Textbook Computational Eeg Analysis Methods and Applications Chang Hwan Im Ebook All Chapter PDF
Textbook Computational Eeg Analysis Methods and Applications Chang Hwan Im Ebook All Chapter PDF
Textbook Computational Eeg Analysis Methods and Applications Chang Hwan Im Ebook All Chapter PDF
https://textbookfull.com/product/nonlinear-analysis-problems-
applications-and-computational-methods-zakia-hammouch/
https://textbookfull.com/product/advances-in-computational-
techniques-for-biomedical-image-analysis-methods-and-
applications-1st-edition-deepika-koundal/
https://textbookfull.com/product/deterministic-flexibility-
analysis-theory-design-and-applications-1st-edition-chuei-tin-
chang/
https://textbookfull.com/product/computational-methods-for-
single-cell-data-analysis-guo-cheng-yuan/
Spatial analysis with R: statistics, visualization, and
computational methods Second Edition Oyana
https://textbookfull.com/product/spatial-analysis-with-r-
statistics-visualization-and-computational-methods-second-
edition-oyana/
https://textbookfull.com/product/computational-methods-for-
nanoscale-applications-particles-plasmons-and-waves-igor-
tsukerman/
https://textbookfull.com/product/engineering-optimization-
applications-methods-and-analysis-first-edition-rhinehart/
https://textbookfull.com/product/computational-methods-for-
processing-and-analysis-of-biological-pathways-1st-edition-
anastasios-bezerianos/
https://textbookfull.com/product/computational-methods-for-deep-
learning-theoretic-practice-and-applications-wei-qi-yan/
Biological and Medical Physics, Biomedical Engineering
Chang-Hwan Im Editor
Computational
EEG Analysis
Methods and Applications
Biological and Medical Physics, Biomedical
Engineering
Editor-in-Chief:
Elias Greenbaum, Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA
Editorial Board:
Masuo Aizawa, Department of Bioengineering, Pierre Joliot, Institute de Biologie
Tokyo Institute of Technology, Yokohama, Japan Physico-Chimique, Fondation Edmond
Olaf S. Andersen, Department of Physiology, de Rothschild, Paris, France
Biophysics and Molecular Medicine, Lajos Keszthelyi, Institute of Biophysics, Hungarian
Cornell University, New York, USA Academy of Sciences, Szeged, Hungary
Robert H. Austin, Department of Physics, Paul W. King, Biosciences Center and Photobiology,
Princeton University, Princeton, New Jersey, USA National Renewable Energy Laboratory, Golden, CO,
USA
James Barber, Department of Biochemistry,
Imperial College of Science, Technology Robert S. Knox, Department of Physics
and Medicine, London, England and Astronomy, University of Rochester, Rochester,
New York, USA
Howard C. Berg, Department of Molecular
and Cellular Biology, Harvard University, Gianluca Lazzi, University of Utah, Salt Lake City, UT,
Cambridge, Massachusetts, USA USA
123
Editor
Chang-Hwan Im
Department of Biomedical Engineering
Hanyang University
Seongdong-gu, Seoul, South Korea
This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface
v
vi Preface
Part I Introduction
1 Basics of EEG: Generation, Acquisition, and Applications
of EEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Chang-Hwan Im
Part II Methods
2 Preprocessing of EEG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Sung-Phil Kim
3 EEG Spectral Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Do-Won Kim and Chang-Hwan Im
4 The Analysis of Event-Related Potentials . . . . . . . . . . . . . . . . . . . . 55
Marco Congedo
5 EEG Source Imaging and Multimodal Neuroimaging . . . . . . . . . . . 83
Yingchun Zhang
6 Methods for Functional Connectivity Analysis . . . . . . . . . . . . . . . . 125
Jeong Woo Choi and Kyung Hwan Kim
vii
viii Contents
ix
x Contributors
xi
xii Abbreviations
Chang-Hwan Im
C.-H. Im (B)
Department of Biomedical Engineering, Hanyang University, Seoul, South Korea
e-mail: ich@hanyang.ac.kr
to the scalp surface at some specific locations or fitted in a cap (or a net) for more
convenient attachment.
