Data-Driven Multichannel Signal Processing for Brain Computer Interfacing

脑机接口是将脑内信息传输到外界的一种技术，具有广泛的工程和医学应用。为了实现这一技术，我们必须建立信号处理的基本原理，用于从观察到的脑电图（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，我们关注到经验模式分解（EMD），这是一个最近众所周知的数据驱动的信号分解技术，我们的目标是建立方法，将这种有前途的技术应用到BCI的信号处理。由于这个项目，我们已经成功地在一个准确的分离心电图（EOG）的组成部分，从观察到的EEG。此外，我们已经成功地开发了一种方法，从脑电信号中提取必要的组件在运动想象任务。此外，我们还开发和实施了在线BCI平台，该平台利用Android智能手机和简单的机器人。

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.