Model-based approach to EEG classification

Nauka i Obrazovanie

S. L. Shishkin¹  A. G. Trofimov²  A. E. Ossadtchi³ 

[1] National Research Centre "Kurchatov Institute";National Research Nuclear University MEPhI;National Research University Higher School of Economics;

关键词: classification;    electroencephalogram, brain-computer interface;    sources of brain electrical activity;    equivalent current dipole;    EEG inverse problem;   

DOI  :  10.7463/0414.0705745

来源: DOAJ

【 摘 要 】

A method to construct a feature space for electroencephalogram (EEG) classification based on the localization of brain’s electrical activity sources is developed.

The purpose of the work is to show that a model-based approach to the construction of feature space for EEG classification allows us to achieve the accuracy comparable to existing classical approaches at the same time giving a number of opportunities to further increase it and having clear neurophysiological interpretation.

Experimental researches on real EEG show that the accuracy of the proposed method is comparable to the accuracy of the classical method of classification in brain-computer interfaces. The simplest statistical characteristics of dipole moments for equivalent current dipoles are chosen as features for classification, and the nearest neighbour algorithm is used for classification.

Application of the proposed algorithm is diagnostics of brain diseases and braincomputer interfaces.

The first section describes a method of modeling the EEG using equivalent current dipoles.

In the second section the statement of the EEG classification problem is formulated.

In the third section we propose a method of constructing a feature space for EEG classification based on the equivalent current dipoles characteristics.

The fourth section is dedicated to the experimental research of the proposed method on real EEG and to discussion of the results achieved.

【 授权许可】

Unknown