The main generators of the EEG, often referred to as EEG sources, are cortical
neurons. It is well-documented that most neurons in the human brain are concen-
trated within the cerebral cortex, which is a thin sheet of gray matter with 2–4 mm
thickness. The apical dendrites of the cortical neurons, often referred to as large
cortical pyramidal neurons, are arranged almost perpendicularly to the surface of
the cerebral cortex. Therefore, the direction of the neuronal current flowing along
the long apical dendrites of cortical pyramidal neurons also becomes perpendicular
to the cortical surface [10, 22]. This physiological basis can be used as an important
constraint for EEG source imaging [1], which will be introduced in Chap. 5.
There are two different sorts of intracellular potentials that may potentially con-
tribute to the generation of scalp EEG signals, which are an action potential and a
postsynaptic potential. The action potential is elicited by sudden changes in trans-
membrane resting potential due to the dynamic movements of intracellular and extra-
cellular ions, such as sodium, chloride and potassium ions. When the action potential
within a neuron propagates to a synapse, a small gap junction between two neurons,
the postsynaptic potential is generated across a pair of neighboring neuronal mem-
branes. If the postsynaptic potential exceeds a threshold level, the action potential of
one neuron is delivered to the other neuron (see Fig. 1.1).
Among the two different types of potentials, the postsynaptic potential is believed
to contribute more to the generation of measurable extracranial electric fields than
the action potential. This is because the action potentials do not fire synchronously
in a large number of neurons [25]. On the contrary, although the magnitude of the
postsynaptic potential is generally smaller than that of the action potential, its rela-
tively longer duration (~30 ms) enables synchronous generation of the postsynaptic
potentials in a large number of neurons (see Fig. 1.1). As aforementioned, since the
apical dendrites of cortical pyramidal neurons are arranged almost perpendicularly
to the cortical surface, the summation of the synchronously generated postsynaptic
Fig. 1.1 (Left) Comparison of waveforms of action potential and postsynaptic potential. (Right)
Synchronous occurrence of postsynaptic potentials can produce unidirectional primary current flow
large enough to be recorded outside the head
1 Basics of EEG: Generation, Acquisition, and Applications of EEG 5
potentials in a small cortical area can induce extracranial electric fields large enough
to be measured on the scalp surface [1]. According to Hämäläinen et al. [10], the
current density on the cortical surface is approximately 100 nA/mm2 . When numer-
ous cortical neurons within a small area are activated synchronously, a unidirectional
neuronal current flow is formed. Figure 1.1 depicts the comparison between action
potential and postsynaptic potential, as well as a schematic illustration of the gener-
ation of the unidirectional neuronal current flow.
The unidirectional neuronal currents, which can be approximately modeled as
equivalent current dipoles (ECDs) in EEG source imaging problems [6] (see Chap. 5
for more details), are called primary or impressed currents [22]. Since the human
body is filled with electrically conductive media, the extracellular currents induced by
the primary currents can flow even to the farthest part of the human body. These extra-
cellular currents are known as secondary, volume, or return currents [22]. According
to the electromagnetic theories, the flow of the secondary currents results in nonuni-
form potential distributions on the scalp. The measurement of the potential difference
between two distant scalp locations over time is the EEG.
Because the EEG measures dynamic changes in potential differences originating
from the secondary current flows, precise evaluation of conductivity profiles of the
volume conductors, i.e., different tissue compartments inside the head, is important,
not only to understand the underlying mechanisms of the EEG, but also to build a
precise head model to calculate electric field quantities generated by primary neu-
ronal currents (this process is called forward calculation). A human head can be
roughly modelled with four different regions: brain, cerebrospinal fluid (CSF), skull,
and scalp. Table 1.1 shows the typical conductivity values when the conductivity of
each region is assumed to be isotropic (having uniform conductivity in all directions)
and homogeneous [9]. The most notable point in the conductivity profile shown in
Table 1.1 is that the conductivity value of the skull is even smaller than those of the
other tissues. Because of the poor electrical conductivity of the skull, the secondary
currents are severely distorted and/or attenuated before they are delivered to the scalp
surface. Since the tissue conductivity is an important factor affecting the reliability
and accuracy of EEG source imaging, anisotropic conductivity characteristics are
sometimes considered. For example, the skull has an anisotropic conductivity prop-
erty, approximately 0.014 and 0.0107 S/m for the directions normal and tangential to
the skull surface, respectively [2]. White matter tissues also have an anisotropic con-
ductivity property: the white matter conducts secondary currents much better along
a fiber direction than in its transverse directions [31]. In practice, however, a rough
approximation of the human head structure as piecewise isotropic and homogeneous
volume conductors (e.g., brain, CSF, skull, and scalp) is most widely used. More
detailed discussion of this topic is provided in Chap. 5.
