Hi!
I'm doing a ROI-to-ROI wRRC analysis with two tasks (a,b) which were collected during a single fMRI run.
The ROIs are pre-defined from two meta-analyses (8 ROIs from one meta-analysis and 4 ROIs from the second meta-analysis).
I notice that the hierarchical clustering becomes different for the two tasks. This makes it hard to compare the results between the two tasks. Although, I guess this in itself is a sort of result. But how would I report it?
Do you recommend that I set the ROI-clustering manually in order to be able to easier compare results between the two tasks.
I can run a wRRC analysis on the contrast (a-b) and get results. I'm not quite sure how this happens, considering that the hierarchical clustering is different between the two tasks. I guess it's run on individual connections? If this question is too time-consuming to answer, is there a reference that explains these matters more in detail? I have read Jafri's article.
All the best.
Thomas.
Hi Thomas,
When you run the [-1 1] contrast comparing the two tasks the same clustering is applied to the data from both tasks (typically clustering is performed based on the similarity between ROIs in their patterns of connectivity with the rest of ROIs -averaged across all subjects and conditions included in your second-level analysis-). In general I would recommend using the clustering that resulted from the [-1 1] contrast (which used the data from both tasks) when displaying the results for each task separately (to do that, for example, in the [-1 1] results figure you can use the 'Save ROI order/groups to clipboard' option, and then in the individual-task results use the 'Load ROI order/groups from clipboard' option so all analyses use the same clustering results)
Hope this helps
Alfonso
Originally posted by Thomas Agren:
Hi!
I'm doing a ROI-to-ROI wRRC analysis with two tasks (a,b) which were collected during a single fMRI run.
The ROIs are pre-defined from two meta-analyses (8 ROIs from one meta-analysis and 4 ROIs from the second meta-analysis).
I notice that the hierarchical clustering becomes different for the two tasks. This makes it hard to compare the results between the two tasks. Although, I guess this in itself is a sort of result. But how would I report it?
Do you recommend that I set the ROI-clustering manually in order to be able to easier compare results between the two tasks.
I can run a wRRC analysis on the contrast (a-b) and get results. I'm not quite sure how this happens, considering that the hierarchical clustering is different between the two tasks. I guess it's run on individual connections? If this question is too time-consuming to answer, is there a reference that explains these matters more in detail? I have read Jafri's article.
All the best.
Thomas.
Hi Alfonso,
I'm dealing with a similar issue but with 2x2 repeated measures design (2 sessions: placebo and drug, in each there were 2 conditions: resting state and music listening). I use 32 networks as ROIs.
On what contrast should be the default clustering based in this case? There is no significant interaction effect with the standard setting for cluster-based inferences, while the drug effect is big. Is the contrast of all drug conditions - placebo conditions alright?
Many thanks,
Eva