Effectiveness of hierarchical Bayesian source current estimation on EEG-based brain-computer interfaces
暂无分享,去创建一个
In recent years, interest on brain-machine interface (BMI) has been rising. Brain-machine interface is a technology for paralyzed people who cannot move their bodies due to an accident or a disease. The main goal of this technology is to allow paralyzed people to interact with the society more freely by giving them control over an external device such as a computer cursor or a robotic arm. Brain-machine interface can be broadly divided into invasive BMI and non-invasive BMI. Motor imaginary paradigm is most popularly used in the non-invasive BMI using EEG signals. The motor imaginary paradigm uses a phenomenon that if a subject imagines a lime movement, the oscillation of a specific frequency range of EEG signals measured over the related brain part decreases. In most cases, subjects participate in the motor imaginary paradigm experiment in the state that they did not get the feel of motor imaginary. Therefore, in order to give subjects getting the feel of motor imaginary and improve the accuracy of the motor imaginary paradigm, we made a real-time motor imaginary training system based on the hierarchical Bayesian estimation method. First of all, currents over each cortical area are estimated in this system. Then, a few important currents over classifying imaginary left and right finger movement are selected by the sparse classifier model. By plotting the changes of the selected currents as bars on the monitor, we made the subjects can get the feel of imaginary movement by themselves. In this study, the difference of EEG signals measured while showing feedback to the subjects and EEG signals measured without feedback will be dealt with.