6 C.-H. Im
Table 1.1 Typical conductivity values for different brain tissues/regions [9]
Regions Absolute conductivity (S/m) Relative conductivity
Brain 0.22 1
CSF 1.79 8
Skull 0.014 1/16
Scalp 0.22 1
Initial analog EEG devices recorded ongoing EEG activities on printed paper, when
no quantitative EEG analysis was possible. Nowadays, owing to the development of
computer technology and digital engineering, EEG signals are stored in computers as
sampled numeric data. The use of a digital EEG enables us to utilize a variety of com-
putational EEG analysis technologies, such as time-frequency analysis, functional
connectivity analysis, and source imaging.
To record EEG data, at least two electrodes must be used, because EEG measures
the potential difference between two distant scalp locations. Recent EEG recording
devices allow simultaneous recording of EEG signals from many scalp locations.
There are two types of EEG recording methods: bipolar and unipolar methods. In
the bipolar method, electrodes are all paired, and the potential differences between
each pair of electrodes are recorded. In the unipolar (or monopolar) method, the
potential differences between each electrode and a reference electrode are recorded.
Theoretically, the reference electrode in unipolar recording can be positioned any-
where; however, because the distribution of potential difference on the scalp surface
varies according to the location of the reference electrode, average reference is fre-
quently used. Average-referenced potential of each electrode can be readily evalu-
ated by subtracting the average of all electrodes from the potential difference of each
electrode. Average reference is particularly useful in depicting spatial distributions
of potentials on the scalp surface, usually referred to as topography or topographic
map.
EEG electrodes are generally attached on the scalp according to international stan-
dard configurations represented by the international 10–20 system. In the 10–20 sys-
tem, electrodes are placed at 10 and 20% fractions of the geodesic distances between
a number of anatomical landmarks such as inion, nasion, and two preauricular points.
Smaller subdivisions (e.g., the 10–5 system) are also used for the placement of more
electrodes. Further information on the electrode systems and electrode naming can
be found in Oostenveld and Praamstra [24] and other sources—e.g., Wikipedia,
https://en.wikipedia.org/wiki/10–20_system_(EEG).
In general, most EEG recording devices are composed of a signal amplifier, analog
filter, and analog-to-digital converter (ADC). Use of high-quality signal amplifiers is
necessary to display and process EEG signals on the order of microvolts. Since the
recorded EEG signals are usually contaminated by unwanted environmental and/or
systemic noises, such as alternating current (AC) power noises, a variety of electronic
1 Basics of EEG: Generation, Acquisition, and Applications of EEG 7
circuits are implemented in the EEG amplifier to remove or reduce the noise. Analog
filters can also be used to remove specific noise components and increase signal-to-
noise ratio (SNR). High-pass and band-reject (notch) filters can be used optionally
to reject low-frequency physiological noise (e.g., respiration artifact) and AC power
noise, respectively. All EEG devices should include an analog low-pass filter with
a cutoff frequency less than half of the sampling rate to prevent aliasing, unwanted
distortion in the sampled EEG signal. This type of analog low-pass filter is generally
referred to as the anti-aliasing filter. This will be dealt with in a more detailed manner
in Chap. 3. ADC converts the amplified and filtered analog signals to digital EEG
signals using sampling and encoding procedures [28].
Once the digital EEG signals have been stored in storage media, a variety of forms
of information characterizing the underlying brain activities can be extracted from
the numeric data. In this book, four major computational EEG analysis methods are
introduced: EEG spectral analysis (Chap. 3), event-related potential (ERP) analy-
sis (Chap. 4), EEG source imaging (Chap. 5), and functional connectivity analysis
(Chap. 6).
In the history of EEG, the most important advancement was the use of stimulus-locked
averaging of event-related EEG. Using event-related potentials (ERP) analysis, one
can observe spatiotemporal components of stimulus-locked brain electrical activities
with reduced background noise. Examples of important ERP components include
P300 [20], N170 [4], mismatch negativity (MMN) [17], and error-related negativity
(ERN) [30], which have been widely used not only for cognitive/clinical neuroscience
studies [21] but also for BCI applications [7]. A series of methods has recently been
proposed to extract more precise spatiotemporal ERP waveforms with fewer repeated
trials, and this will be introduced in a detailed manner in Chap. 4.
tive impairment [26], and post-traumatic stress disorder [13]. In particular, functional
connectivity analysis is useful to study epilepsy because epilepsy is thought to be
one of the most representative brain network disorders [18]. Detailed descriptions
of the functional connectivity measures can be found in Chap. 6.
