On the issue of low-frequency EEG generators and methods of their spatial localization.

BACKGROUND: The article reviews the theoretical foundations of low-frequency EEG generation and the problems of existing methods and algorithms for localizing low-frequency EEG NEW METHOD: A new method based on spatial filtering for signal recovery from a specific, pre-set point in space (area of interest) is presented. The algorithm uses the dynamics and correlation of signals in EEG leads with the addition of artificially generated data. The method assumes a volume charge (unipole) as a source and a function of linear potential decay from values inverse to the distance from a given point to scalp electrodes.

RESULTS: It is shown that the dipole model based on the summation of postsynaptic potentials cannot adequately describe the low-frequency component of the EEG; arguments are given in favor of the unipolar model based on the summation of trace potentials.

COMPARISON WITH EXISTING METHOD: Based on deep brain stimulation data, the new method was verified and its effectiveness was compared with some existing algorithms (sLORETA, dSPM, etc., included in the Brainstorm package).

CONCLUSIONS: As a result, it is shown that the new "Virtually implanted electrode" method makes it possible to accurately determine the localization of unipolar current sources. At the same time, it is shown that other methods based on the dipole model gave false solutions. In this regard, it is necessary to revise the results of localization of the low-frequency component of the EEG previously obtained using these tools, especially when evaluating functional relationships based on these data.