周波数標識された定常体性感覚誘発電位に基づくBMI

摘要

In this study, the right and left index fingers of the subjects were constantly stimulated by a mechanical vibration device to examine the possibility of an asynchronous brain machine interface (BMI) based on steady-state somatosensory evoked potentials (SSSEPs). In the experiment, the subjects were instructed to attend on the finger indicated by the visual que. EEG signals were recorded from electrodes on the entire head. First, the narrow band pass filters which pass the same frequency of the stimulation were applied. Secondly, principal component analysis (PCA) and independent component analysis (ICA) were applied to obtain the weight vectors of the filtered SSSEPs. Lastly, the time courses of independent component (IC) scores were reconstructed. The reconstructed IC for which the subject attended on tended to be increased. The result demonstrates that the presented methods were useful for detecting the SSSEPs and a BMI based on SSSEP may be promising.%本研究では，異なる周波数で標識した20Hz近傍の機械振動刺激を左右の示指に定常的に与え，視覚刺激によりいずれかに注意を向けるか指示する課題により，定常体性感覚誘発電位（SSSEP）に基づくブレインーマシン・インタフェースの検討を行った．実験では視覚刺激によってどちらの示指に注意を向けるかを教示した．EEGは頭部全体に配置した電極で記録した．解析は，機械振動刺激で与えているものと同一の周波数が通過する狭帯域フィルタを通した．次に，SSSEP抽出のための荷重ベクトルを求めるため，主成分分析および独立成分分析を行った．最後に，独立成分の時間波形を再構成した．被験者が注意を向けた場合，その時間波形の振幅が増大した．この結果は，本手法がSSSEPを検出することに有用であり，SSSEPに基づくBMIの構築の可能性を示唆している．