In the early stage of development of EEG, visual inspection of EEG waveforms was
the only way to use EEG in practical applications. Indeed, visual inspection of EEG
waveforms is still useful in studying sleep and diagnosing some neurological dis-
orders, such as epilepsy. Dissemination of digital EEGs expanded the application
fields of EEGs from limited research and diagnostic applications to more-extensive
applications, including cognitive neuroscience study, diagnosis of psychiatric dis-
eases, neuromarketing, neuroergonomics, sports science, and human brain mapping.
Recently, owing to the rapid development of digital engineering, EEGs can be applied
to real-time applications, such as BCI and neurofeedback.
The use of EEG in practical applications has steadily increased and is expected
to continue to increase. Indeed, EEG has many advantages over the other methods
to study brain functions, as follows:
• EEG is perfectly noninvasive, without any exposure to radiation or high magnetic
field
• EEG is economical
• EEG devices can be made small and portable
• EEG has high temporal resolution
• EEG devices do not generate any noise
• EEG can be recorded in an open environment
• EEG can be acquired without active response from subjects.
Traditionally, EEG data were acquired in laboratory or clinical environments,
where there are high-end EEG recording devices with a large number of channels
and well-motivated participants who have agreed to participate in experiments with
long durations. Recently, however, the advancement of wireless technology and high-
performance biosensors enabled the development of wearable EEG devices that are
easy to wear and comfortable for long-term use, expediting the development of
novel applications of EEG that do not necessarily require laboratory settings, e.g.,
monitoring the brain activity of healthy persons during daily life [5, 19, 29].
Despite the recent development of EEG technology, EEG still has some intrinsic
limitations that need to be overcome, examples of which include low spatial reso-
lution and low SNR. Therefore, development of new computational EEG analysis
methods is still necessary to enhance the reliability and usability of EEG.
10 C.-H. Im
References
1. S. Baillet, J.C. Mosher, R.M. Leahy, Electromagnetic brain mapping. IEEE Signal Process.
Mag. 18, 14–30 (2001)
2. S. Baillet, J.J. Riera, G. Marin, J.F. Mangin, J. Aubert, L. Garnero, Evaluation of inverse
methods and head models for EEG source localization using a human skull phantom. Phys.
Med. Biol. 46, 77–96 (2001)
3. E. Başar, A review of gamma oscillations in healthy subjects and in cognitive impairment. Int.
J. Psychophysiol. 90, 99–117 (2013)
4. V.C. Blau, U. Maurer, N. Tottenham, B.D. McCandliss, The face-specific N170 component is
modulated by emotional facial expression. Behav. Brain Funct. 3, 7 (2007)
5. W.D. Chang, H.S. Cha, K. Kim, C.H. Im, Detection of eye blink artifacts from single prefrontal
channel electroencephalogram. Comput. Methods Programs Biomed. 124, 19–30 (2016)
6. J.C. de Munck, B.W. van Dijk, H. Spekreijse, Mathematical dipoles are adequate to describe
realistic generators of human brain activity. IEEE Trans. Biomed. Eng. 35(11), 960–966 (1988)
7. R. Fazel-Rezai, B.Z. Allison, C. Guger, E.W. Sellers, S.C. Kleih, A. Kübler, P300 brain com-
puter interface: current challenges and emerging trends. Front Neuroeng. 5, 14 (2012)
8. D.C. Hammond, Neurofeedback treatment of restless legs syndrome and periodic leg move-
ments in sleep. J. Neurother. 16(2), 155–163 (2012)
9. J. Haueisen, C. Ramon, M. Eiselt, H. Brauer, H. Nowak, Influence of tissue resistivities on
neuromagnetic fields and electric potentials studied with a finite element model of the head.
