![American Psychiatry Association [1]. Schizophrenia treatment involves long-term
medication and is a great burden on healthcare systems and families [2]. The early
prediction of schizophrenia involves a large number of aspects [3]. The reliability
and comparability of studies have increased dramatically due to the introduction of
standardized tools for evaluating symptoms and diagnosis. However, the problems
of selecting proper methods and evaluation tools, and repeatability, etc, remain.
Electroencephalogram (EEG) signals have gained much attention in the diagnosis of
schizophrenia due to their non-invasive nature and ease of use [4–7]. EEG signals are
electrical measures of the brain activity of billions of neurons connected together to
form a network. An EEG signal is acquired from the scalp and has played a key role
in clinical diagnosis and the dynamics of brain research. EEG signals provide
increased coherence that reflects the presence of anomalous cortical organization in
schizophrenics rather than transient states or medication effects related to severe
clinical disturbance [8]. The temporal, occipital, frontal and parietal portions of the
scalp play significant roles in analysing the changes during schizophrenia [9].
To date, researchers have proposed various methods for the detection and
diagnosis of schizophrenia using EEG signals. The detection of schizophrenia by
clustering the EEG signals with the help of the k-means method has been proposed
[10]. The psychopharmacological and physiological changes occurring in the EEGs
of schizophrenic and healthy patients have been monitored [11]. The biological and
clinical association of the alpha and gamma frequency bands and power has been
studied to separate schizophrenic and normal patients [12–14]. The identification of
schizophrenic and normal patients has been carried out using spectral analysis
[15, 16]. A rhythm based risk rate evaluation of healthy and schizophrenic patients is
used in [17]. The alpha, delta, beta and gamma rhythms of occipital, central and
frontal sites have been classified using the support vector machine (SVM) [18]. The
separation of rhythm based features using filtering methods, multilayer back-
propagation and self-organizing maps has been used [19]. Rhythm based features
using a band pass filtering method with SVM, Sammon map and deep neural
network classification techniques have been utilized for the identification of
schizophrenia [20–23] as has the evaluation and classification of frequency based
features using linear discriminant analysis [24]. The detection of schizophrenic
patients using matched filtering and the fast Fourier transform (FFT) is proposed in
[25]. The separation of rhythms using the Grey Walter passive filter has also been
used to identify schizophrenic patients.
Positive and negative schizophrenia have been separated using the FFT of EEG
signals of the frontal, temporal, parietal and occipital regions [26]. The brain
activities of schizophrenic patients have been detected by evaluation of the spectral
energy using the FFT [27]. The utility of the FFT along with principal component
analysis, the Wilcoxon method and Welch’s averaged periodogram method has been
demonstrated in identifying the changes in the delta, beta, alpha and gamma bands
of schizophrenic patients [28–33]. The different spikes, namely the focal, paroxysmal
and independent spikes, occurring in the EEGs of schizophrenic patients have been
analysed [34]. Post-imperative negative variation and contingent negative variation
analysis have been used for identifying schizophrenic patients [35]. The steady-state
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-2](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-28-320.jpg)
![visual evoked potential (SSVEP) and Fisher score have been used to evaluate
different features classified using quadratic discriminant analysis (QDA), linear
discriminant analysis (LDA), SVM with k-nearest neighbors (k-NN), second order
polynomial kernels and logistic regression analysis (LRA) [36]. Features based on
band power, autoregressive coefficients, Lampel–Ziv complexity (LZC), fractal
dimensions (FDs) and entropies have been classified using SVM, adaptive boosting
and LDA to detect schizophrenia [37–40]. The Welch periodogram technique for
spectral estimation has been used to detect schizophrenia using the Kolmogorov–
Smirnov test [41]. A genetic algorithm with a Butterworth filter and SVM has also
been used to identify schizophrenia [42].
Ensemble synchronization measurement and Hilbert phase synchronization
based methods have been used for classification using a logistic regression classifier
[43]. The Hilbert–Huang transform, PCA, ICA and local discriminant bases have
been used to extract the features of schizophrenic patients [44]. The utility of LZC
for the identification of patients with schizophrenia is described in [45]. Power
analysis of the alpha and delta bands has been carried out to distinguish control and
affected patients [46]. The Higuchi, Katz and Petrosian methods have been used to
extract classification features using LDA [47]. The weighted nearest neighbor, band
power, FDs and autoregressive methods have been used to classify schizophrenic
patients and control patients [48]. An auto-correlation and autoregressive coefficient
have been classified using the independent t-test and neural networks [49–54]. LZC
and correlation have been explored for measuring the alpha band activity of
schizophrenic patients [55]. Multi-set canonical correlation analysis (MCCA) and
SVM with recursive feature elimination (SVM-RFE) have been used for discrim-
inating schizophrenia [56]. Entropy measurements and mean coherence with SVM
have been used to discriminate schizophrenia [57]. Hurst exponent and FDs have
been used to differentiate schizophrenic and control patients [22]. Kolmogorov
complexity (KC), entropy and LZC methods have been used to find a useful
discriminative tool for diagnostic purposes [58]. Feature vectors based on LZC and
ANN have also been used to identify schizophrenic patients [59].
Autoregressive (AR), band power and FD coefficient based features extracted
after preprocessing have been classified using LDA, multi-LDA (MLDA) and
adaptive boosting (Adaboost) [60]. FDs and Pearson’s correlation coefficient have
been used to apply the brief psychiatric rating scale (BPRS) for the detection of
schizophrenia [61]. Power spectral density based features have been classified using a
combination of factor analysis based on maximum likelihood theory [62, 63].
Spectral features extracted from combinatorial analysis have been classified using
the Kora-N algorithm [64]. The weighted minimum distance to mean and
Riemannian geometric mean have been used for the classification of schizophrenia
[65]. The equivalent current dipole power and asymmetry coefficient have been used
for the analysis of the positive symptoms of schizophrenia [66]. Factor analysis and
Kaiser’s criteria have been used to identify patients suffering from schizophrenia
[67]. Eigenvector power spectrum estimation and SVM have been used for the
identification of schizophrenia [68]. Higuchi’s method of computation of FDs has
also been used to detect schizophrenia [69]. Energy and power based features have
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-3](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-29-320.jpg)
![been classified using a high order pattern discovery algorithm [70]. The ε-complexity
of continuous vector functions with RFs and an SVM have been used for binary
classification of healthy and schizophrenic patients [71]. A fuzzy accuracy based
classifier system has been used to generate fuzzy rules for discriminating healthy and
schizophrenic subjects [72]. Autoregression based directed connectivity (DC) and
graph-theoretical complex network (CN) based features have been classified using
deep neural networks [73]. Inherent spatial pattern of network (SPN) features have
been classified using LDA and an SVM to separate healthy patients from
schizophrenic patients [74]. The analysis of entropy has been used for the evocation
of emotions from visual cues in schizophrenia [75]. Mutual information (MI) has
used to construct functional brain networks for analysis using graph theory [76].
Statistic significance probability maps based on the BPRS and a scale for the
assessment of negative symptoms have been used for morphological findings in
schizophrenia [77]. The spectral power of 192-channel resting EEG has been
analysed using the Pearson correlation coefficient [78]. Spectral, complexity and
variability measures evaluated from EEG signals have been classified using k-NN
[79]. The long-term replicability of EEG spectra and auditory evoked potentials
have been analysed to identify patients suffering from schizophrenia [80]. Sample
covariance matrix and linear eigenvalue statistics have been used to classify
schizophrenic patients using decision tree, random forest, SVM and naïve Bayes
classifiers [81]. The Lyapunov exponent and Kolmogorov entropy have been
evaluated to identify the classification accuracy of schizophrenic and controlled
patients [82]. Average reference potential maps corresponding to global field power
peaks in rhythms have been used to classify patients with schizophrenia [83].
Higuchi’s FD, entropy and Kolmogorov complexity based features have been
classified using SVM [84].
Independent component analysis (ICA) and time–frequency representation using
the Stockwell transform have been used to find the most significant rhythms in
schizophrenic patients [85]. ICA, spectral analysis and analysis of variance
(ANOVA) have been carried out on the frequency bands to identify control patients
from schizophrenic patients [86]. Filtering, ICA and Fisher analysis have been used
to classify patients using connectivity maps [87]. ICA for the spectral analysis of 200
bands and RFs have been used for accurately detecting schizophrenia based on one-
minute EEG recordings [88]. Fourier statistical analysis, evaluative power spectra,
averaged power spectra and spectral variance have been used to identify the traits of
schizophrenia among patients [89]. Time–frequency distributions (TFDs), FFTs,
eigenvector methods, the wavelet transform (WT) and AR method have been used
for the extraction of features, with advantages and disadvantages [90]. Filtering,
FFT, STFT and entropy based features have been used for the classification of
schizophrenia using an SVM and multilayer perceptron (MLP) [91]. Short-time
Fourier transforms with a sliding window have been used to distinguish schizo-
phrenic patients [92]. Feature extraction based on wavelet filtering with a genetic
algorithm and SVM has been used to identify control patients [93]. Classification
based on PCA, wavelet transform and k-NNs is proposed in [94]. Time–frequency
analysis has been used, with a Morlet wavelet having a Gaussian shape in time and
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-4](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-30-320.jpg)
![frequency, for the detection of schizophrenia [95]. Analysis of schizophrenic patients
has been carried out using wavelet decomposition and Welch power spectral density
(PSD) methods [96]. Analysis of alpha band frequencies has been carried out to
detect the activity in schizophrenic patients [97]. Discriminant analysis (DA) has
also been employed for classifying schizophrenic patients [98]. Features evaluated
using Kolmogorov entropy, permutation entropy, correlation dimensions and
spectral entropy have been selected using the Fisher criterion and classified using
k-NN, SVM and back-propagation neural networks [99]. The phase lock value and
phase coherence value of the intrinsic mode functions of empirical mode decom-
position have been used to differentiate schizophrenia [100]. Multi-domain convolu-
tional neural networks have been used for the classification of EEG based brain
connectivity networks in schizophrenia [101].
Various methods for the detection of schizophrenia have been proposed in the
literature. The FFT suffers from time–frequency localization. Other rigid methods
such as STFT, wavelet transform and filtering use a basis which is independent of
the processed signal. Moreover, the majority of these methods involve the direct
evaluation of features from the raw EEG signals. The empirical wavelet transform
(EWT) is capable of building an adaptive wavelet to extract the AM–FM
components of a signal. The adaptive selection of the wavelet can capture useful
hidden information from non-stationary EEG signals. In this chapter, a wavelet
based decomposition method is employed to decompose the signal into AM–FM
components. Dominant time domain features evaluated from the AM–FM compo-
nents are selected using the Kruskal–Wallis test. The selected features are given as
the input for the classifiers to distinguish patients with schizophrenia from control
patients. The performance of the system is tested by evaluating four performance
parameters and the receiver operating characteristics curve. The remainder of the
chapter is organized as follows: section 1.2 presents the methodology, the results and
discussion are provided in section 1.3, and section 1.4 concludes the chapter.
