To whom it may concern,
I am planning to use permutation statistics to analyze the distribution (at sensor level) of an EEG-based metric distinct from power (a ratio between the power of distinct frequency bands). This would the same steps of analyzing the topographical distribution of a sole frequency band: doing permutation stats between the two groups of interest and do cluster-based correction for multiple comparisons. Therefore, I wonder if I could use brainstorm algorithms on this different metric. Is there a specific way in which I could structure the data and import it so that I can use brainstorm algorithms?
Thank you in advance for your help,
Aimee Flores
I wonder if I could use brainstorm algorithms on this different metric.
Yes, you can use the non-parametric tests available in Brainstorm or FieldTrip with any metric. These approaches do not assume anything about the distribution of the values, and can therefore be used on any kind of data.
https://neuroimage.usc.edu/brainstorm/Tutorials/Statistics#Nonparametric_permutation_tests
Is there a specific way in which I could structure the data and import it so that I can use brainstorm algorithms?
The easiest is maybe to first create some template files in the Brainstorm database which are similar to the files you want to compute (same dimensions of the data matrix and same expected displays), by importing data and processing it with the Brainstorm GUI. Then replace the existing data with your own metric (export to Matlab/modify/import back, or directly modify the .mat files), and run the statistical test on the modified files.
The data structures are documented in the sections "On the hard drive" of the tutorials, and the tutorial Scripting will give you a lot of information on how to manipulate Brainstorm files and structures:
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting#File_structures
https://neuroimage.usc.edu/brainstorm/Tutorials/Scripting#Custom_processing