Serial LORETA volume calculation

Hi everybody!

I am looking to do serial sLORETA source localization on the same spike population in a single patient in repeat EEGs over time - and would like to calculate the volume of an sLORETA solution in each instance.

I understand that I may have to (1) threshold the solution (2) do a correction for channel baseline power, and (3) possible other corrections (e.g. head circumference changes over time).

Thus, how do I find out the number of voxels involved in the solution, and the number of voxels in the brain in total, and the voxel size?
Is it possible in Brainstorm?

Any suggestions?

Some of it possible in Brainstorm, but I'm not sure what approach I would recommend.

@Sylvain @John_Mosher @pantazis
Can you please share your recommendations regarding the statistical approach for this volumetric approach of inverse results?

Once you have identified a clear method, we (the software engineers) could help you implement it in Brainstorm.

A pragmatic approach would be to set an arbitrary threshold to apply to the sLORETA amplitude values to define the boundaries of the volume and apply that same threshold to the source model of each spike. You can then test whether this volumes changes from spike to spike.

Hi All,

Thank you for such a quick response!
So, if I understand it correctly- I need to know how many voxels there are in the head model in total, and how many of those voxels are involved in the solution. The solution needs to be thresholded in so the weakly activated ones don’t count. Threshold need to be applied consistently within patients and across patients. Since the amplitude of the spike determines in part the size of the field, we need to consider normalizing it for baseline band power?
Do I understand it more or less correctly?

You understand correctly. The threshold needs to be set in absolute (not %tage fraction of the maximum) value of the sLORETA statistics.

In this case- How to get this information?
(I'm a clinician, sorry for basic questions...)

how many voxels there are in the head model in total, and how many of those voxels are involved in the solution

In this case- How to get this information?

There is no solution for doing this from the interface or the functions of Brainstorm.
You would need to read the data structures from the files on the hard drive (source maps, head models, cortex surface, MRI), and compute the metrics of interest by yourself.
For this, advanced skills in Matlab programming are required, as well as good familiarity with the Brainstorm workflows.

To get started with Brainstorm scripting: