[Eeglablist] ROIconnect: group-level plots

[Arnaud Delorme]

Hi Tyron,

> We have a couple questions about the MIM matrix and the roiconnect_plot
> functionality.
> 
> We have a study where we are looking at a control and a treatment group and
> we want to compare the functional connectivity (FC) between time points and
> groups. We are looking at aggregating the MIM matrix as an average for each
> group and were curious if that was allowed because, when you consider that
> you already averaged FC at each Hz level to get the value for each band,
> you are then taking an average of an average. If there are better ways to
> look at this type of data please let us know.

Yes, you could compute statistics between the two groups of matrices (and correct for multiple comparisons), as well as average them and compute the average between the difference. If you use parametric statistics, make sure to check the distribution of values.

> As part of that we would like to visualize a FC matrix (i.e. heatmap) of
> overall changes, but the pop_roi_connectplot function only has input for
> one EEG. As a workaround we saved the averaged freqxroixroi MIM matrix
> overwriting the MIM for that one and using POP_roi_connect plot, because as
> far as we can tell from the code it only is using the frequencies and ROI
> names which should be consistent from the EEG structure, but we are not
> MATLAB experts just intermediate. We wanted to reach out to confirm that
> this approach of aggregating the multiple MIM matrices was sound?

There are documented low level functions to plot connectivity matrices, in particular, 

roi_largeplot
roi_plotcoloredlobes
roi_plotbrainmovie
plotconnectivity

Best wishes,

Arno