1.2 Methodology
This section includes descriptions of the dataset, empirical wavelet transform,
features and classification techniques. The EEG signals are decomposed into
AM–FM components using the EWT. Multiple time domain features are extracted
from the obtained AM–FM components. Highly discriminant features are selected
using the Kruskal–Wallis test and are classified using different classification
techniques. The flowchart of the proposed methodology is shown in figure 1.1.
1.2.1 Dataset
The dataset used in this chapter contains EEG recordings of 14 female and 67 male
patients. The average age and years of education are 39 years and 14.5 years,
respectively. The details of the dataset are available online [102]. Three press button
tasks were performed by the subjects, namely (1) pressing a button to immediately
generate a tone, (2) passively listening to the same tone and (3) pressing a button
without generating a tone to study the corollary discharge in people with
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-5](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-31-320.jpg)
![separation of these modes is similar to Fourier spectrum segmentation and filtering.
The EWT is defined in the same manner as the traditional wavelet transform. The
EWT has detailed and approximation coefficients [103]. The detailed coefficients
( α
W n t
( , )
f ) are defined as the inner products with the empirical wavelet given as
∫
ϕ τ ϕ τ τ
ω ϕ ω
= ⟨ ⟩ = −
= ˆ ˆ
α
( )
W n t f f t
f
( , ) , ( ) ( )d
( ) ( ) .
(1.1)
f n n
n
v
The approximation coefficients ( α
W t
(0, )
f ), defined as the inner product with a
scaling function, can be written as
∫
φ τ φ τ τ
ω φ ω
= ⟨ ⟩ = −
= ˆ ˆ
α
ν
W t f f t
f
(0, ) , ( ) ( )d
( ( ) ( )) ,
(1.2)
f 1 1
1
where f is the input signal in the time domain, and φ and ϕ are the wavelet and
scaling functions, respectively. The reconstruction of the signal can be denoted as
∑
∑
φ ω ϕ
ω φ ω ω ϕ ω
= × + ×
= ˆ × ˆ + ˆ × ˆ
=
=
α α
α α
f t W t t W t t
W W n
( ) (0, ) ( ) ( , ) ( )
(0, ) ( ) ( , ) ( ) .
(1.3)
n
N
n
N
1
1
f f n
f f n
1
1
⎛
⎝
⎜
⎜
⎞
⎠
⎟
⎟
The AM–FM components obtained from the signal are shown in figure 1.3.
1.2.3 Feature extraction
Features are the statistical measures evaluated from the AM–FM components of
signals. These statistical measures play an important role in the dimensionality
Figure 1.3. Modes obtained from EWT.
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-7](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-33-320.jpg)
![1.2.4 Classification techniques
The purpose of classifiers is to classify input data into two or more classes. The
feature matrix is given as an input for classifiers. In this chapter, five different
classification techniques are employed to classify binary class data. The k-NN, DA,
ensemble method, SVM and decision tree based classifiers are used. To classify the
input signals different kernels are employed. In the case of k-NN, six kernels are
used, namely the fine, medium, coarse, cosine, cubic and weighted kernels. Linear
and quadratic kernels are employed with discriminant analysis. Four kernels,
namely the bagged tree, boosted tree, subspace k-NN (SS-k-NN) and subspace
discriminant (SS-D) kernels are used with an ensemble based classifier. Linear,
medium Gaussian and coarse Gaussian kernels are used with the SVM. For the
decision tree based classifiers, simple tree, complex tree and medium tree are used.
The details of the classification methods can be found in [104–108]. The process of k-
NN is denoted as
∑
= − + − + … + −
= −
=
−
−
=
d y y y y y y y y
y y
V
V A
A A
( , ) ( ) ( ) ( )
( )
min
max min
,
(1.9)
i t i t i t ip tp
i
n
i i
1 1
2
2 2
2 2
1 1 2
2
1
where d is the distance, yi is an input with p features, n is the total inputs and p is the
total number of features. V1
is the max–min normalization matrix. In this chapter
the total number of neighbors is selected as 5.
The mathematical modeling of the SVM is formed by minimizing the objective
function K(w), by taking the constraint + ⩾ = …
z w y b i N
( ) 1( 1, 2, , )
i i
T
:
= ∣∣ ∣∣
K w w
( ) min
1
2
. (1.10)
2
⎛
⎝
⎜
⎞
⎠
⎟
By augmenting the objective function, the Lagrangian function for the SVM thus
formed is denoted by
∑
λ λ
Ψ = − + −
=
w b w w z w y b
( , , )
1
2
[ ( ) 1], (1.11)
i
N
1
t
i i i
T
where, K(w) is the kernel, w is the weight matrix, b is the bias and y is the input.
The ensemble method for classification is mathematically represented by
∑
ˆ • = •
=
G c G
( ) ( ), (1.12)
i
N
1
i i
ens
where ˆ •
G ( )
ens is the ensemble based function estimator, •
G ( )
i is the reweighted
original data and ci is the averaging weights.
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-9](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-35-320.jpg)
![signals available for schizophrenia patients and 3608 control signals. Every signal of
each class has a total of 3072 samples. N-100 channels play an important role [102]
in the detection of schizophrenia. Based on this, the first ten N-100 channels are
considered for the evaluation. To maintain uniformity, a common experimental
platform is used for both classes. Each signal is given as an input to the EWT. The
boundary conditions are kept the same for both classes. The boundaries are chosen
as {4 8 12 30}. Different numbers of AM–FM components are obtained from the
signals using EWT. The minimum number of AM–FM components is eight and the
maximum is 28. However, to maintain the synchronism between all signals for
further computation, the number of AM–FM components is considered to be eight.
Various statistical parameters are evaluated from the AM–FM components. Highly
discriminable features are selected based on the results of the Kruskal–Wallis (KW)
test. The KW test is a non-parameterized analysis of variance. It is used to find the
discrimination ability of features by evaluating the probability of χ. A probability
value ⩽0.05 is considered to be significant for classification. A total of five features
are selected based on the KW test. These are kurtosis, variance, root mean square,
minima and maxima, respectively. The probabilistic values of all the features are
shown in tables 1.1–1.5, respectively. It is evident from these tables that most of the
AM–FM components and channels are highly discriminable.
Inspired by the obtained results presented in tables 1.1–1.5, the selected features
are given as the input for different classifiers. All the channels of every feature of
each AM–FM component are combined. For every AM–FM component, the
feature matrices obtained for schizophrenia and normal patients are 4108 × 50
and 3608 × 50, respectively. In this methodology, the ten-fold cross-validation
method is employed for classification. Here, the input data are partitioned randomly
into ten disjoint sets. Nine sets are used for training the input data and the remaining
set is utilized for testing. The patients with schizophrenia are separated from the
normal patients using five types of classification techniques.
Table 1.6 shows the classification accuracy obtained by the k-NN classifier. Six
kernels are used for the classification. The classification accuracies obtained with the
fine, medium, coarse, cosine, cubic and weighted kernels are, respectively, 81%,
84.1%, 82.1%, 84%, 83.3% and 84.3% for M-1, 72.1%, 75.6%, 71.5%, 75.2%, 72.4%
and 75.9% for M-2, 66%, 70.2%, 68.2%, 69.7%, 67.5% and 71.1% for M-3, 66%,
69.5%, 67.6%, 68.3%, 66.2% and 70% for M-4, 65%, 69.6%, 70%, 68.6%, 65.7% and
70.7% for M-5, 61.3%, 65.4%, 67.9%, 65%, 62.8% and 66.4% for M-6, 59.3%,
61.8%, 64.8%, 61.5%, 61.2% and 63.8% for M-7, and 60.4%, 63.4%, 65.6%, 63.9%,
62.4% and 64.6% for M-8. The maximum accuracies obtained with the fine,
medium, coarse, cosine, cubic and weighted kernel are, respectively, 81%, 84.1%,
82.1%, 84%, 83.3% and 84.3% for M-1.
Table 1.7 shows the accuracy of four classifiers, namely the discriminant analysis,
SVM, ensemble and decision tree classifiers. The classification accuracies with the
linear kernel are 74.9%, 59.3%, 56.9%, 56.4%, 55.8%, 55.7%, 56.3% and 57.1% for,
respectively, M-1, M-2, M-3, M-4, M-5, M-6, M-7 and M-8, and with QDA the
accuracies are 59%, 60.4%, 58.6%, 58.2%, 58.2%, 58.3%, 58.6% and 58.8%,
respectively. The maximum accuracies for LDA and QDA are 74.9% and 60.4%
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-11](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-37-320.jpg)
![1.4 Conclusion
Patients with schizophrenia cannot be easily identified through visual inspection.
Doctors recommend a number of neurological tests to identify the symptoms of
schizophrenia. However, these tests are not always effective. Electroencephalogram
signals provide vital information about the neurological changes that happen in the
schizophrenic state. In this chapter, a novel method based on the empirical wavelet
transform is proposed for the identification of schizophrenia. The majority of the
information resides in the first two AM–FM components, as these provide the
highest correct classifications of schizophrenic patients and normal patients. The
classification abilities of different classification techniques are tested. It is found that
SVM is the best method, followed by the ensemble, k-nearest neighbors, decision
tree and, finally, discriminant analysis classifier. The medium kernel of the SVM
provides the best performance parameters with an accuracy of 88.7%, a sensitivity of
91.13%, a specificity of 89.29% and a precision of 83.78%.
References
[1] Samuel B G 1995 Diagnostic and statistical manual of mental disorders 4th edn (DSM-IV)
Am. J. Psychiatry 152 1228
[2] McGlashan T H 1998 Early detection and intervention of schizophrenia: rationale and
research Br. J. Psychiatry 172 3–6
[3] Lett T A, Voineskos A N, Kennedy J L, Levine B and Daskalakis Z J 2014 Treating
working memory deficits in schizophrenia: a review of the neurobiology Biol. Psychiatry 75
361–70
[4] Hee N S, Jin S H, Kim S Y and Ham B J 2002 EEG in schizophrenic patients: mutual
information analysis Clin. Neurophysiol. 113 1954–60
[5] Knox H F and Macfie C C 1941 Electroencephalography in schizophrenia Am. J.
Psychiatry 98 374–81
[6] Koukkou M, Lehmann D, Wackermann J, Dvorak I and Henggeler B 1993 Dimensional
complexity of EEG brain mechanisms in untreated schizophrenia Biol. Psychiatry 33 397–
407
[7] Karson C N, Coppola R, Daniel D G and Weinberger D R 1988 Computerized EEG in
schizophrenia Schizophr. Bull. 14 193–7
[8] Merrin E L, Floyd T C and Fein G 1989 EEG coherence in unmedicated schizophrenic
patients Biol. Psychiatry 25 60–6
[9] Davis P A 1940 Evaluation of the electroencephalograms of schizophrenic patients Am. J.
