I have a set of ERD/ERS data for different subjects groups and different conditions and would like to compare the source estimates for these reactions between groups. I initially expected to use LORETA for this, but the ERD/ERS values themselves (in contrast to the ERP) are not the correct entries for the LORETA analysis, as it was developed for the electric potential data, and not the spectral power relative modulations. (By the way, the BrainVision Analyzer2 software somehow calculates the sources for ERD/ERS with implemented LORETA algorithm and the results seem quite meaningful, but I'm not sure if I can trust it.) The LORETA suggests an option for time-frequency source analysis, but it would require coming back to the epoched raw EEG, calculating cross-spectra etc. For the whole dataset, it would be time- and computationally consuming and laborous and in a way will repeat the work which is already done. So, could anybody suggest a method (preferrably, realized in some free software), which could use the ERD/ERS time series for the set of EEG electrodes to calculate the cortical origin of these frequency-specific activity modulations? Or the task is actually ill-defined and I anyway have to start the work with single-trial EEG?
Due to time constraints I can not give an extensive account, but as far as I know one of the best methods to do this is to use distributed 3D source modeling based on multiple sparse priors (e.g. implemented in SPM8 for MATLAB). For each participant you create a forward model (e.g. using three-shell boundary element method (BEM)), then compute source estimates using multiple sparse priors (first see, Friston K, Henson R, Phillips C, Mattout J (2006) Bayesian estimation of evoked and induced responses. Hum Brain Mapp 27:722–735. CrossRef Medline; then: Friston K, Harrison L, Daunizeau J, Kiebel S, Phillips C, Trujillo-Barreto N, Henson R, Flandin G, Mattout J (2008) Multiple sparse priors for the M/EEG inverse problem. Neuroimage 39:1104 –1120.).
I hope this short account can help you. best, Sebastian
I would suggest to use LORETA, even if some spectral analysis calculation have to be redone. LORETA calculates cross-spectra rapidly from large amount of raw EEG data (I use eLORETA). I would try thiis way: calculate the power spectrum for the "rest/reference" epochs and find the source of the rhythm of your interest. Then do the same for the "action" periods. Finally use the non parametric statistical mapping routine, included in the eLORETA package, to compare the sources. You shoud obtain a statistical map showing spectral differences between the two conditions, so something related to the source of ERD/ERS.
I did this in Fieldtrip using Dynamic Imaging of Coherent Sources (DICS) as my inverse model and FEM as my forward model (https://www.fieldtriptoolbox.org/tutorial/beamformer/). The tutorial covers most of the things needed. I used Brainettome atlas to further average the activations ( which were subtracted from baseline) to average ERD in 20 ROIs
- Badges
- Science topic