pass the 0.95, let's say it was dr_stage3_ic0002_tfce_corrp_tstat1.nii.gz, and the ic0002 stands for DMN for example. Thus I considered that there's a increased functional connectivity in DMN in patient's group. [Is that right?]
The next step I opened the melodic_IC.nii.gz, standard MNI152 space and the dr_stage3_ic0002_tfce_corrp_tstat1.nii.gz, together in the same fslview window, and I switch the volume of melodic_IC to volume 2 to correspond with the stage3_ic0002, set the min threshold = 0.95 and max threshold = 1. Now the question is that the PIONT with highest intensity (coordinate 52 76 36) is not in the DMN region. Why does it happen? Did I do something wrong or does it mean something else? Can I still consider that there's significant difference between two groups?
BTW: How can I do to correspond the coordinations x, y, z with structural, anatomical regions, not visually? I.e. is there any tool with an input of coordination can output a result of region?
Again a lot of questions. Any advise is appreciated.
Thanks in advance.
Jayci
|