Psychiatry 96 851–60
[10] Lehmann D, Faber P L, Galderisi S, Herrmann W M and Koenig T 2005 EEG microstate
duration and syntax in acute, medication-naïve, first-episode schizophrenia: a multi-center
study Psychiatry Res.: Neuroimaging 138 141–56
[11] Kornetsky C and Mirsky A F 1965 On certain psychopharmacological and physiological
differences between schizophrenic and normal persons Psychopharmacologia 8 309–18
[12] Sponheim S R, Clementz B A, Iacono W G and Beiser M 2000 Clinical and biological
concomitants of resting state EEG power abnormalities in schizophrenia Biol. Psychiatry 48
1088–97
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-20](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-46-320.jpg)
![[13] Roach B J and Mathalon D H 2008 Event-related EEG time-frequency analysis: an
overview of measures and an analysis of early gamma band phase locking in schizophrenia
Schizophr. Bull. 34 907–26
[14] Moran L and Hong L 2011 High vs low frequency neural oscillations in schizophrenia
Schizophr. Bull. 37 659–63
[15] Giannitrapani D and Kayton L 1974 Schizophrenia and EEG spectral analysis
Electroencephalogr. Clin. Neurophysiol. 36 377–86
[16] Boutros N N, Arfken C, Galderisi S, Warrick J, Pratt G and Iacono W 2008 The status of
spectral EEG abnormality as a diagnostic test for schizophrenia Schizophr. Res. 99 225–37
[17] Clementez B A, Sponheim S R, Iacono W G and Beiser M 1994 Resting EEG in first‐
episode schizophrenia patients, bipolar psychosis patients, and their first‐degree relatives
Psychophysiology 31 486–94
[18] Johannesen J, Bi J, Jiang R, Kenney J and Chen C M 2016 Machine learning identification
of EEG features predicting working memory performance in schizophrenia and healthy
adults Neuropsychiatr. Electrophysiol. 2 3
[19] Li Y and Fan F 2005 Classification of schizophrenia and depression by EEG with ANNs
2005 IEEE Engineering in Medicine and Biology 27th Annual Conf. pp 2679–82
[20] Shim M, Hwang H, Kim D, Lee S and Im C 2016 Machine-learning-based diagnosis of
schizophrenia using combined sensor-level and source-level EEG features Schizophr. Res.
176 314–9
[21] Winterer G, Ziller M, Dor H, Frick K, Mulert C, Wuebben Y and Herrmann W M 2000
Frontal dysfunction in schizophrenia—a new electrophysiological classifier for research and
clinical applications Eur. Arch. Psychiatry Clin. Neurosci. 250 207–14
[22] Chu L, Qiu R, Liu H, Ling Z and Shi X 2017 Individual recognition in schizophrenia using
deep learning methods with random forest and voting classifiers: insights from resting state
EEG streams, arXiv:1707.03467
[23] Bose T, Sivakumar S and Kesavamurthy B 2016 Identification of schizophrenia using EEG
alpha band power during hyperventilation and post-hyperventilation J. Med. Biol. Eng. 36
901–11
[24] Ince N F, Pellizzer G, Tewfik A H, Nelson K, Leuthold A, McClannahan K and Stephane
M 2009 Classification of schizophrenia with spectro-temporo-spatial MEG patterns in
working memory Clin. Neurophysiol. 120 1123–34
[25] Flor H P, Lind J C and Koles Z J 2004 A source-imaging (low-resolution electromagnetic
tomography) study of the EEGs from unmedicated males with depression Psychiatry Res.:
Neuroimaging 130 191–207
[26] Begić D, Hotujac L and Jokić‐Begić N 2000 Quantitative EEG in ‘positive’ and ‘negative’
schizophrenia Acta Psychiatr. Scand. 101 307–11
[27] Morstyn R, Duffy F H and McCarley R W 1983 Altered topography of EEG spectral
content in schizophrenia Electroencephalogr. Clin. Neurophysiol. 56 263–71
[28] Michelogiannis S, Paritsis N and Trikas P 1991 EEG coherence during hemispheric
activation in schizophrenics Eur. Arch. Psychiatry Clin. Neurosci. 241 31–4
[29] Dierks T K, Maurer R and Ihl A 1989 Schmidtke Topographic Brain Mapping of EEG and
Evoked Potentials (Berlin: Springer) pp 507–17
[30] Michel C M, Koukkou M and Lehmann D 1993 EEG reactivity in high and low
symptomatic schizophrenics, using source modelling in the frequency domain Brain
Topogr. 5 389–94
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-21](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-47-320.jpg)
![[31] Umesh S, Nizamie S H, Goyal N, Tikka S and Bose S 2016 Social anhedonia and gamma
band abnormalities as a composite/multivariate endophenotype for schizophrenia: a dense
array EEG study Early Interv. Psychiatry 12 362–71
[32] Alfimova M and Uvarova L 2003 Cognitive peculiarities in relatives of schizophrenic and
schizoaffective patients: heritability and resting EEG-correlates Int. J. Psychophysiol. 49
201–16
[33] Lacroix D, Chaput Y, Rodriguez J P, Filion M, Morrison D, St. Denis P and Albert J-M
1995 Quantified EEG changes associated with a positive clinical response to clozapine in
schizophrenia Progr. Neuro-Psychopharmacol. Biol. Psychiatry 19 861–76
[34] White P T, Demyer W and Demyer M K 1964 EEG abnormalities in early childhood
schizophrenia: a double-blind study of psychiatrically disturbed and normal children during
promazine sedation Am. J. Psychiatry 120 950–8
[35] Verleger R, Wascher E, Arolt V, Daase C, Strohm A and Kömpf D 1999 Slow EEG
potentials (contingent negative variation and post-imperative negative variation) in
schizophrenia: their association to the present state and to Parkinsonian medication effects
Clin. Neurophysiol. 110 1175–92
[36] Alimardani F, Cho J H, Boostani R and Hwang H J 2018 Classification of bipolar disorder
and schizophrenia using steady-state visual evoked potential based features IEEE Access 6
40379–88
[37] Sabeti M, Katebi S and Boostani R 2009 Entropy and complexity measures for EEG signal
classification of schizophrenic and control participants Artif. Intell. Med. 47 263–74
[38] Sabeti M, Boostani R, Katebi S D and Price G W 2007 Selection of relevant features for EEG
signals classification of schizophrenic patients Biomed. Signal Process. Control 2 122–34
[39] Sabeti M, Katebi S D, Boostani R and Price G W 2011 A new approach for EEG signal
classification of schizophrenic and control participants Expert Syst. Appl. 38 2063–71
[40] Taghavi M, Boostani R, Sabeti M and Taghavi S M 2011 Usefulness of approximate
entropy in the diagnosis of schizophrenia Iran. J. Psychiatry Behav. Sci. 5
[41] Akbar Y, Khotimah S N and Haryanto F 2016 Spectral and brain mapping analysis of
EEG based on Pwelch in schizophrenic patients J. Phys. Conf. Ser. 694 012070
[42] Sabeti M, Boostani R and Katebi S D 2007 A new approach to classify the schizophrenic
and normal subjects by finding the best channels and frequency bands 2007 15th Int. Conf.
on Digital Signal Processing
[43] Prasad P D, Halahalli H N, John J P and Majumdar K K 2014 Single-trial EEG
classification using logistic regression based on ensemble synchronization IEEE J.
Biomed. Health Inform. 18 1074–80
[44] Azlan W A and Low Y F 2014 Feature extraction of electroencephalogram (EEG) signal—
a review 2014 IEEE Conf. on Biomedical Engineering and Sciences (IECBES)
[45] Li Y, Tong S, Liu D, Gai Y, Wang X, Wang J, Qiu Y and Zhu Y 2008 Abnormal EEG
complexity in patients with schizophrenia and depression Clin. Neurophysiol. 119 1232–41
[46] Fenton G W, Fenwick P B, Dollimore J, Dunn T L and Hirsch S R 1980 EEG spectral
analysis in schizophrenia Br. J. Psychiatry 136 445–55
[47] Boostani R, Khadijeh S and Sabeti M 2009 An efficient classifier to diagnose of
schizophrenia based on the EEG signals Expert Syst. Appl. 36 6492–9
[48] Parvinnia E, Sabeti M, Zolghadri M J and Boostani R 2014 Classification of EEG signals
using adaptive weighted distance nearest neighbor algorithm J. King Saud Univ. Comput.
Inf. Sci. 26 1–6
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-22](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-48-320.jpg)
![[49] Abrams R and Taylor M A 1979 Differential EEG patterns in affective disorder and
schizophrenia Arch. Gen. Psychiatry 36 1355–8
[50] Saletu B, Anderer P, Saletu Z G M, Arnold O and Pascual-Marqui R D 2002 Classification
and evaluation of the pharmacodynamics of psychotropic drugs by single-lead pharmaco-
EEG, EEG mapping and tomography (LORETA) Methods Find. Exp. Clin. Pharmacol. 24
97–120
[51] Streit M, Wölwer W, Brinkmeyer J, Ihl R and Gaebel W 2001 EEG-correlates of facial
affect recognition and categorisation of blurred faces in schizophrenic patients and healthy
volunteers Schizophr. Res. 49 145–55
[52] Jeong J H, Kim D J, Chae J H, Kim S Y, Ko H J and Paik I H 1998 Nonlinear analysis of
the EEG of schizophrenics with optimal embedding dimension Med. Eng. Phys. 20 669–76
[53] Bob P 2012 Consciousness, schizophrenia and complexity Cogn. Syst. Res. 13 87–94
[54] Vasios C E, Papageorgiou C, Matsopoulos G K, Nikita K S and Uzunoglu N 2003 A
decision support system of evoked potentials for the classification of patients with first-
episode schizophrenia Ger. J. Psychiatry 5 05
[55] Cerquera A, Gjini K, Bowyer S M and Boutros N 2017 Comparing EEG nonlinearity in
deficit and nondeficit schizophrenia patients: preliminary data Clin. EEG Neurosci. 48 376–82
[56] Sui J, Castro E, He H, Bridwell D A, Du Y, Pearlson G D, Jiang T and Calhoun V D 2014
Combination of FMRI-SMRI-EEG data improves discrimination of schizophrenia patients
by ensemble feature selection 2014 36th Annual Int. Conf. of the IEEE Engineering in
Medicine and Biology Society pp 3889–92
[57] Aydın S, Arica N, Ergul E and Tan O 2015 Classification of obsessive compulsive disorder
by EEG complexity and hemispheric dependency measurements Int. J. Neural Syst. 25
1550010
[58] Akar S A, Kara S, Latifoğlu F and Bilgiç V 2016 Analysis of the complexity measures in the
EEG of schizophrenia patients Int. J. Neural Syst. 26 1650008
[59] Fan F-Y, Li Y-J, Qiu Y-H and Zhu Y-S 2005 Use of ANN and complexity measures in
cognitive EEG discrimination 2005 IEEE Engineering in Medicine and Biology 27th Annu.
Conf.