IEEE Trans. Biomed. Eng. 44(8), 727–735 (1997)
10. M.S. Hämäläinen, R. Hari, R.J. Ilmoniemi, J. Knuutila, O.V. Lounasmaa, Magnetoencephalog-
raphy. Theory, instrumentation and applications to the noninvasive study of human brain func-
tion. Rev. Mod. Phys. 65, 413–497 (1993)
11. H.J. Hwang, K. Kwon, C.H. Im, Neurofeedback-based motor imagery training for brain-
computer interface (BCI). J. Neurosci. Meth. 179(1), 150–156 (2009)
12. H.J. Hwang, S. Kim, S. Choi, C.H. Im, EEG-based brain-computer interfaces: a thorough
literature survey. Int. J. Hum. Comput. Interact. 29(12), 814–826 (2013)
13. C. Imperatori, B. Farina, M.I. Quintiliani, A. Onofri, P. Castelli Gattinara, M. Lepore, V. Gnoni,
E. Mazzucchi, A. Contardi, G. Della Marca, Aberrant EEG functional connectivity and EEG
power spectra in resting state post-traumatic stress disorder: a sLORETA study. Biol. Psychol.
102, 10–17 (2014)
14. K.Y. Jung, Y.S. Koo, B.J. Kim, D. Ko, G.T. Lee, K.H. Kim, C.H. Im, Electrophysiologic dis-
turbances during daytime in patients with restless legs syndrome: further evidence of cognitive
dysfunction? Sleep Med. 12(4), 416–421 (2011)
15. Y.J. Jung, H.C. Kang, K.O. Choi, J.S. Lee, D.S. Kim, J.H. Cho, S.H. Kim, C.H. Im, H.D. Kim,
Localization of ictal onset zones in Lennox-Gastaut syndrome using directional connectivity
analysis of intracranial electroencephalography. Seizure 20(6), 449–457 (2011)
16. W. Klimesch, R. Fellinger, R. Freunberger, Alpha oscillations and early stages of visual encod-
ing. Front. Psychol. 2, 118 (2011)
17. D. Ko, S. Kwon, G.T. Lee, C.H. Im, K.H. Kim, K.Y. Jung, Theta oscillation related to the
auditory discrimination process in mismatch negativity: oddball versus control paradigm. J.
Clin. Neurol. 8(1), 35–42 (2012)
18. C. Lee, S.M. Kim, Y.J. Jung, C.H. Im, D.W. Kim, K.Y. Jung, Causal influence of epileptic
network during spike-and-wave discharge in juvenile myoclonic epilepsy. Epilepsy Res. 108(2),
257–266 (2014)
19. C.T. Lin, L.D. Liao, Y.H. Liu, I.J. Wang, B.S. Lin, J.Y. Chang, Novel dry polymer foam
electrodes for long-term EEG measurement. IEEE Trans. Biomed. Eng. 58, 1200–1207 (2011)
20. D.E. Linden, The p300: where in the brain is it produced and what does it tell us? Neuroscientist
11(6), 563–576 (2005)
21. S.J. Luck, An Introduction to the Event-Related Potential Technique (The MIT Press, Boston,
2005)
1 Basics of EEG: Generation, Acquisition, and Applications of EEG 11
Sung-Phil Kim
Abstract Preprocessing of the EEG signal, which is virtually a set of signal process-
ing steps preceding main EEG data analyses, is essential to obtain only brain activity
from the noisy EEG recordings. It has been shown that the design of preprocess-
ing procedures can affect subsequent EEG data analysis outcomes. Preprocessing of
EEG largely includes a number of processes, such as line noise removal, adjustment
of referencing, elimination of bad EEG channels, and artifact removal. This chapter
presents an overview of the methods available for each process and discusses prac-
tical considerations for applying these methods to the EEG signals. In particular,
considerable attention is paid to the state-of-the-art artifact removal methods since
there are still plenty of opportunities to enhance the artifact removal techniques for
EEG, in the perspectives of both signal processing and neuroscience. It is desirable
that this chapter provides the readers an overall view of EEG preprocessing pipelines
and serves as a handbook guide for the practice of EEG preprocessing.
2.1 Introduction
Preprocessing of the EEG signal is an indispensable step for the analysis of EEG in
most circumstances. Although there is still a lack of the standard pipeline of EEG
preprocessing [8, 37, 58] it generally includes any necessary digital signal processing
operations to polish up raw EEG signals with an aim to leave only brain activity
signals for subsequent analyses. Often, EEG preprocessing also involves procedures
to enhance spatiotemporal characteristics of the EEG signal related to the task used
in a study [65].
A number of studies have demonstrated the influences of EEG preprocessing on
the subsequent data analysis results [8, 33, 90, 110, 112]. For instance, the classi-
fication of different mental states from EEG or the control performance of a brain-
computer interface (BCI) could be dependent on how EEG preprocessing treated the
recorded EEG signals. In fact, it is obvious that any analytic result from the EEG
signals containing significant noise and artifacts is likely to draw misleading conclu-
sions. Recent reports also emphasize the standardization of preprocessing routines
for multi-site data collection in divergent experimental environments [8, 37].
