Data-Driven Multichannel Signal Processing for Brain Computer Interfacing

脑机接口是一种将大脑中的信息传递给外界的技术，在工程和医学上有着广泛的应用。为了实现这一技术，我们必须建立信号处理的基础，以便从观察到的脑电(EEG)中提取必要的成分和/或去除不必要的成分。在这个KAKENHI项目中，我们关注了经验模式分解(EMD)，这是一种最近广为人知的数据驱动信号分解技术，我们的目标是建立将这一有前途的技术应用于脑-机接口信号处理的方法。由于这个项目的结果，我们已经成功地从观察到的脑电中准确地分离出心电(EOG)分量。此外，我们还成功地开发了一种在运动想象任务中从脑电信号中提取必要成分的方法。此外，我们还开发和实施了利用Android智能手机和简单机器人的在线BCI平台。

Brain-computer interfacing is a technology that transmit information in the brain to the outside, has a variety of engineering and medial applications. To realize this technology, we have to establish fundamentals of signal processing for extracting necessary components from the observed electroencephalogram(EEG) and/or for removing unnecessary components. In this KAKENHI project, we paid attention into the empirical mode decomposition(EMD), which is a recently well-known data-driven signal decomposition technique, and we aimed at establishing methodologies that apply this promising technique to signal processing for BCI. As a result of this project, we have succeeded in an accurate separation of electrocardiogram(EOG) components from the observed EEG. Moreover, we have successfully developed a method for extracting necessary components from EEG signals during motor-imagery tasks. Additionally, we have developed and implemented on-line BCI platforms that utilize an Android smartphone and simple robots.