Dear CONN toolbox experts,

I am interested in analyzing the connectivity within different networks (DMN, DAN, ECN and salient network). However, I don't understand how I should carry out the analysis. Would it be more appropriate to do a seed to voxel or a roi to roi analysis?


In the case of roi to roi analysis, I understand that I could select just the rois of a particular network and analyze each network separetely.

However, in the case of seed to voxel analysis, I don't know how I should select the seeds/sources.
Could I select them individually? For example, first select the MPFC of the DMN and see the connections to these regions, then select the LP of the DMN and see the connections to these regions, etc. Or should I select all the seeds from the same network at the same time?

The problem is that if I select various seeds at the same time, then I dont know to which seed the significant association corresponds. For example, if I select the PPC (L) seed I find a significant negative connection to the inferior frontal gyrus. However, if I select at the same time the seeds of PPC (L) and the LPFC (L), I get significant positive connections with the subcallosal cortex, but I do not know how to see which of both seeds is the one connected with this subcallosal area. I do not understand how should I carry out the analysis and how to interpret them.

Thank you very much in advance.

Best regards,

Agurne

Dear Agurne,

All are perfectly reasonable options, but if you select the seeds individually then you should apply an additional multiple-comparison correction (e.g. bonferroni on cluster-level p- threshold) to take into account the multiple analyses that you are performing. Instead, the more standard approach is to select all seeds simultaneously (because this allows you to make inferences about the presence of effects in any of the selected seeds), and then use post-hoc analyses to determine which among the selected seeds may be responsible for each of the effects that you are finding (for example by clicking on the 'plot effects' button -the button with an image of a barplot-). These post-hoc analyses allow you to better characterize and interpret the effects that you are finding (e.g. is the effect present across all of the selected seeds or just among a subset?), but restricted only to clusters that were already significant in the primary analysis. 

Best
Alfonso

Originally posted by Agurne Sampedro:

Dear Alfonso,

Thank you very much for your answer.

I did the seed to voxel analysis with the four seeds from the dorsal attention network and the rest of the brain, obtaining a significant cluster in the inferior temporal gyrus. I clicked on the plot effects button to see better the results, as suggested. Results showed that the bars of the time1 (pre) were lower than the bars of time2 (post) in all seeds of the dorsal attention network. Does this mean that the connectivity between the inferior temporal gyrus and all the seeds of the dorsal attention network was greater in the post than in the pre? But is this significant in all the seeds of the network? Is there any way to interpret whether this is significant in all areas of the network?

And on the other hand, when I clicked on plot effects, I got F(4 9) = 8.91 p(FDR) = 0.003. What do these values mean? They are different from the ones that appear in the first window before clicking on plot effects.

Thank you very much.

Kind regards,

Agurne