At the center of EEG processing lies the removal of any unnecessary covert and
overt components of the EEG signals. In this chapter, we denote such unneces-
sary components as noise and artifacts. Following the previous notion [65], noise is
regarded as neurological activities irrelevant to an examined behavioral task whereas
artifacts are regarded to originate from external sources unrelated to neurological
activities, such as eye movements, respiration or electrical interference. As most
EEG preprocessing techniques pay attention to removing artefacts, we will also nar-
row our focus on the methods used to eliminate artifacts to clean up the EEG signals.
Note that the topics covered by this chapter do not include the extraction of fea-
tures from the EEG signals for particular applications, which should be discussed
separately.
This chapter begins with the description of early-stage procedures to remove
basic artifacts, sort out contaminated channels and possibly adjust references. It then
discusses a range of methods to remove artifacts from the EEG signals, followed by
brief discussion on EEG preprocessing.
A basic and brief summary of the characteristics of background EEG activity is given
as follows [104]. The frequency range of EEG is reportedly limited approximately
from 0.01 to 100 Hz. The amplitudes of EEG generated from the brain typically range
within ±100 µV. The power spectral density of EEG is known to follow the power
law [44]. Background brain rhythms are present in EEG, generally being classified
in terms of oscillatory frequency into five disjoint bands: delta (0.5–4 Hz), theta
(4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz) and gamma (30–100 Hz). More details
2 Preprocessing of EEG 17
of the implications and functions of these rhythms can be found in other resources
(e.g. see [10, 41, 63, 98]).
It is reasonable to consider the EEG signal as stochastic due to the lack of genuine
EEG measurements [93]. In addition, over a long-term period, the EEG signals should
be viewed as a non-stationary time series [57, 66]. However, EEG within a short
time window can be approximately stationary with static statistical properties. The
length of such a window containing stationary EEG signals varies with environments,
generally ranging from several seconds to minutes [51].
Most efforts to eliminate line noise from the EEG signal rely on notch filtering at
60 Hz. A notch filter is typically implemented with a certain frequency width sur-
rounding 60 Hz (e.g. a width of 10 Hz). Consequently, notch filtering, although
successfully removing line noise, could cause unintended distortions in signal com-
ponents oscillating between 50 and 70 Hz. Also, the notch filter can reportedly
generate a transient oscillation in baseline activity, leading to a potential issue in
data interpretation [18]. Follow-up low-pass filtering with a cutoff frequency lower
than 50 Hz may remedy this problem, but instead give rise to other issues such as
alteration of temporal structures of EEG [106] or spurious interactions between EEG
channels [40].
One suggestion to overcome this problem is estimating line noise embedded in
the recorded EEG signals as precise as possible and subtracting it from the data [8,
80]. This method employs multi-taper decomposition to find line noise components
in the signal. A short-time window slides over the course of the signal in which the
transformation of EEG time series based on multi-tapers is carried out [5]. This trans-
formation can effectively estimate spectral energy within each frequency band. Then,
a regression model is applied to estimate the amplitude and phase of sinusoidal line
noise (e.g. sinusoids at 60 Hz) in the transformed frequency domain. The Thompson
F-test evaluates a significance of the magnitude of the estimated line noise. A time
series of sinusoidal line noise is reconstructed if the magnitude is significant. This
process is repeated over the sliding windows. The reconstructed line noise signal
is subtracted from the original EEG signal. The entire process is repeated until the
magnitude at the frequency of line noise becomes non-significant (Fig. 2.1). In this
way, line noise components can be removed without damaging background spectral
components [83].
2.2.3 Referencing
We often subtract a reference (with the same time resolution as the recorded EEG
signals) from the original EEG signal at each channel. The reference signal should
18 S.-P. Kim
remain unchanged relative to the EEG signals during the recording such that dif-
ferences of the EEG signals from reference can effectively represent brain activity
related to a study. Typical choices of reference include a signal recorded at a mastoid
channel, an EEG signal at a particular channel, the average of two mastoid signals
or the average of the entire EEG channels. In any case, it is strongly recommended
that a researcher should inspect a chosen reference signal carefully to ensure that its
amplitude level is on par with those of other EEG signals and it has no correlation
with task-induced brain activity.