[60] Abhishekh H A, Thirthalli J, Manjegowda A, Phutane V H, Muralidharan K and
Gangadhar B N 2013 Ictal EEG fractal dimension in ECT predicts outcome at 2 weeks
in schizophrenia Psychiatry Res. 209 155–9
[61] Abhishekh H A, Thirthalli J, Manjegowda A, Phutane V H, Muralidharan K and
Gangadhar B N 2013 Ictal EEG fractal dimension in ECT predicts outcome at 2 weeks
in schizophrenia Psychiatry Res. 209 155–9
[62] Khodayari-Rostamabad A, Reilly J P, Hasey G, deBruin H and MacCrimmon D 2010
Diagnosis of psychiatric disorders using EEG data and employing a statistical decision
model 2010 Annual Int. Conf. of the IEEE Engineering in Medicine and Biology pp 4006–9
[63] Julieta R, Luis F, Cerdán M A, Guevara C A and Araceli S 2001 Abnormal EEG patterns
in treatment-resistant schizophrenic patients Int. J. Neurosci. 109 47–59
[64] Kaplan A Y, Borisov S V and Zheligovskiy V A 2005 Classification of the adolescent EEG
by the spectral and segmental characteristics for normals Zh. Vyssh. Nerv. Deiat. Im. I P
Pavlova 55 478–86
[65] AliMardani F, Boostani R and Blankertz B 2016 Presenting a spatial-geometric EEG
feature to classify BMD and schizophrenic patients Int. J. Adv. Telecommun. Electrotech.
Signals Syst. 5 79–85
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-23](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-49-320.jpg)
![[66] Youn T, Park H-J, Kim J-J, Kim M S and Kwon J S 2003 Altered hemispheric asymmetry
and positive symptoms in schizophrenia: equivalent current dipole of auditory mismatch
negativity Schizophr. Res. 59 253–60
[67] Shagass C, Roemer R A, Straumanis J J and Josiassen R C 1984 Psychiatric diagnostic
discriminations with combinations of quantitative EEG variables Br. J. Psychiatry 144
581–92
[68] Lou E, Zhang S and Qiao S 2009 Application of eigenvector estimation and SVM for EEG
signals classification 2009 9th Int. Conf. on Electronic Measurement & Instruments
[69] Raghavendra B S, Dutt D N, Halahalli H N and John J P 2009 Complexity analysis of
EEG in patients with schizophrenia using fractal dimension Physiol. Meas. 30 795–808
[70] Zhang S, Qiao S and Wang W 2010 Classification of schizophrenia’s EEG based on high
order pattern discovery 2010 IEEE Fifth Int. Conf. on Bio-Inspired Computing: Theories and
Applications (BIC-TA)
[71] Piryatinska A, Darkhovsky B and Kaplan A 2017 Binary classification of multichannel-
EEG records based on the ϵ-complexity of continuous vector functions Comput. Methods
Programs Biomed. 152 131–9
[72] Sabeti M, Mohammad H S and Pric G W 2007 Fuzzy accuracy-based classifier systems for
EEG classification of schizophrenic patients First Joint Congress on Fuzzy and Intelligent
Systems
[73] Phang C, Ting S, Samdin B and Ombao H 2019 Classification of EEG-based effective brain
connectivity in schizophrenia using deep neural networks 9th Int. IEEE/EMBS Conf. on
Neural Engineering (NER)
[74] Li F et al 2019 Differentiation of schizophrenia by combining the spatial EEG brain
network patterns of rest and task P300 IEEE Trans. Neural Syst. Rehabil. Eng. 27 594–602
[75] Chu W, Huang M, Jian B and Cheng K 2017 Analysis of EEG entropy during visual
evocation of emotion in schizophrenia Ann. Gener. Psychiatry 16 34
[76] Yin Z, Li J, Zhang Y, Ren A, Von Meneen K M and Huang L 2017 Functional brain
network analysis of schizophrenic patients with positive and negative syndrome based on
mutual information of EEG time series Biomed. Signal Process. Control 31 331–8
[77] Kirino E and Inoue R 1999 The relationship of mismatch negativity to quantitative EEG
and morphological findings in schizophrenia J. Psychiatr. Res. 33 445–56
[78] Tikka S K, Yadav S, Nizamie S H, Das B, Tikka D L and Goyal N 2014 Schneiderian first
rank symptoms and gamma oscillatory activity in neuroleptic naïve first episode schizo-
phrenia: a 192 channel EEG study Psychiatry Invest. 11 465–75
[79] Sabeti M, Behroozi R and Moradi E 2016 Analysing complexity, variability and spectral
measures of schizophrenic EEG signal Int. J. Biomed. Eng. Technol. 21 109
[80] Lifshitz K, Lee K L and Susswein S 1987 Long-term replicability of EEG spectra and
auditory evoked potentials in schizophrenic and normal subjects Neuropsychobiology 18
205–11
[81] Liu H, Zhang T, Ye Y, Pan C, Yang G, Wang J and Qiu R C 2017 A data driven approach
for resting-state EEG signal classification of schizophrenia with control participants using
random matrix theory, arXiv:1712.05289
[82] Röschke J, Fell J and Beckmann P 1995 Nonlinear analysis of sleep EEG data in
schizophrenia: calculation of the principal Lyapunov exponent Psychiatry Res. 56 257–69
[83] Merrin E L, Meek P, Floyd T C and Callaway E 1990 Topographic segmentation of
waking EEG in medication-free schizophrenic patients Int. J. Psychophysiol. 9 231–6
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-24](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-50-320.jpg)
![[84] Thilakvathi B, Shenbagadevi S, Bhanu K S and Malaippan M 2017 EEG signal complexity
analysis for schizophrenia during rest and mental activity Biomed. Res. 28 1–9
[85] Dvey-Aharon Z, Fogelson N, Peled A and Intrator N 2015 Schizophrenia detection and
classification by advanced analysis of EEG recordings using a single electrode approach
PLoS One 10 e0123033
[86] Kim J W, Lee Y S, Han D H, Min K J and Lee K 2015 Diagnostic utility of quantitative
EEG in un-medicated schizophrenia Neurosci. Lett. 589 126–31
[87] Dvey-A Z, Fogelson N, Peled A and Intrator N 2017 Connectivity maps based analysis of
EEG for the advanced diagnosis of schizophrenia attributes PLoS One 12 e0185852
[88] Buettner R, Beil D, Scholtz S and Djemai A 2020 Development of a machine learning based
algorithm to accurately detect schizophrenia based on one-minute EEG recordings HICSS-
53 Proc.: 53nd Hawaii Int. Conf. on System Sciences
[89] Etevenon P 1984 Intra and inter-hemispheric changes in alpha intensities in EEGs of
schizophrenic patients versus matched controls Biol. Psychol. 19 247–56
[90] Al-Fahoum A S and Al-Fraihat A 2014 Methods of EEG signal features extraction using
linear analysis in frequency and time-frequency domains ISRN Neurosci. 2014 730218
[91] Khaleghi A, Sheikhani A, Mohammadi M R, Nasrabadi A M, Vand S R, Zarafshan H and
Moeini M 2015 EEG classification of adolescents with type I and type II of bipolar disorder
Australas. Phys. Eng. Sci. Med. 38 551–9
[92] Bachiller A et al 2014 Decreased entropy modulation of EEG response to novelty and
relevance in schizophrenia during a P300 task Eur. Arch. Psychiatry Clin. Neurosci. 265 525–35
[93] Hiesh M, Lam A Y, Shen C, Chen W, Lin F, Sung H, Lin J, Chiu M and Lai F 2013
Classification of schizophrenia using genetic algorithm-support vector machine (GA-SVM)
2013 35th Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society
(EMBC) pp 6047–50
[94] Nasehi S and Pourghassem H 2011 A novel effective feature selection algorithm based on S-
PCA and wavelet transform features in EEG signal classification 2011 IEEE 3rd Int. Conf.
on Communication Software and Networks
[95] Doege K, Bates A T, White T P, Das D, Boks M P and Liddle P F 2009 Reduced event-
related low frequency EEG activity in schizophrenia during an auditory oddball task
Psychophysiology 46 566–77
[96] Akar S A, Kara S, Latifoglu F and Bilgic V 2012 Wavelet–Welch methodology for analysis of
EEG signals of schizophrenia patients 2012 Cairo Int. Biomedical Engineering Conf. (CIBEC)
[97] Pauline A D 1942 Comparative study of the EEGs of schizophrenic and manic-depressive
patients Am. J. Psychiatry 99 210–7
[98] Haring C, Neudorfer C, Schwitzer J, Hummer M, Saria A, Hinterhuber H and
Fleischhacker W W 1994 EEG alterations in patients treated with clozapine in relation
to plasma levels Psychopharmacology 114 97–100
[99] Zhao Q, Hu B J, Li Y, Peng H, Li L, Liu Q, Li Y D, Shi Q and Feng J 2013 An alpha
resting EEG study on nonlinear dynamic analysis for schizophrenia 2013 6th Int. IEEE/
EMBS Conf. on Neural Engineering (NER) pp 484–8
[100] Ziqiang Z and Puthusserypady S 2007 Analysis of schizophrenic EEG synchrony using
empirical mode decomposition 2007 15th Int. Conf. on Digital Signal Processing
[101] Phang C R, Ting C M, Numan F and Ombao H 2019 Classification of EEG-based brain
connectivity networks in schizophrenia using a multi-domain connectome convolutional
neural network, arXiv:1903.08858
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-25](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-51-320.jpg)
![[102] Ford J M, Palzes V A, Roach B J and Mathalon D H 2014 Did I do that? Abnormal
predictive processes in schizophrenia when button pressing to deliver a tone Schizophr. Bull.
40 804–12
[103] Gilles J 2013 Empirical wavelet transform IEEE Trans. Signal Process. 61 3999
[104] Song Y Y and Lu Y 2015 Decision tree methods: applications for classification and
prediction Shanghai Arch. Psychiatry 27 130–5
[105] Awad M and Khanna R 2015 Support vector machines for classification Efficient Learning
Machines (Berkeley, CA: Apress)
[106] Adeniyi D A, Wei Z and Yong-quan Y 2016 Automated web usage data mining and
recommendation system using k-nearest neighbor (KNN) classification method Appl.
Comput. Inform. 12 1
[107] Bühlmann P 2012 Bagging, boosting and ensemble methods Handbook of Computational
Statistics (Berlin: Springer)
[108] Tharwat A 2016 Linear vs quadratic discriminant analysis classifier: a tutorial Int. J. Appl.
Pattern Recogn. 3 145
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
1-26](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-52-320.jpg)
![2.1 Introduction
This section provides a brief introduction to the background of post-traumatic
stress disorder (PTSD), as well as resting state eyes-closed and eyes-open EEG
signals. In the following, we will mention some points about phase synchrony
calculation.
According to the American Psychiatric Association (APA), post-traumatic stress
disorder (PTSD) is defined as a psychiatric disorder which may follow a traumatic
event [1]. PTSD is a chronic debilitating anxiety condition which is characterized by
unremitting distressing repetition of the traumatic experience, avoidance, hyper-
arousal, hyper-vigilance, dissociation, emotional numbing and negative alteration in
cognition [2].
fMRI studies have reported functional abnormalities in the cortical and sub-
cortical regions of PTSD patients’ brains [3, 4]. Functional connectivity is a powerful
approach for investigating the brain as a sophisticated network. Many disorders
have been studied using functional connectivity, including PTSD [5–8]. Although
studying abnormal patterns in the presence of emotional elicitation is of great
importance, dysregulated patterns of the resting state functional connectivity may
provide valuable knowledge about PTSD pathology [9, 10]. Correlations embedded
in haemodynamic activity levels among various brain regions are addressed by
resting state functional connectivity. Functional connectivity uses synchronization
of the neural activation of the aforementioned regions [11], which can be calculated
using different modalities, based on various methods.
