NECOM meeting : "About the information content of Local fild potentials non invasively estimated from the EEG " by Rolando Grave de Peralta (Brain Imaging Group)
Abstract: " Local field potentials (LFP) are produced by the electrical activity of large groups of neurons. The temporal structure of LFP codes, in the form of oscillations, reproducible information about neural processes. This information has been proven more efficient in predicting animal behavior or cognitive states than the information carried by the spike rate of individual neurons.
Here, we evaluate the information content (IC) conveyed by non invasively estimated LFP computed from the scalp EEG using the irrotational source model ELECTRA and compare it with EEG and intracranial potentials (IP). The IC is defined as the predictive power attained by each modality in a leave-one-out cross validation procedure. Following results were obtained from more than 25 subjects and several experimental paradigms.
1) In the vast majority of the cases the IC attained by LFP surpass the IC provided by the EEG. Even if the LFP are a “back-projection” of the EEG into the source space, the inversion procedure results in 3D images with increased discriminative power.
2) LFP and IP obtained for the same task have similar predicting power indicating that ELECTRA correctly retrieves the main attributes of the temporal activity of different brain regions and thus is a safe alternative to invasive procedures (as used in BCI).
In conclusion, LFP estimated non-invasively from the scalp EEG are as informative as intracranial potentials obtained by invasive procedures and have more discriminative power than the EEG used to compute it."