Dear Jake
You don't mention whether this is DCM for fMRI or EEG / MEG.
If this is DCM for EEG / MEG, you can create models with different numbers of regions and compare the evidence (free energy).
If this is fMRI, you cannot compare models with different numbers of regions, so you will need to decide which regions you want in advance. Given you are only talking about 2 regions in each hemisphere, I would probably model both. However, if you hypothesise that both hemispheres are doing the same thing and are not interacting in a way that's important for testing your hypotheses, then it's fine to just model one hemisphere.
Do not model each hemisphere with separate models.
Best
Peter
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Jake Bowling
Sent: 13 October 2017 17:21
To: [log in to unmask]
Subject: [SPM] DCM Model Specification Bilateral vs. Unilateral
Hello, I am interested in modeling a network with 4 total VOI's, 2 analogous ones in each hemisphere. I am only interested in within-hemisphere connectivity. A question has come up in model specification as to whether we should:
1) Assume cross-hemisphere endogenous connections (leading to a model space with 4 VOI's that looks sort of like a square) or
2) Create separate DCM's for each hemisphere, 2 VOI's each
I have noticed that many DCM studies use unilateral models. Is it best practice to use DCM within-hemisphere only?
Thank you,
Jake
University of North Carolina-Chapel Hill
|