EEG is a low cost, commonly available modality with high temporal resolution.
It can provide optimal observation of brain activities [12, 13]. Only a few studies
have focused on the resting state functional connectivity of PTSD patients based on
EEG signals. Resting state EEG can be recorded in two forms, including resting
state eyes-closed (REC), and resting state eyes-open (REO). REC and REO have
different basic properties. For example, the alpha band EEG is defined using
synchronization in REC but desynchronization in REO as the feature [14].
Comparing REC and REO demonstrates differences in the delta, theta, alpha and
beta sub-bands, both in adult and young participants [15]. One hypothesis states that
the differences between REC and REO can be translated as a signal which reflects
the brain’s activity in response to visual stimuli. To test this hypothesis, REC and
REO EEG signals have been recorded in a completely dark environment. The
reported results imply significantly different spectral powers and coherence values in
the delta, theta, alpha1, alpha2, beta1, beta2 and gamma sub-bands. These findings
suggest that the differences between REC and REO are independent of external
stimuli to the visual system. This study also proposes that such differences actually
reflect externally directed attention and, in contrast, internally directed attention,
specific to REO and REC, respectively [16].
In this study we aim to examine the REC and REO signals recorded from combat
related PTSD participants, and two control groups including combat trauma
exposed non-PTSD participants and healthy controls who have not experienced
any serious trauma including war. For this purpose, we will use functional
Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
2-2](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-54-320.jpg)
![This ebook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this ebook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.
Title: Over the Ocean; or, Sights and Scenes in Foreign Lands
Author: Curtis Guild
Release date: October 30, 2012 [eBook #41233]
Most recently updated: October 23, 2024
Language: English
Credits: Produced by K Nordquist, Charlie Howard, and the
Online
Distributed Proofreading Team at http://www.pgdp.net
(This
book was produced from scanned images of public
domain
material from the Google Print project.)
*** START OF THE PROJECT GUTENBERG EBOOK OVER THE
OCEAN; OR, SIGHTS AND SCENES IN FOREIGN LANDS ***](https://image.slidesharecdn.com/18855745-250612191729-e653e542/85/Modelling-And-Analysis-Of-Active-Biopotential-Signals-In-Healthcare-Volume-1-Sinha-Bajaj-60-320.jpg)
Modelling And Analysis Of Active Biopotential Signals In Healthcare Volume 1 Sinha Bajaj
- 1. Modelling And Analysis Of Active Biopotential Signals In Healthcare Volume 1 Sinha Bajaj download https://ebookbell.com/product/modelling-and-analysis-of-active- biopotential-signals-in-healthcare-volume-1-sinha-bajaj-37711490 Explore and download more ebooks at ebookbell.com
- 2. Here are some recommended products that we believe you will be interested in. You can click the link to download. Modelling And Analysis Of Sustainability Related Issues In New Era Wenhsien Tsai Editor https://ebookbell.com/product/modelling-and-analysis-of- sustainability-related-issues-in-new-era-wenhsien-tsai-editor-36371384 Modelling And Analysis Of Hybrid Supervisory Systems A Petri Net Approach 1st Edition Emilia Villani Phd https://ebookbell.com/product/modelling-and-analysis-of-hybrid- supervisory-systems-a-petri-net-approach-1st-edition-emilia-villani- phd-4192262 Modelling And Analysis Of Fine Sediment Transport In Wavecurrent Bottom Boundary Layer Liqin Zuo https://ebookbell.com/product/modelling-and-analysis-of-fine-sediment- transport-in-wavecurrent-bottom-boundary-layer-liqin-zuo-7160976 Numerical Modelling And Analysis Of Fluid Flow And Deformation Of Fractured Rock Masses First Edition 1st Xing Zhang https://ebookbell.com/product/numerical-modelling-and-analysis-of- fluid-flow-and-deformation-of-fractured-rock-masses-first-edition-1st- xing-zhang-2010838
- 3. Mathematical Modelling And Analysis Of Infectious Diseases 1st Ed Khalid Hattaf https://ebookbell.com/product/mathematical-modelling-and-analysis-of- infectious-diseases-1st-ed-khalid-hattaf-22503324 The Modelling And Analysis Of The Mechanics Of Ropes 1st Edition Cm Leech Auth https://ebookbell.com/product/the-modelling-and-analysis-of-the- mechanics-of-ropes-1st-edition-cm-leech-auth-4625914 Empirical Finance Modelling And Analysis Of Emerging Financial And Stock Markets 1st Edition Dr Sardar M N Islam https://ebookbell.com/product/empirical-finance-modelling-and- analysis-of-emerging-financial-and-stock-markets-1st-edition-dr- sardar-m-n-islam-4267950 Structural Mechanics Modelling And Analysis Of Frames And Trusses 1st Edition Karlgunnar Olsson https://ebookbell.com/product/structural-mechanics-modelling-and- analysis-of-frames-and-trusses-1st-edition-karlgunnar-olsson-5393636 Formal Techniques Modelling And Analysis Of Timed And Faulttolerant Systems Joint International Conferences On Formal Modeling And Analysis Of Timed Systmes Formats 2004 And Formal Techniques In Realtime And Fault Tolerant Systems Ftrtft 2004 Gre 1st Edition Paul Feautrier Auth https://ebookbell.com/product/formal-techniques-modelling-and- analysis-of-timed-and-faulttolerant-systems-joint-international- conferences-on-formal-modeling-and-analysis-of-timed-systmes- formats-2004-and-formal-techniques-in-realtime-and-fault-tolerant- systems-ftrtft-2004-gre-1st-edition-paul-feautrier-auth-4604272
- 6. Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1
- 7. IPEM–IOP Series in Physics and Engineering in Medicine and Biology Editorial Advisory Board Members Frank Verhaegen Maastro Clinic, the Netherlands Carmel Caruana University of Malta, Malta Penelope Allisy-Roberts formerly of BIPM, Sèvres, France Rory Cooper University of Pittsburgh, USA Alicia El Haj University of Birmingham, UK Kwan Hoong Ng University of Malaya, Malaysia John Hossack University of Virginia, USA Tingting Zhu University of Oxford, UK Dennis Schaart TU Delft, the Netherlands Indra J Das New York University, USA About the Series Series in Physics and Engineering in Medicine and Biology will allow IPEM to enhance its mission to ‘advance physics and engineering applied to medicine and biology for the public good.’ Focusing on key areas including, but not limited to: • clinical engineering • diagnostic radiology • informatics and computing • magnetic resonance imaging • nuclear medicine • physiological measurement • radiation protection • radiotherapy • rehabilitation engineering • ultrasound and non-ionising radiation. A number of IPEM–IOP titles are published as part of the EUTEMPE Network Series for Medical Physics Experts.
- 8. Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 Edited by Varun Bajaj PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India G R Sinha Myanmar Institute of Information Technology, Mandalay, Myanmar IOP Publishing, Bristol, UK
- 9. ª IOP Publishing Ltd 2020 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the publisher, or as expressly permitted by law or under terms agreed with the appropriate rights organization. Multiple copying is permitted in accordance with the terms of licences issued by the Copyright Licensing Agency, the Copyright Clearance Centre and other reproduction rights organizations. Permission to make use of IOP Publishing content other than as set out above may be sought at permissions@ioppublishing.org. Varun Bajaj and G R Sinha have asserted their right to be identified as the authors of this work in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. ISBN 978-0-7503-3279-8 (ebook) ISBN 978-0-7503-3277-4 (print) ISBN 978-0-7503-3280-4 (myPrint) ISBN 978-0-7503-3278-1 (mobi) DOI 10.1088/978-0-7503-3279-8 Version: 20200801 IOP ebooks British Library Cataloguing-in-Publication Data: A catalogue record for this book is available from the British Library. Published by IOP Publishing, wholly owned by The Institute of Physics, London IOP Publishing, Temple Circus, Temple Way, Bristol, BS1 6HG, UK US Office: IOP Publishing, Inc., 190 North Independence Mall West, Suite 601, Philadelphia, PA 19106, USA
- 10. Dedicated to my father Late Mahendra Bajaj and my family members. Varun Bajaj Dedicated to my late grandparents, my teachers and Revered Swami Vivekananda. G R Sinha
- 12. Contents Preface xiv Acknowledgements xv Editor biographies xvi Contributor list xviii 1 Classification of schizophrenia patients through empirical wavelet transformation using electroencephalogram signals 1-1 Smith K Khare, Varun Bajaj, Siuly Siuly and G R Sinha 1.1 Introduction 1-1 1.2 Methodology 1-5 1.2.1 Dataset 1-5 1.2.2 Empirical wavelet transform 1-6 1.2.3 Feature extraction 1-7 1.2.4 Classification techniques 1-9 1.2.5 Performance parameters 1-10 1.3 Results and discussion 1-10 1.4 Conclusion 1-20 References 1-20 2 Fuzzy scale invariant feature transform phase locking value and its application to PTSD EEG data 2-1 Zahra Ghanbari and Mohammad H Moradi 2.1 Introduction 2-2 2.2 Method 2-4 2.2.1 FSIFT-PLV 2-4 2.2.2 Functional connectivity graph indices 2-8 2.3 Data 2-9 2.3.1 Synthetic data 2-9 2.3.2 EEG data 2-10 2.4 Results 2-12 2.4.1 Synthetic EEG data 2-13 2.4.2 Real EEG data 2-15 2.5 Conclusion 2-21 Acknowledgments 2-23 References 2-23 vii
- 13. 3 Weighted complex network based framework for epilepsy detection from EEG signals 3-1 Supriya Supriya, Siuly Siuly, Hua Wang and Yanchun Zhang 3.1 Introduction 3-1 3.2 Weighted complex network based framework 3-5 3.2.1 Conversion of EEG signals into the WCN 3-5 3.2.2 Statistical feature extraction from the WCN 3-6 3.2.3 Evaluation of the AWD using classifiers 3-7 3.2.4 Evaluation of performance 3-12 3.3 Experimental results and discussion 3-13 3.3.1 Experimental data 3-13 3.3.2 Results and discussion 3-14 3.4 Conclusion 3-18 References 3-18 4 Epileptic seizure prediction and onset zone localization using intracranial and scalp electroencephalographic and magnetoencephalographic signals 4-1 Hamid Reza Marateb, Carolina Migliorelli, Alejandro Bachiller, Tayebe Azimi, Farzad Ziaie Nezhad, Marjan Mansourian, Joan Francesc Alonso, Javier Aparicio, Maria Victoria San Antonio-Arce, Sergio Romero and Miguel Ángel Mañanas 4.1 Epileptic seizure prediction 4-2 4.2 Seizure onset zone identification 4-6 4.3 Performance indices 4-12 4.4 Conclusion and future scope 4-12 Acknowledgments 4-13 References 4-13 5 Automatic drowsiness detection based on variational non-linear chirp mode decomposition using electroencephalogram signals 5-1 Smith K Khare, Varun Bajaj and G R Sinha 5.1 Introduction 5-1 5.2 Methodology 5-4 5.2.1 Dataset 5-4 5.2.2 Variational non-linear chirp mode decomposition (VNCMD) 5-5 5.2.3 Feature extraction 5-8 5.2.4 Classifiers 5-10 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 viii
- 14. 5.3 Results and discussion 5-10 5.4 Conclusion 5-18 References 5-19 6 Noise removal and classification of EEG signals using the Fourier decomposition method 6-1 Virender Kumar Mehla, Ashish Kumar, Amit Singhal and Pushpendra Singh 6.1 Introduction 6-1 6.2 Related work 6-2 6.3 Proposed work 6-4 6.3.1 Dataset 6-4 6.3.2 The Fourier decomposition method 6-5 6.4 Classification 6-13 6.5 Experimental results and discussion 6-16 6.6 Conclusion and proposed future scope 6-24 References 6-24 7 Reliable and accurate information extraction from surface electromyographic signals 7-1 Hamid Reza Marateb, Mislav Jordanic, Monica Rojas-Martı́nez, Joan Francesc Alonso, Leidy Yanet Serna, Mehdi Shirzadi, Marjan Nosouhi, Miguel Ángel Mañanas and Kevin C McGill 7.1 Surface electromyography 7-1 7.2 Surface EMG applications 7-3 7.3 Challenges in sEMG recording 7-4 7.4 Detection of atypical signals in HD-sEMG 7-7 7.4.1 Feature extraction 7-9 7.4.2 Detection methods 7-9 7.5 Myoelectric prosthesis control, a hot topic 7-10 7.6 Conclusion and future scope 7-12 Acknowledgments 7-12 References 7-13 8 Classification of physical actions from surface EMG signals using the wavelet packet transform and local binary patterns 8-1 Ömer F Alçin, Ümit Budak, Muzaffer Aslan, Yaman Akbulut, Zafer Cömert, Muhammed H Akpınar and Abdulkadir Şengür 8.1 Introduction 8-2 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 ix
