Hi,

I know that mri_glmfit and mri_surfcluster can be used to threshold activity on the surface using a monte carlo simulation, but I was wondering if it is possible to do this on data that has been constrained by an anatomical label.  

I think this could be described as the Monte Carlo analog of tksurfer's FDR "marked label only" option.

The documentation lists 3 steps for the more general case:

  1. Estimation: run the analysis on your data without simulation. Note: at this point you can view your results with FDR thresholding in tksurfer. FDR is often conservative relative to cluster-based thresholding.

  2. Simulation: run the simulator with the same parameters as the estimation to get the Cluster Simulation Data (CSD).

  3. Clustering: run mri_surfcluster, passing it the CSD from the simulator and the output of the estimation. These programs will print out clusters along with their p-values.

Where in this flow, and how, could func2roi be used to limit everything to activity taking place in an ROI?  

I have a guess, but I think it is at best overcomplicated and worst not possible to do (this way).

1. Run mkcontrast and stxgrinder to generate the significance maps for the contrasts, given the list of subjects.
2. Using mris_preproc, resample the data onto average7
    This generates mgh files.
3. Run func2roi to constrain the contrasts, for subject average7, using the ROI labels.
    I'm not so sure about this step.  Can this input be specified to func2roi?  Is it possible to do this?
    The output side of this step is a little more clear than the input.  One of the ouputs of func2roi are bfloat files.  
    bfloat: "This is where the hemodynamic averages (as computed by selxavg-sess) averaged within the ROI are stored."
4. Using mri_surf2surf, convert the bfloat files to mgh files.                                                                                         
5.  Create a group descriptor file to compare between groups.
    Is this done manually, or using a specific program?
6. Run mri_glmfit on the mgh files to estimate at a predefined significance level.
7. Run mri_glmfit again on the same files to simulate random null hypothesis data.
8. Run mri_surfcluster to generate clusters at a p-value based on a monte carlo algorithm using the null data vs. 'real' hypothesis data.

Thanks in advance,
Rob 
-- 

Robert P. Levy, B.A.
Research Assistant, Manoach Lab
Massachusetts General Hospital
Charlestown Navy Yard
149 13th St., Room 2611
Charlestown, MA 02129
email: levy@nmr.mgh.harvard.edu
phone: 617-726-1908
fax: 617-726-4078

http://nmr.mgh.harvard.edu/manoachlab