Eventos Anais de eventos
CONEM 2018
X Congresso Nacional de Engenharia Mecânica
Selection of Subject-Specific EEG Channels and Features for Online Performance with a BCI
Submission Author:
Gabriel Chaves de Melo , MG
Co-Authors:
Gabriel Chaves de Melo, Marco Antonio Meggiolaro
Presenter: Gabriel Chaves de Melo
doi://10.26678/ABCM.CONEM2018.CON18-0338
Abstract
A person with limited or complete absence of voluntary muscle control may find in a brain-computer interface (BCI) an alternative way to communicate with other people and interact with the environment. A non-invasive electroencephalogram (EEG) based BCI translates brain signals measured over the scalp into commands to a computer. One of the major problems in developing efficient BCI algorithms is the inter-subject variability of scalp recorded potentials. Spatial characteristics of brain mapping and spectral and temporal particularities of each person’s brain signals contribute to this issue. In this paper, a method for improving the accuracy on classification by choosing subject-specific features and channels in an EEG based BCI is proposed and compared with a publicly available dataset. The methods are later tested with a BCI system consisting of a commercial EEG headset and a microcontroller for simulating real time applications. The EEG signals considered in this paper are related to motor-imagery (MI), as it is in many publications in the field when aiming the actuation of robotic devices.
Keywords
Brain-computer interfaces, Electroencephalogram, Motor-Imagery