Referencing to a mastoid channel has a potential problem because it generates a
single point of failure. If the contact to a mastoid becomes poor at any point during
the recording, referencing to the mastoid can increase signal variance tremendously,
resulting in irreversible contamination of EEG data. The same problem exists for ref-
erencing to a particular EEG channel. Using the common average reference (CAR)
may reduce the effect of single-point failure [9], but still suffer from an outlier chan-
nel. One simple solution to this problem is detecting and removing bad channels
before using CAR [8]. There are other systematic re-referencing methods developed
to address the issues of reference, based on physical considerations and electrody-
namics [38, 113, 114] or on statistical approaches [48, 69, 73].
low-frequency components and detect a bad channel showing a ratio higher than a
threshold.
Once being detected, bad channels are replaced with virtual healthy channels
created by the interpolation from neighboring channels, in order to reconstruct the
global brain responses [8, 31]. There exist a number of interpolation schemes useful
for channel reconstruction, including spherical splines [87], higher-order polynomi-
als [4], nearest-neighbor averaging [15] and radial basis function [53]. Using spher-
ical splines allows accurate estimation of scalp potentials if the electrode mapping
is sufficiently dense [38, 97]. Interpolation using a statistical method such as radial
basis functions has advantages of cost-effectiveness with less computational loads.
In this section, we briefly review the potential sources of artifacts mixed in the EEG
signal and the techniques to remove or reduce artifacts. We primarily deal with artifact
removal techniques, forgoing other steps of artifact management such as artifact
detection. However, it does not mean that other methods including artifact detection
or artifact avoidance are less crucial than artifact removal. In fact, artifact removal
is often accompanied by artifact detection for efficient processing of artifacts. There
have been a number of methods for artifact detection that the interested readers can
refer to [3, 14, 32, 52, 81, 84].
The sources of EEG artifacts can be categorized into two classes: internal and external
sources. The internal sources originate from the physiological systems of self and
include electromagnetic activities of heart, eyes, muscle and so on. The external
sources include all other possible signals from environments that can contaminate
EEG such as wireless telecommunication signals, electrode attachment, recording
equipment and cable movements [93]. Recently. the handling of external artifacts
has become more important as EEG applications tend to move out of laboratories
toward in-home healthcare systems [100]. Yet, the external sources, owing to their
origins, can be inhibited once being identified. On the other hand, the internal artifacts
physiologically permeate EEG, making it difficult to prevent them from occurring
in advance. Therefore, most artifact removal methods have been focused on dealing
with the internal artifacts and here we also pay our attention to the most pronounced
internal artifacts that have been handled by EEG artifact removal methods.
Ocular artifacts include electric activities generated by eye movements or eye
blinking [22, 23]. Interference by ocular artifacts is strong enough to be visible
in EEG waveforms. EEG channels proximal to eyes are more vulnerable to ocular
artifacts. Ocular artifacts can be detected by electrooculogram (EOG) measurements.
Another random document with
no related content on Scribd:
SWEET CARROTS. (ENTREMETS.)
(A Dutch Receipt.)
Prepare some finely flavoured carrots as above, and dry them
over a gentle fire like mashed turnips; then for a dish of moderate
size mix well with them from two to three ounces of good butter, cut
into small bits, keeping them well stirred. Add a seasoning of salt
and cayenne, and serve them very hot, garnished or not at pleasure
with small sippets (croutons) of fried bread.
CARROTS AU BEURRE, OR BUTTERED CARROTS.
(French.)
Either boil sufficient carrots for a dish quite tender, and then cut
them into slices a quarter of an inch thick, or first slice, and then boil
them: the latter method is the most expeditious, but the other best
preserves the flavour of the vegetable. Drain them well, and while
this is being done just dissolve from two to three ounces of butter in
a saucepan, and strew in some minced parsley, some salt, and white
pepper or cayenne; then add the carrots, and toss them very gently
until they are equally covered with the sauce, which should not be
allowed to boil: the parsley may be omitted at pleasure. Cold carrots
may be re-warmed in this way.
CARROTS IN THEIR OWN JUICE.
Boil them until they are about half done, lift them out, and let them
cool; slice them rather thickly, sprinkle them with fine salt and white
pepper, and fry them a pale brown in good butter. Serve them with
roast meat, or dish them under it.
JERUSALEM ARTICHOKES.
Wash the artichokes, pare them quickly, and throw them as they
are done into a saucepan of cold water, or of equal parts of milk and
water; and when they are about half boiled add a little salt to them.