- 15. 8.2 Materials and methods 8-6 8.2.1 The wavelet transform 8-6 8.2.2 The one-dimensional LBP 8-7 8.2.3 The support vector machine classifier 8-8 8.2.4 The decision tree classifier 8-10 8.2.5 The ensemble bagging classifier 8-11 8.2.6 The ensemble boosting classifier 8-11 8.2.7 The k-nearest neighbor classifier 8-12 8.2.8 The linear discriminant classifier 8-12 8.3 Experimental work and results 8-14 8.4 Conclusion 8-18 References 8-20 9 Empirical wavelet transform based classification of surface electromyogram signals for hand movements 9-1 Anurag Nishad and Abhay Upadhyay 9.1 Introduction 9-2 9.2 Dataset 9-4 9.3 Overview of empirical wavelet transform 9-5 9.4 The proposed method 9-8 9.4.1 EWT based decomposition 9-8 9.4.2 Feature computation 9-9 9.4.3 Feature ranking 9-10 9.4.4 Classification 9-11 9.5 Simulation results 9-12 9.6 Discussion 9-22 9.7 Conclusion and future scope 9-28 References 9-28 10 Analysis of the muscular activity pattern of recurring physical action 10-1 Ajay Somkuwar and Vandana Somkuwar 10.1 Introduction 10-1 10.2 Analytical expressions of joint moments 10-2 10.2.1 Data collection and joint moment analysis 10-5 10.3 Myoelectric signals during recursive work 10-8 10.3.1 The major muscles of the lower extremity 10-10 10.3.2 Data collection and subjects 10-10 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 x
- 16. 10.3.3 The myoelectrical signal, electrodes and recording 10-10 10.3.4 Crosstalk and muscle movement artefacts 10-11 10.3.5 The measured electromyogram 10-12 10.3.6 Muscle activity pattern 10-13 10.4 Joint force estimation 10-14 10.4.1 Joint moment pattern and performance measurement 10-19 10.5 Conclusions 10-20 References 10-21 11 Cloud-based cardiac health monitoring using event-driven ECG processing and ensemble classification techniques 11-1 Saeed M Qaisar and Abdulhamit Subasi 11.1 Introduction 11-1 11.2 Background and literature review 11-4 11.3 ECG in healthcare 11-5 11.4 The proposed approach 11-7 11.4.1 Dataset 11-7 11.4.2 The event-driven acquisition 11-8 11.4.3 The event-driven segmentation 11-9 11.4.4 The adaptive rate resampling and denoising 11-9 11.4.5 Extraction of features 11-10 11.4.6 Machine learning methods 11-11 11.5 The performance evaluation measures 11-13 11.5.1 Compression ratio 11-13 11.5.2 Computational complexity 11-14 11.5.3 Classification accuracy 11-15 11.6 Experimental results and discussion 11-15 11.6.1 Experimental results 11-15 11.6.2 Discussion 11-20 11.7 Conclusion 11-21 Acknowledgments 11-22 References 11-22 12 Electrocardiogram beat classification using deep convolutional neural network techniques 12-1 Zafer Cömert, Yaman Akbulut, Muhammed H Akpinar, Ömer F Alçin, Ümit Budak, Muzaffer Aslan and Abdulkadir Şengür 12.1 Introduction 12-2 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 xi
- 17. 12.2 Material and methods 12-7 12.2.1 The MIT-BIH database 12-7 12.2.2 Producing ECG beat images 12-7 12.2.3 Convolutional neural networks (CNNs) 12-8 12.2.4 Deep transfer learning (DTL) 12-10 12.2.5 Support vector machines 12-11 12.2.6 Performance metrics 12-12 12.3 Experimental work and results 12-12 12.4 Discussion 12-17 12.5 Conclusion 12-21 References 12-21 13 ECG signal watermarking to enhance the security of telecardiology 13-1 Siddharth Bhalerao, Irshad A Ansari and Anil Kumar 13.1 Introduction 13-1 13.2 Preliminaries 13-5 13.3 Prediction error expansion 13-5 13.4 Prediction scheme and ECG database 13-6 13.4.1 Deep neural network 13-6 13.4.2 ECG database 13-7 13.5 Training and embedding 13-8 13.5.1 Training 13-9 13.5.2 Embedding scheme 1 13-9 13.5.3 Embedding scheme 2 13-10 13.5.4 Embedding scheme 3 13-12 13.6 Improved embedding scheme 13-14 13.6.1 The effect of ECG abnormalities 13-19 13.6.2 Performance on the ECG-ID database 13-20 13.7 Conclusion 13-21 References 13-23 14 Statistical measures and analysis in electrocardiogram (ECG) signal processing 14-1 Ranjeet Kumar 14.1 Introduction 14-1 14.2 The electrocardiogram (ECG) signal and its characteristics 14-3 14.2.1 ECG signal generation 14-4 14.2.2 ECG signal characteristics 14-9 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 xii
- 18. 14.3 Statistical measures and analysis 14-10 14.4 Statistical analysis in ECG signal processing 14-13 14.5 Conclusion 14-16 References 14-16 15 The impact of regional atrophy on Alzheimer’s disease and its identification using 3D texture analysis 15-1 Shiwangi Mishra and Pritee Khanna 15.1 Introduction 15-1 15.2 Regional atrophy and Alzheimer’s disease 15-4 15.3 Related works 15-5 15.3.1 VBM based methods 15-5 15.3.2 Texture analysis based methods 15-5 15.3.3 Shape analysis based methods 15-7 15.3.4 Other methods 15-7 15.4 Materials and methods 15-8 15.4.1 Dataset 15-8 15.4.2 The proposed methodology 15-9 15.5 Experiments and results 15-16 15.5.1 Experiment 1: Voxel as features (VAF) obtained from GM and WM regions 15-17 15.5.2 Experiment 2: Volumetric features evaluated on the 3D-DWT sub-bands obtained from all 116 regions 15-17 15.5.3 Experiment 3: Volumetric features evaluated on 3D-DWT sub-bands obtained from the top five regions 15-18 15.5.4 Experiment 4: Features obtained after applying feature selection on the features of the top five selected regions 15-18 15.5.5 Performance comparison with state-of-art methods 15-19 15.6 Conclusions 15-21 Acknowledgments 15-21 References 15-22 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 xiii
- 19. Preface Bio-potential signals are often used by physicians for monitoring the pathological and physiological conditions of human organs. These signals originate from biological systems which include the nervous, cardiovascular and musculoskeletal systems, etc. Modelling and analysis of bio-potential signals involves the manipu- lation or transformation of the signals to enhance the relevant information for the improvement of healthcare systems. This book, Modelling and Analysis of Active Bio-potential Signals in Healthcare, provides modelling of biomedical signals for understanding physiology, which can help to improve healthcare systems for the diagnosis and identification of disorders related to human organs. Therefore, this book addresses the need for a better understanding of the behaviours, concepts, fundamentals, principles, case studies, etc, of human organs for healthcare systems so that future research can be planned and carried out more effectively and the results strengthened. Biomedical signal applications are and will be used extensively in a huge number of research works and real-time applications for the improvement of healthcare systems using science, engineering and technology. This first volume of the book provides the framework for modelling, the concepts and the applications of bio-potential signal processing. This book also emphasizes the real-time challenges in bio-potential signal processing due to the complex and non-stationary nature of the signals that are used for a variety of applications in the analysis, classification and identification of different states for the improvement of healthcare systems. Each chapter begins with a description of a biomedical example and the significance of the methods, with discussion to connect the technology with an understanding of the human organs. This book also provides information for the identification of diseases such as schizophrenia, epileptic seizures, physical action, cardiac health monitoring, etc. Moreover, the chapters can be read independently by research scholars, practising physicians, R&D engineers and graduate students who wish to explore research in the field of biomedical engineering. xiv
- 20. Acknowledgements Dr Bajaj expresses his heartfelt appreciation to his mother Prabha, wife Anuja and daughter Avadhi for their wonderful support and encouragement throughout the completion of this important book. His deepest gratitude goes to his mother-in-law and father-in-law for their constant motivation. This book is the outcome of sincere effort that could only be achieved due to the great support of his family. He also extends his thanks to Professor Sanjeev Jain, Director of PDPM IIITDM, Jabalpur, for his support and encouragement. Dr Sinha expresses his gratitude and sincere thanks to his family members, wife Shubhra, daughter Samprati, parents and teachers. We would like to thank all our friends, well-wishers and all those who keep us motivated to do more and more, better and better. We sincerely thank all the contributors for their writing on the relevant theoretical background and real-time applications of bio-potential signals for healthcare. We are also deeply grateful to many whose names are not mentioned here but whose help during this work we appreciate and wish to acknowledge. We express our humble thanks to Michael Slaughter (Senior Commissioning Editor), Sarah Armstrong (Editorial Assistant) and the staff of IOP Publishing for their great support, necessary help, appreciation and quick responses. We also wish to thank Jessica Fricchione and Emily Tapp for their support during the review and approval of the book proposal. We also wish to thank IOP Publishing for giving us this opportunity to contribute on a relevant topic with a reputed publisher. Finally, we want to thank everyone who, in one way or another, helped us in editing this book. Dr Bajaj, in particular, thanks his family who provided encouragement through- out the editing of this book. This book is whole-heartedly dedicated to his father who took the lead to heaven before the completion of this book. Last, but not least, we would also like to thank God for showering us with his blessings and strength to perform novel and quality work of this type. Varun Bajaj G R Sinha xv
- 21. Editor biographies Varun Bajaj Varun Bajaj has been working as a faculty in the discipline of Electronics and Communication Engineering, at Indian Institute of Information Technology, Design and Manufacturing (IIITDM) Jabalpur, India since 2014. He worked as a visiting faculty in IIITDM Jabalpur from September 2013 to March 2014. He served as an Assistant Professor at Department of Electronics and Instrumentation, Shri Vaishnav Institute of Technology and Science, Indore, India during 2009–2010. He received B.E. degree in Electronics and Communication Engineering from Rajiv Gandhi Technological University, Bhopal, India in 2006, M.Tech. Degree with Honors in Microelectronics and VLSI design from Shri Govindram Seksaria Institute of Technology & Science, Indore, India in 2009. He received his PhD degree in the Discipline of Electrical Engineering, at Indian Institute of Technology Indore, India in 2014. He is also serving as a Subject Editor-in-Chief of IET Electronics Letters. He served as a Subject Editor of IET Electronics Letters November 2018 to June 2020. He is Senior Member IEEE June 2020, MIEEE 16-20, and also contributing as active technical reviewer of leading International journals of IEEE, IET, and Elsevier, etc. He has authored more than 100 research papers in various reputed international journals/conferences like IEEE Transactions, Elsevier, Springer, IOP etc. He has edited Modelling and Analysis of Active Biopotential Signals in Healthcare Volumes 1 and 2 published by IOP Publishing. He also edited a book by CRC Press. The citation impact of his publications is around 1800 citations, an h- index of 19, and i10 index of 40 (Google Scholar July 2020). He has guided Six (03 Competed 3 Ongoing) PhD Scholars, 5 M. Tech. Scholars. He is a recipient of various reputed national and international awards. His research interests include biomedical signal processing, image processing, time-frequency analysis, and com- puter-aided medical diagnosis. G R Sinha G R Sinha is an Adjunct Professor at the International Institute of Information Technology Bangalore (IIITB) and currently deputed as a Professor at the Myanmar Institute of Information Technology (MIIT), Mandalay, Myanmar. He obtained his BE in Electronics Engineering and MTech in Computer Technology with a gold medal from the National Institute of Technology Raipur, India. He received his PhD in Electronics and Telecommunication Engineering from Chhattisgarh Swami Vivekanand Technical University (CSVTU), Bhilai, India. He is a Visiting Professor (Honorary) at the Sri Lanka Technological Campus Colombo for the year 2019–20. He has published 254 research papers, book chapters and books at the international and national level, xvi
