Chapter 9
Computational EEG Analysis
for Brain-Computer Interfaces
Garett D. Johnson and Dean J. Krusienski
Abstract EEG activity can be actively or passively modulated in a way to provide
commands to external devices. The feedback provided by interacting with the EEGcontrolled device creates a closed-loop system with the user in the loop. Such a system
is known as a Brain-Computer Interface (BCI). The selection of an analysis approach
for BCIs should be guided by the nature of the signals in consideration. This chapter
presents the most fundamental and widely-used EEG analysis techniques organized
by the type of control signal.
9.1 Brain-Computer Interfaces
9.1.1 Introduction
A Brain-Computer Interface (BCI) uses brain responses to deliberately-designed sensory stimuli or spontaneous mental activity to provide commands to external devices.
A block diagram of a typical BCI is shown in Fig. 9.1. The digitized signals are commonly preprocessed, which includes preemptive elimination of known interference
(i.e., artifacts) or irrelevant information, and/or the enhancement of spatial, spectral,
or temporal characteristics of the signal that are particularly relevant to the application. The preprocessed signals are passed to the feature extraction stage, which can
represent a variety of techniques for effectively isolating the relevant information in
the signals for BCI control. Commonly, more than one feature are extracted from the
signals and the resulting set of features for a given observation interval is processed
as a feature vector. This feature vector is then passed to the classifier (or regressor),
G. D. Johnson · D. J. Krusienski (B)
Department of Electrical and Computer Engineering, Old Dominion University,
Norfolk, VA, USA
e-mail: dkrusien@odu.edu
© Springer Nature Singapore Pte Ltd. 2018
C.-H. Im (ed.), Computational EEG Analysis, Biological and Medical Physics,
Biomedical Engineering, https://doi.org/10.1007/978-981-13-0908-3_9
193
Computational EEG Analysis
for Brain-Computer Interfaces
Garett D. Johnson and Dean J. Krusienski
Abstract EEG activity can be actively or passively modulated in a way to provide
commands to external devices. The feedback provided by interacting with the EEGcontrolled device creates a closed-loop system with the user in the loop. Such a system
is known as a Brain-Computer Interface (BCI). The selection of an analysis approach
for BCIs should be guided by the nature of the signals in consideration. This chapter
presents the most fundamental and widely-used EEG analysis techniques organized
by the type of control signal.
9.1 Brain-Computer Interfaces
9.1.1 Introduction
A Brain-Computer Interface (BCI) uses brain responses to deliberately-designed sensory stimuli or spontaneous mental activity to provide commands to external devices.
A block diagram of a typical BCI is shown in Fig. 9.1. The digitized signals are commonly preprocessed, which includes preemptive elimination of known interference
(i.e., artifacts) or irrelevant information, and/or the enhancement of spatial, spectral,
or temporal characteristics of the signal that are particularly relevant to the application. The preprocessed signals are passed to the feature extraction stage, which can
represent a variety of techniques for effectively isolating the relevant information in
the signals for BCI control. Commonly, more than one feature are extracted from the
signals and the resulting set of features for a given observation interval is processed
as a feature vector. This feature vector is then passed to the classifier (or regressor),
G. D. Johnson · D. J. Krusienski (B)
Department of Electrical and Computer Engineering, Old Dominion University,
Norfolk, VA, USA
e-mail: dkrusien@odu.edu
© Springer Nature Singapore Pte Ltd. 2018
C.-H. Im (ed.), Computational EEG Analysis, Biological and Medical Physics,
Biomedical Engineering, https://doi.org/10.1007/978-981-13-0908-3_9
193