Hello experts,

I have two questions:

1. I'm currently using 13.h with SPM8 and Matlab R2009b, and got this following error message in the preprocessing stage:

--------------
Step 2/4: Preprocessing functional data
??? Index exceeds matrix dimensions.

Error in ==> conn_designmatrix at 48
        if isfield(X1,'xyz'),for
        n0=1:size(x,2),xyz{end+1}=X1.xyz{idx}+(n0-1)*(confounds.deriv{n1}+1);end;end

Error in ==> conn_process at 623
                X{nses}=conn_designmatrix(CONN_x.Preproc.confounds,X1{nses},X2{nses});
--------------

It all worked well at first, but this error appeared after I reduced the number of ROIs used.
(It seems that this error occurs when the number of ROIs is smaller than the number of covariates.)
Could you please figure out what's going on here?


2. Is it valid to use CONN in event-related designs? 


Thank you!

Regarding question 1, thanks for reporting this, could you please try the attached patch and let me know whether this fixes the issue? (simply save this file in the conn distribution folder overwriting the file with the same name there) 

Regarding question 2, the operations implemented in the conn toolbox are designed and validated only for resting state and block designs. While it is possible to use the conn toolbox on event related data the optimal way to do this and the interpretation of the results is still open to debate. If you are not interested in task-related effects but rather would like to use your event-related data to infer 'resting-state' functional connectivity I would recommend the approach described in Fair et al. (2007, 'A method for using blocked and event-related fMRI data to study “resting state” functional connectivity'), where one simply regresses out the event-related effects (as 'confounds') and uses the residual BOLD timeseries as a 'resting state' scan (e.g. not analyzing task-related connectivity but instead creating a new 'rest' condition containing all of the scans, entering the main effects of the true task conditions as 'confounds' and analyzing the functional connectivity only on the 'rest' condition). If instead you would like to obtain 'task-related' functional connectivity measures from event-related data, you could still use the default approach implemented in the toolbox to analyze your data but the resulting measures may not be directly interpretable as or comparable to standard functional connectivity measures (the interpretation of the resulting measures varies depending mainly on whether this is a slow or a fast event-related design, whether the main task effects are entered as confounds and the form of temporal weighting used), and to my knowledge there are no papers describing or validating this general approach, so I would be cautious when using this approach or interpreting these results until there is a better consensus on the proper methodology and interpretation of these measures.

Best
Alfonso

Originally posted by Hongmi Lee:

Works now. Thank you for the advice!