視覺誘發電位(Visual evoked potential, VEP)被廣泛應用於臨床視覺檢查、 以及腦波人機介面的應用上。藉由針對視覺刺激進行不同頻率或相位的編 碼，可以利用分類器分辨不同視覺刺激源造成的腦波訊號，下達多種控制 指令。視覺腦波人機介面(VEP-based BCI) 透過非侵入式腦波訊號 (Electroencephalogram, EEG)的辨識與擷取，可以讓使用者與外界互動，並 且不需要他人輔助或肌肉移動。此外，視覺腦波具有高傳輸率與較少訓練 時間的優點，因此也獲得世界上許多腦波人機介面研究團隊的注意。但是 目前以視覺誘發電位為基礎的腦波人機介面，都屬於依賴性的系統 (dependent system)，使用者的眼睛必須注視於閃光刺激光源，才能進行系統 操作。這使得使用者的目光無法離開視覺刺激閃光，而造成使用上的限制。 因此，本研究致力於發展一種不具依賴性(independent system)的視覺誘 發電位腦波人機介面，藉由討論視覺誘發電位受到注意力調變增強的效應， 分辨受試者注意力所停留的選項，並進行腦波操控。 本研究所使用的閃光方式為閃爍視覺誘發電位(Flash Visual Evoked Potential, FVEP)，藉由人腦視覺腦波對於閃光刺激的亮滅具有時間鎖定與相 位鎖定的特性，以隨機編碼方式產生閃光序列，經由簡單的平均方法分離 出左視野與右視野所產生的閃爍視覺誘發電位，辨識出左右視野視覺誘發 電位的特徵峰值P2 與N2，並計算兩波峰之差值Amponset，經過比較振幅大 小後可以得出與視覺間注意力之間具有正相關性，本研究的成果將來可以 應用於非依賴性視覺腦波人機介面的應用上。 Visual evoked potential (VEP) has been widely used in clinical visual diagnoses and has been utilized for the application of brain computer interface. By encoding the temporal sequeces of visual stimuli with distinct frequencies or phases, brain waves induced from different visual stimuli can be recognized using classifiers, and the recognized brain waves are subsequently used for delivering control commands. Visual evoked potential – based brain computer interface (VEP-based BCI) enables users to interact with external environments independent of other people’s help or peripheral neuralmuscular activities. In addition, VEP-based BCI has the advantages of high information transfer rate and less training effort which has drawn attention from serveral BCI research teams. However, current VEP-based BCI systems are dependent system. Users’ eyes should always gaze at their intended visual stimuli which results in limitation to BCI applications. This thesis aims to develop a new independent VEP-based BCI system. By studying the effects of visuospatial attention on the modulation of VEP amplitudes. User’s attended targets can be distinguished from other targets, and the recognized target can then be used to control external devices. Owing to the time-locked and phase-locked characteristics of VEP, the present study utilized flash visual evoked potential (FVEP) to design our system. Two visual stimuli located at left and right visual fields were driven by distinct randomly generated sequences, and VEP induced from visual stimuli at left and right visual fields were detected separately by means of a simply averaging process. The amplitude difference between P2 and N2 peaks, denoted as AMPonset, was calculated in each VEP. Therefore, the visuospatial attention effect on user’s attended targets can then be discussed. The research results of this study can be applied to design independent VEP-based BCI system in future applications.
