EEG Signal Classification: Introduction to the Problem

Abstract

The contribution describes the design, optimization and verification of the off-line single-trial movement classification system. Four types of movements are used for the classification: the right index finger extension vs. flexion as well as the right shoulder (proximal) vs. right index finger (distal) movement. The classification system utilizes hidden information stored in the characteristic shapes of human brain activity (EEG signal). The great variability of EEG potentials requires using of context information and hence the classifier based on Hidden Markov Models (HMM). The suitable parameterization, model structure as well as training and classification process are suggested on the base of spectral analysis results and experience with the speech recognition. The training and the classification are performed with the disjoint sets of EEG realizations. Classification experiments are performed with 10 randomly chosen sets of EEG realizations. The final average score of the distal/proximal movement classification is 80%; the standard deviation of classification results is 9%. The classification of the extension / flexion gives comparable results.