Take them up the instant they are perfectly tender: this will be in from
fifteen to twenty-five minutes, so much do they vary in size and as to
the time necessary to dress them. If allowed to remain in the water
after they are done, they become black and flavourless. Melted
butter should always be sent to table with them.
15 to 25 minutes.
TO FRY JERUSALEM ARTICHOKES. (ENTREMETS.)
Boil them from eight to twelve minutes; lift them out, drain them on
a sieve, and let them cool; dip them into beaten eggs, and cover
them with fine bread-crumbs. Fry them a light brown, drain, pile them
in a hot dish, and serve them quickly.
JERUSALEM ARTICHOKES, À LA REINE.
Boil them tender, press the water well from them, and then
proceed exactly as for mashed turnips, taking care to dry the
artichokes well, both before and after the milk or cream is added to
them; they will be excellent if good white sauce be substituted for
either of these.
HARICOTS BLANCS.
Wash the roots delicately clean, but neither scrape nor cut them,
for should even the small fibres be taken off before they are cooked,
their beautiful colour would be much injured. Throw them into boiling
water, and, according to their size, which varies greatly, as they are
sometimes of enormous growth, boil them from one hour and a half
to two and a half, or longer if requisite. Pare and serve them whole,
or cut into thick slices and neatly dished in a close circle: send
melted butter to table with them. Cold red beet root is often
intermingled with other vegetables for winter salads; and it makes a
pickle of remarkably brilliant hue. A common mode of serving it at
the present day is in the last course of a dinner with the cheese: it is
merely pared and sliced after having been baked or boiled tender.
1-1/2 to 2-1/2 hours, or longer.
TO BAKE BEET ROOT.
Bake or boil it tolerably tender, and let it remain until it is cold, then
pare and cut it into slices; heat and stew it for a short time in some
good pale veal gravy (or in strong veal broth for ordinary occasions),
thicken this with a teaspoonful of arrow-root, and half a cupful or
more of good cream, and stir in, as it is taken from the fire, from a
tea to a tablespoonful of chili vinegar. The beet root may be served
likewise in thick white sauce, to which, just before it is dished, the
mild eschalots of page 128 may be added.
TO STEW RED CABBAGE.
(Flemish Receipt.)
Strip the outer leaves from a fine and fresh red cabbage; wash it
well, and cut it into the thinnest possible slices, beginning at the top;
put it into a thick saucepan in which two or three ounces of good
butter have been just dissolved; add some pepper and salt, and stew
it very slowly indeed for three or four hours in its own juice, keeping it
often stirred, and well pressed down. When it is perfectly tender add
a tablespoonful of vinegar; mix the whole up thoroughly, heap the
cabbage in a hot dish, and serve broiled sausages round it; or omit
these last, and substitute lemon-juice, cayenne pepper, and a half-
cupful of good gravy.
The stalk of the cabbage should be split in quarters and taken
entirely out in the first instance.
3 to 4 hours.
BRUSSELS SPROUTS.
Boil the salsify tender, as directed above, drain, and then press it
lightly in a soft cloth. Make some French batter (see Chapter V.),
throw the bits of salsify into it, take them out separately, and fry them
a light brown, drain them well from the fat, sprinkle a little fine salt
over them after they are dished, and serve them quickly. At English
tables, salsify occasionally makes its appearance fried with egg and
bread-crumbs instead of batter. Scorgonera is dressed in precisely
the same manner as the salsify.
BOILED CELERY.
Cut five or six fine roots of celery to the length of the inside of the
dish in which they are to be served; free them from all the coarser
leaves, and from the green tops, trim the root ends neatly, and wash
the vegetable in several waters until it is as clean as possible; then,
either boil it tender with a little salt, and a bit of fresh butter the size
of a walnut, in just sufficient water to cover it quite, drain it well,
arrange it on a very hot dish, and pour a thick béchamel, or white
sauce over it; or stew it in broth or common stock, and serve it with
very rich, thickened, Espagnole or brown gravy. It has a higher
flavour when partially stewed in the sauce, after being drained
thoroughly from the broth. Unless very large and old, it will be done
in from twenty-five to thirty minutes, but if not quite tender, longer
time must be allowed for it. A cheap and expeditious method of
preparing this dish is to slice the celery, to simmer it until soft in as
much good broth as will only just cover it, and to add a thickening of
flour and butter, or arrow-root, with some salt, pepper, and a small
cupful of cream.
25 to 30 minutes, or more.