- 22. that include Biometrics (published by Wiley India, a subsidiary of John Wiley), Medical Image Processing (published by Prentice Hall of India), and five edited books with IOP, Elsevier and Springer. He is an active reviewer and editorial member of more than 12 reputed international journals published by IEEE, IOP, Springer, Elsevier, etc. He has teaching and research experience of 21 years. He has been the Dean of Faculty and an Executive Council Member of CSVTU, and is currently a member of the Senate of MIIT. Dr Sinha has been delivering ACM lectures across the world as an ACM Distinguished Speaker in the field of DSP since 2017. A few of his more important roles include the Expert Member for Vocational Training Programme by Tata Institute of Social Sciences (TISS) for two years (2017–19), Chhattisgarh Representative of IEEE MP Sub-Section Executive Council (2016–19) and Distinguished Speaker in the field of Digital Image Processing by the Computer Society of India (2015). He is the recipient of many awards and recognitions, such as the TCS Award 2014 for Outstanding Contributions in the Campus Commune of TCS, Rajaram Bapu Patil ISTE National Award 2013 for Promising Teacher in Technical Education by ISTE New Delhi, Emerging Chhattisgarh Award 2013, Engineer of the Year Award 2011, Young Engineer Award 2008, Young Scientist Award 2005, IEI Expert Engineer Award 2007, ISCA Young Scientist Award 2006 Nomination and Deshbandhu Merit Scholarship for five years. He served as a Distinguished IEEE Lecturer in the IEEE India Council for the Bombay section. He is a Senior Member of IEEE, a Fellow of the Institute of Engineers India and a Fellow of IETE India. He has delivered more than 50 keynote/invited talks and has chaired many technical sessions at international conferences across the world. His Special Session on ‘Deep Learning in Biometrics’ was included in the IEEE International Conference on Image Processing 2017. He is also a member of many national professional bodies such as ISTE, CSI, ISCA and IEI. He is a member of various committees of the University and has been Vice President of the Computer Society of India for the Bhilai chapter for two consecutive years. He is a consultant for various skill development initiatives of NSDC, the Government of India. He is a regular referee of project grants under the DST-EMR scheme and several other schemes of the Government of India. He has received important consultancy support, such as grants and travel support. Dr Sinha has supervised eight PhD scholars, 15 MTech scholars and is currently supervising another PhD scholar. His research interests include biometrics, cognitive science, medical image processing, computer vision, outcome based education (OBE) and ICT tools for developing employability skills. Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 xvii
- 23. Contributor list Smith K Khare Electronics and Communication Department, PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India Varun Bajaj Electronics and Communication Department, PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India Siuly Siuly Institute for Sustainable Industries and Liveable Cities, Footscray Park Victoria University, Australia G R Sinha Myanmar Institute of Information Technology (MIIT), Mandalay, Myanmar Zahra Ghanbari Amirkabir University of Technology, Tehran, Iran Mohammad Hassan Moradi Amirkabir University of Technology, Tehran, Iran Supriya Supriya Institute for Sustainable Industries and Liveable Cities, Footscray Park Victoria University, Australia Hua Wang Institute for Sustainable Industries and Liveable Cities, Footscray Park Victoria University, Australia Yanchun Zhang Institute for Sustainable Industries and Liveable Cities, Footscray Park Victoria University, Australia Hamid Reza Marateb Biomedical Engineering Department, Engineering Faculty, University of Isfahan, Isfahan, Iran Carolina Migliorelli Falcone CIBER-BBN, UPC, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), Barcelona, Spain Alejandro Bachiller Matarranz Universitat Politècnica de Catalunya—BarcelonaTech, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), Barcelona, Spain Tayebe Azimi Biomedical Engineering Department, Engineering Faculty, University of Isfahan, Isfahan, Iran xviii
- 24. Farzad Ziaie Nezhad Biomedical Engineering Department, Engineering Faculty, University of Isfahan, Isfahan, Iran Marjan Mansourian Biostatistics and Epidemiology Department, Faculty of Health, Isfahan University of Medical Sciences, Isfahan, Iran Joan Francesc Alonso López Universitat Politècnica de Catalunya—BarcelonaTech, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), Barcelona, Spain Javier Aparicio Epilepsy Unit, Department of Neuropediatrics, Universitary Hospital Sant Joan de Déu Barcelona, Spain Maria Victoria San Antonio Arce Epilepsy Unit, Department of Neuropediatrics, Universitary Hospital Sant Joan de Déu Barcelona, Spain Freiburg Epilepsy Centre, Medical Center—University of Freiburg, Germany Sergio Romero Lafuente Universitat Politècnica de Catalunya—BarcelonaTech, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), CREB, Barcelona, Spain Miguel Angel Mañanas Villanueva Universitat Politècnica de Catalunya—BarcelonaTech, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), Barcelona, Spain Virender Mehla Bennett University, Greater Noida, Uttar Pradesh, India Ashish Singh Bennett University, Greater Noida, Uttar Pradesh, India Pushpendra Singh National Institute of Technology Hamirpur, Himachal Pradesh, India Amit Singhal Bennett University, Greater Noida, Uttar Pradesh, India Mislav Jordanić Universitat Politècnica de Catalunya—BarcelonaTech, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), Barcelona, Spain Mónica Rojas-Martínez Universidad El Bosque, Programa de Bioingeniería, Universidad El Bosque, Bogotá, Colombia Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 xix
- 25. Leidy Yanet Serna Higuita Universitat Politècnica de Catalunya—BarcelonaTech, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), Barcelona, Spain Mehdi Shirzadi Universitat Politècnica de Catalunya—BarcelonaTech, Escola Tècnica Superior d’Enginyeria Industrial de Barcelona (ETSEIB), Barcelona, Spain Marjan Nosouhi Biomedical Engineering Department, Engineering Faculty, University of Isfahan, Isfahan, Iran Kevin McGill US Department of Veterans Affairs, United States Ömer Faruk Alçin Malatya Turgut Ozal University, Faculty of Engineering and Natural Sciences, Department of Electrical Engineering, Malatya, Turkey Ümit Budak Bitlis Eren University, Engineering Faculty, Electrical and Electronics Engineering Department, Bitlis, Turkey Muzaffer Aslan Bingol University, Engineering Faculty, Electrical-Electronics Engineering Department, Bingol, Turkey Yaman Akbulut Firat University, Informatics Department, Elazig, Turkey Zafer Cömert Samsun University, Engineering Faculty, Software Engineering Department, Samsun, Turkey Muhammed H Akpınar Firat University, Technology Faculty, Electrical and Electronics Engineering Department, Elazig, Turkey Abdulkadir Şengür Firat University, Technology Faculty, Electrical and Electronics Engineering Department, Elazig, Turkey Anurag Nishad EEE Department, BITS Pilani, KK Birla Goa Campus, India, India Abhay Upadhyay IET, Bundelkhand University Jhansi, UP, India, India Ajay Somkuwar Department of Electronics and Communication, MANIT Bhopal MP, INDIA Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 xx
- 26. Vandana Somkuwar Department of Mechanical Engineering, NITTTR Bhopal MP, INDIA Abdulhamit Subasi Effat University, College of Engineering, Jeddah, Saudi Arabia Saeed Mian Qaisar Effat University, College of Engineering, Jeddah, Saudi Arabia Siddharth Bhalerao Electronics and Communication Department, PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India Irshad Ahmad Ansari Electronics and Communication Department, PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India Anil Kumar Electronics and Communication Department, PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India Ranjeet Kumar School of Electronics Engineering (SENSE), VIT University, Tamilnadu, India Shiwangi Mishra Electronics and Communication Department, PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India Pritee Khanna Electronics and Communication Department, PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 xxi
- 27. IOP Publishing Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 Varun Bajaj and G R Sinha Chapter 1 Classification of schizophrenia patients through empirical wavelet transformation using electroencephalogram signals Smith K Khare1 , Varun Bajaj1 , Siuly Siuly2 , G R Sinha3 1 PDPM-Indian Institute of Information Technology, Design and Manufacturing, Jabalpur, India 2 Institute for Sustainable Industries & Liveable Cities, Victoria University, Australia 3 Myanmar Institute of Information Technology, Mandalay, Myanmar Schizophrenia is a chronic and complex mental health disorder characterized by symptoms such as delusions, disorganized speech or behavior, hallucinations and impaired cognitive ability. Electroencephalogram (EEG) signals can provide detailed information about the brain activity associated with the behavioral changes associated with schizophrenia. Accurate and timely detection of this disease can help in diagnosis. In this chapter, empirical wavelet transformation is used to decompose the highly non- stationary EEG signals into modes in a Fourier spectrum. Linear and non-linear time domain features are extracted from the modes. Highly discriminant features are selected using the Kruskal–Wallis test. Different types of classification techniques are employed to classify the healthy and patients with schizophrenia. The effectiveness of the system is measured by evaluating various performance parameters such as accuracy, sensitivity, precision and specificity. Accuracy, precision, sensitivity and specificity of 88.7% 83.78%, 91.13% and 89.29%, respectively, are obtained. 1.1 Introduction Schizophrenia is a mental disorder which mostly occurs during adulthood, causing deficits such as interpersonal engagement and relationships, etc. About 1% of the global population is affected by schizophrenia. Patients with schizophrenia show symptoms such as disorganized speech, hallucinations or delusions, according to the doi:10.1088/978-0-7503-3279-8ch1 1-1 ª IOP Publishing Ltd 2020
