Análise de sinais EEG BCI 200 para classificação de intenção de movimentos: uma abordagem prática

Abstract

Brain Computer Interfaces (BCI) is becoming a vital field of engineering and biomedical computing. It uses electrical signals obtained from electroencephalogram (EEG) scans to deliver assistive technologies (AT) to humans. The objective of ICM is to interpret brain activity in digital form and act with commands in electromechanical devices, so that people unable to produce a motor response can communicate with a computer and obtain greater self-sufficiency, social inclusion and freedom. One of the main challenges in current BCI research is the extraction of features from random time-varying EEG signals and their classification as accurately as possible. Data extraction techniques, such as: power spectral density (PSD), spectral centroids, standard deviation and entropy are used in the filtering and investigation of the signals from two different mental exercises; The selected signals are classified using Linear Discriminant Analysis (LDA) and Common Spatial Patterns (CSP) algorithms. The best accuracy was achieved by power spectral density. The accuracies of this feature range from 75% to 99%, depending on the amount and quantity and quality of data samples collected. Finally, the translation algorithm will be built using selected and classified EEG features to control the ICM devices. Thus, this paper presents the results of analysis of specific movement intention EEG signals to extract the appropriate signal features, using statistical classification techniques and algorithms that can be employed to control ICM devices that can be used by people with disabilities or paralyzed.