See responses below.
> On 2 Dec 2014, at 9:38 am, Yann Quidé <[log in to unmask]> wrote:
>
> Dear Donald and other gPPI users,
>
>
> I am planning to investigate the relationship between 3 ROIs during a
> block designed task. I was thinking about using physio-physiological
> interaction or psycho-physio-physiological interaction analyses, but I am a
> bit confused by the exact way to handle this with gPPI, and by the
> interpretation of such analyses. I am using the PPPI_v2012_1_22 version of
> gPPI.
>
>
> To be sure I am on the good track, can anyone confirm my thoughts about
> these?
>
>
> 1) Conceptual questions:
>
> a- The physio-physiological interaction analysis will constrain my PPI
> analysis to the relationship between 2 (or more?) defined ROIs. Am I right?
>
>
No. The physio-physiological intereactions will should you where you have a
linear relationship with region 1, a linear relationship with region 2, and
where you have an interaction of region 1 and region 2 on other parts of
the brain.
> b- The psycho-physio-physiological interaction analysis will explain how
> two regions interact (A and B) to affect a third region C, right?
>
> No. This is an extension of (a) where the task interacts with the
phsyio-physio interaction. In the psycho-physio-physio, you have a linear
relationship with region 1, a linear relationship with region 2, and where
you have an interaction of region 1 and region 2, interaction of your task
with region 1, interaction of task and region 2, and interaction of task
with the interaction of region 1 and region 2 on other parts of the brain
(e.g. a third region C).
> In that case, are all possible interactions between these 3 regions valid?
> i.e. the relationship between A and B to affect C, but also the
> relationship between B and C to affect A, and the relationship between A
> and C to affect B?
>
> C is each voxel of the brain or the dependent variable in the model. Thus,
you don't have interactions between C and the other regions.
>
> 2) Methodological questions:
>
> a- For these analyses, I understand I need to introduce a second (and a
> third?) VOI in the wrapper, introduce their names in 'Region', choose the
> appropriate analysis ('psy' or 'psyphy'). However, while it seems clear for
> the physio-physiological interaction analysis, I am confused about the VOI2
> option for the psycho-physio-physiological interaction analysis. Should I
> create a VOI3 field?
>
> Only physio-physio and psycho-physio-physio interactions use the VOI2
option. You'd have to recode the gPPI toolbox for additional VOI terms.
It's worth noting that adding VOI terms increases the number of interaction
terms as well.
> b- For both analyses, how to feed the wrapper correctly?
>
> psyphyphy -- psycho-physio-physiological interaction
phyphy -- physio-physiological interaction
Both require VOI2.
> c- Once everything is set, what would be the outputs, and how should they
> be used? To be more clear, in the classic psycho-physiological interaction
> analysis the toolbox creates a contrast that can be used in a new GLM
> analysis for within/between groups investigations. But what will happen
> with the physio-physiological interaction and with the
> psycho-physio-physiological interaction analysis?
>
> I'm not sure that the contrasts are generated automatically for these
models. You might have to generate them yourself. Ultimately, once you
generate the contrasts, the second-level/group analysis is no different
than the other analyses.
> d- in the case of the psycho-physio-physiological interaction analysis, if
> any possible interaction (see question 1b) is possible, how do I know which
> one is which one?
>
> No. Only interactions between VOI1 and VOI2 are possible. C is the
dependent variable/brain data. PPI does not provide direction of
connections.
>
> I am sorry for these questions, but I could not figure out how to do it
> with the PPPI_Manual_v7_12 and want to be sure of what Im doing. Is there
> an updated version of the Manual that I missed somewhere?
>
>
The latest manual is posted on the nitrc website. It is not complete, but
the most common options have been described. If someone wants to add to the
manual, I'd appreciate the help.
>
> Thanks for your time!
>
>
> Cheers,
>
>
> Yann
>
>
>
