以中央腦區擷取移動想像和穩態視覺誘發電位特徵提昇複合式腦機介面成效

Abstract

複合型腦機介面是近年來在腦機介面領域中新興的一項熱門研究，和傳統腦機介面相比，其特色為結合了至少二種以上腦機介面指令，不僅能增加特徵資訊的擷取，也能有效提升正確率表現，更能降低腦電波電極量測數量以減少計算複雜度。 本篇研究以複合型的智慧型腦機介面為基礎，提出了一種結合移動想像與穩態視覺誘發電位的複合型腦機介面系統，並且導入一項重要的演算法稱為共同頻域模式，其發展來自於共同空間模式，是一種有效且被證實的特徵擷取方法。其優勢為:只需要擷取兩個中央電極 (C3 與C4 電極)，就可以有效維持97%的正確率。實驗證明，共同頻域模式可以突破共同空間模式的缺點，即需要多腦波電極點的量測限制，且未來可結合目前流行的穿戴型智慧系統，讓腦機介面實際應用於生活層面中。

Recently hybrid brain computer interface (BCI) has received increasing attention recently, which consists of at least two different systems, is taken more seriously. Compare to a conventional BCI, the advantages of a hybrid BCI are information-rich in recognizing the specific patterns and channel reduction. In this study, we propose a hybrid BCI system combining two of common BCI approaches: motor imagery (MI) and steady state visual evoke potential (SSVEP) to develop a hybrid BCI system based on only two EEG channels (C3 and C4). This system introduces an important method called common frequency pattern (CFP), which is an extension of common spatial pattern (CSP). Experimental results show that the purposed hybrid BCI system with only two EEG channels can maintain the same performance to the 32 EEG channels. This study reports also shows that the hybrid BCI outperformed conventional BCI.