Steady-state visual evoked potential (SSVEP) is widely used to design a brain-computer interface (BCI). If the user wants to enter the command at any time, it is necessary to extract user's attentional behavior for visual stimulus. In this study, we aimed to improve the performance of SSVEP-based BCI by classifying user's state into gazing or resting. The canonical correlation analysis was used to extract the features of SSVEP. Three-class discrimination of non-gaze and two kinds of gaze was performed by two-stage Fisher linear discriminators. Feature space was evaluated by within-class variance between-class variance ratio. Experiments were conducted using a wireless EEG system by changing the frequency combination of visual stimuli. The results showed that the separation performance was improved by setting the stimulation frequency and its harmonics to avoid 10 Hz which is the center of the alpha band, resulting in high accuracy of 88.3%.
We developed a low-cost, event-related desynchronization (ERD) measurement system using consumer-grade EEG system (Emotiv Epoc) for application of motor-imagery (MI) related brain machine interface (BMI). We investigated accuracy of ERD power detection of Epoc by comparing the ERD power calculated from simultaneously recorded EEG via Epoc and medical-grade EEG system (g.USBamp: g.tec medical engineering). Eleven healthy participants performed MI of holding a rolling tennis ball and opening the grasped hand under the observation of hand movement video. The mean ERD power was comparable between Epoc and g. USBamp with both hand gestures tested. ERD detection accuracy of Epoc was 70.5% of the all trials tested. Considering the price of Epoc, our results suggest that Epoc could be a good substitute for medical-grade EEG systems for the purpose of MI-based BMI.
近年，脳情報を入力としてユーザの意図を反映するBrain Computer Interface（BCI）の研究が盛んに行われており，麻痺患者等のコミュニケーション支援システムとして期待されている．本研究では，近赤外分光法（NIRS）を用いて前頭前野の脳活動を計測し，得られた脳情報によってカーソルコントロールを行うBCIを作成し，その有効性について検証を行った．
近年の非侵襲的脳機能計測技術の発展を受け，人間の脳機能の解明が進んでいる．これらは神経経済学といった経済・経営活動の解明にも利用され，今後の実用化が期待されている．本研究では応用例として，商品選好による購買意思決定支援のためBrain Computer Interface（BCI）を試作した．さらに，試作したBCIを用いて実験を行い，その有効性および可能性について検討した．
[Purpose] Evaluating the effect of brain-computer interface (BCI)-based functional electrical stimulation (FES) training on brain activity in children with spastic cerebral palsy (CP) was the aim of this study. [Subjects and Methods] Subjects were randomized into a BCI-FES group (n=9) and a functional electrical stimulation (FES) control group (n=9). Subjects in the BCI-FES group received wrist and hand extension training with FES for 30 minutes per day, 5 times per week for 6 weeks under the BCI-based program. The FES group received wrist and hand extension training with FES for the same amount of time. Sensorimotor rhythms (SMR) and middle beta waves (M-beta) were measured in frontopolar regions 1 and 2 (Fp1, Fp2) to determine the effects of BCI-FES training. [Results] Significant improvements in the SMR and M-beta of Fp1 and Fp2 were seen in the BCI-FES group. In contrast, significant improvement was only seen in the SMR and M-beta of Fp2 in the control group. [Conclusion] The results of the present study suggest that BCI-controlled FES training may be helpful in improving brain activity in patients with cerebral palsy and may be applied as effectively as traditional FES training.
We investigated the transition of event-related desynchronization (ERD) intensity through 6 days of brain-machine interface (BMI) training using Digital Mirror Box (DMB), which is a potential rehabilitation system for stroke patients with hand paralysis. Eleven healthy participants performed motor-imagery of grasping their dominant hand under the observation of hand movement video. The overall ERD strength of all participants showed a significant increase from day 1 to day 6. When the participants were divided into high and low ERD groups by their initial ERD strength of higher or lower than 0% at the initial measurement before BMI training, respectively, participants in low ERD group showed larger training effect. These results suggest that BMI training is particularly useful for increasing the ERD strength of stroke patients who have lower ERD strength.
脳信号を用いて，コンピュータを中心とした外部機器の操作を行う仕組みをBrain-Computer Interface (BCI)という．BCIは，ユーザーの脳波を計測・判別することで動作するが，脳波のS/N比は非常に悪いため，判別に先立ってノイズを低減するための前処理が必要である．本稿では，多チャンネルの信号源分離によって，BCIにおける判別に有効な成分を抽出する手法について検討する．