二状態で変化がある脳波共分散行列の同時対角化に基づく信号源推定法

Abstract

運動時, 安静時等, 二状態間で変化がある脳波(EEG)信号源の推定法として, 各状態の共分散行列を同時対角化し, その差分を用いて推定する, 同時対角化ビームフォーマを提案する.β律動の増大信号源を一次運動野に配置した際のEEGをシミュレーションにより作成し, 提案手法および従来法であるミニマムバリアンスビームフォーマにより信号源推定して性能の比較を行った.その結果, 提案手法の信号源推定の誤差が10mm以下となる信号対雑音比の下限値が, ミニマムバリアンスビームフォーマの信号対雑音比の下限値の1/3より小さく, 二状態間で変化がある信号源の推定に適用可能であることが示唆された.

We proposed a signal source estimation method based on simultaneous diagonalization of electroencephalogram (EEG) covariance matrices of two different states such as movement and resting states. We compared performances of the proposed method and conventional minimum variance beamformer them to simulation data in which an increasing signal source of β rhythmic spontaneous EEG activity was located in the primary motor cortex. As a result, it was found that the lower limit of signal-to-noise ratio of the proposed method, in which location bias became lower than 10 mm, was one third of that of minimum variance beamformer. This result indicates that indicates that the proposed method is applicable for signal source estimation that has two different states.