- 28. American Psychiatry Association [1]. Schizophrenia treatment involves long-term medication and is a great burden on healthcare systems and families [2]. The early prediction of schizophrenia involves a large number of aspects [3]. The reliability and comparability of studies have increased dramatically due to the introduction of standardized tools for evaluating symptoms and diagnosis. However, the problems of selecting proper methods and evaluation tools, and repeatability, etc, remain. Electroencephalogram (EEG) signals have gained much attention in the diagnosis of schizophrenia due to their non-invasive nature and ease of use [4–7]. EEG signals are electrical measures of the brain activity of billions of neurons connected together to form a network. An EEG signal is acquired from the scalp and has played a key role in clinical diagnosis and the dynamics of brain research. EEG signals provide increased coherence that reflects the presence of anomalous cortical organization in schizophrenics rather than transient states or medication effects related to severe clinical disturbance [8]. The temporal, occipital, frontal and parietal portions of the scalp play significant roles in analysing the changes during schizophrenia [9]. To date, researchers have proposed various methods for the detection and diagnosis of schizophrenia using EEG signals. The detection of schizophrenia by clustering the EEG signals with the help of the k-means method has been proposed [10]. The psychopharmacological and physiological changes occurring in the EEGs of schizophrenic and healthy patients have been monitored [11]. The biological and clinical association of the alpha and gamma frequency bands and power has been studied to separate schizophrenic and normal patients [12–14]. The identification of schizophrenic and normal patients has been carried out using spectral analysis [15, 16]. A rhythm based risk rate evaluation of healthy and schizophrenic patients is used in [17]. The alpha, delta, beta and gamma rhythms of occipital, central and frontal sites have been classified using the support vector machine (SVM) [18]. The separation of rhythm based features using filtering methods, multilayer back- propagation and self-organizing maps has been used [19]. Rhythm based features using a band pass filtering method with SVM, Sammon map and deep neural network classification techniques have been utilized for the identification of schizophrenia [20–23] as has the evaluation and classification of frequency based features using linear discriminant analysis [24]. The detection of schizophrenic patients using matched filtering and the fast Fourier transform (FFT) is proposed in [25]. The separation of rhythms using the Grey Walter passive filter has also been used to identify schizophrenic patients. Positive and negative schizophrenia have been separated using the FFT of EEG signals of the frontal, temporal, parietal and occipital regions [26]. The brain activities of schizophrenic patients have been detected by evaluation of the spectral energy using the FFT [27]. The utility of the FFT along with principal component analysis, the Wilcoxon method and Welch’s averaged periodogram method has been demonstrated in identifying the changes in the delta, beta, alpha and gamma bands of schizophrenic patients [28–33]. The different spikes, namely the focal, paroxysmal and independent spikes, occurring in the EEGs of schizophrenic patients have been analysed [34]. Post-imperative negative variation and contingent negative variation analysis have been used for identifying schizophrenic patients [35]. The steady-state Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-2
- 29. visual evoked potential (SSVEP) and Fisher score have been used to evaluate different features classified using quadratic discriminant analysis (QDA), linear discriminant analysis (LDA), SVM with k-nearest neighbors (k-NN), second order polynomial kernels and logistic regression analysis (LRA) [36]. Features based on band power, autoregressive coefficients, Lampel–Ziv complexity (LZC), fractal dimensions (FDs) and entropies have been classified using SVM, adaptive boosting and LDA to detect schizophrenia [37–40]. The Welch periodogram technique for spectral estimation has been used to detect schizophrenia using the Kolmogorov– Smirnov test [41]. A genetic algorithm with a Butterworth filter and SVM has also been used to identify schizophrenia [42]. Ensemble synchronization measurement and Hilbert phase synchronization based methods have been used for classification using a logistic regression classifier [43]. The Hilbert–Huang transform, PCA, ICA and local discriminant bases have been used to extract the features of schizophrenic patients [44]. The utility of LZC for the identification of patients with schizophrenia is described in [45]. Power analysis of the alpha and delta bands has been carried out to distinguish control and affected patients [46]. The Higuchi, Katz and Petrosian methods have been used to extract classification features using LDA [47]. The weighted nearest neighbor, band power, FDs and autoregressive methods have been used to classify schizophrenic patients and control patients [48]. An auto-correlation and autoregressive coefficient have been classified using the independent t-test and neural networks [49–54]. LZC and correlation have been explored for measuring the alpha band activity of schizophrenic patients [55]. Multi-set canonical correlation analysis (MCCA) and SVM with recursive feature elimination (SVM-RFE) have been used for discrim- inating schizophrenia [56]. Entropy measurements and mean coherence with SVM have been used to discriminate schizophrenia [57]. Hurst exponent and FDs have been used to differentiate schizophrenic and control patients [22]. Kolmogorov complexity (KC), entropy and LZC methods have been used to find a useful discriminative tool for diagnostic purposes [58]. Feature vectors based on LZC and ANN have also been used to identify schizophrenic patients [59]. Autoregressive (AR), band power and FD coefficient based features extracted after preprocessing have been classified using LDA, multi-LDA (MLDA) and adaptive boosting (Adaboost) [60]. FDs and Pearson’s correlation coefficient have been used to apply the brief psychiatric rating scale (BPRS) for the detection of schizophrenia [61]. Power spectral density based features have been classified using a combination of factor analysis based on maximum likelihood theory [62, 63]. Spectral features extracted from combinatorial analysis have been classified using the Kora-N algorithm [64]. The weighted minimum distance to mean and Riemannian geometric mean have been used for the classification of schizophrenia [65]. The equivalent current dipole power and asymmetry coefficient have been used for the analysis of the positive symptoms of schizophrenia [66]. Factor analysis and Kaiser’s criteria have been used to identify patients suffering from schizophrenia [67]. Eigenvector power spectrum estimation and SVM have been used for the identification of schizophrenia [68]. Higuchi’s method of computation of FDs has also been used to detect schizophrenia [69]. Energy and power based features have Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-3
- 30. been classified using a high order pattern discovery algorithm [70]. The ε-complexity of continuous vector functions with RFs and an SVM have been used for binary classification of healthy and schizophrenic patients [71]. A fuzzy accuracy based classifier system has been used to generate fuzzy rules for discriminating healthy and schizophrenic subjects [72]. Autoregression based directed connectivity (DC) and graph-theoretical complex network (CN) based features have been classified using deep neural networks [73]. Inherent spatial pattern of network (SPN) features have been classified using LDA and an SVM to separate healthy patients from schizophrenic patients [74]. The analysis of entropy has been used for the evocation of emotions from visual cues in schizophrenia [75]. Mutual information (MI) has used to construct functional brain networks for analysis using graph theory [76]. Statistic significance probability maps based on the BPRS and a scale for the assessment of negative symptoms have been used for morphological findings in schizophrenia [77]. The spectral power of 192-channel resting EEG has been analysed using the Pearson correlation coefficient [78]. Spectral, complexity and variability measures evaluated from EEG signals have been classified using k-NN [79]. The long-term replicability of EEG spectra and auditory evoked potentials have been analysed to identify patients suffering from schizophrenia [80]. Sample covariance matrix and linear eigenvalue statistics have been used to classify schizophrenic patients using decision tree, random forest, SVM and naïve Bayes classifiers [81]. The Lyapunov exponent and Kolmogorov entropy have been evaluated to identify the classification accuracy of schizophrenic and controlled patients [82]. Average reference potential maps corresponding to global field power peaks in rhythms have been used to classify patients with schizophrenia [83]. Higuchi’s FD, entropy and Kolmogorov complexity based features have been classified using SVM [84]. Independent component analysis (ICA) and time–frequency representation using the Stockwell transform have been used to find the most significant rhythms in schizophrenic patients [85]. ICA, spectral analysis and analysis of variance (ANOVA) have been carried out on the frequency bands to identify control patients from schizophrenic patients [86]. Filtering, ICA and Fisher analysis have been used to classify patients using connectivity maps [87]. ICA for the spectral analysis of 200 bands and RFs have been used for accurately detecting schizophrenia based on one- minute EEG recordings [88]. Fourier statistical analysis, evaluative power spectra, averaged power spectra and spectral variance have been used to identify the traits of schizophrenia among patients [89]. Time–frequency distributions (TFDs), FFTs, eigenvector methods, the wavelet transform (WT) and AR method have been used for the extraction of features, with advantages and disadvantages [90]. Filtering, FFT, STFT and entropy based features have been used for the classification of schizophrenia using an SVM and multilayer perceptron (MLP) [91]. Short-time Fourier transforms with a sliding window have been used to distinguish schizo- phrenic patients [92]. Feature extraction based on wavelet filtering with a genetic algorithm and SVM has been used to identify control patients [93]. Classification based on PCA, wavelet transform and k-NNs is proposed in [94]. Time–frequency analysis has been used, with a Morlet wavelet having a Gaussian shape in time and Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-4
- 31. frequency, for the detection of schizophrenia [95]. Analysis of schizophrenic patients has been carried out using wavelet decomposition and Welch power spectral density (PSD) methods [96]. Analysis of alpha band frequencies has been carried out to detect the activity in schizophrenic patients [97]. Discriminant analysis (DA) has also been employed for classifying schizophrenic patients [98]. Features evaluated using Kolmogorov entropy, permutation entropy, correlation dimensions and spectral entropy have been selected using the Fisher criterion and classified using k-NN, SVM and back-propagation neural networks [99]. The phase lock value and phase coherence value of the intrinsic mode functions of empirical mode decom- position have been used to differentiate schizophrenia [100]. Multi-domain convolu- tional neural networks have been used for the classification of EEG based brain connectivity networks in schizophrenia [101]. Various methods for the detection of schizophrenia have been proposed in the literature. The FFT suffers from time–frequency localization. Other rigid methods such as STFT, wavelet transform and filtering use a basis which is independent of the processed signal. Moreover, the majority of these methods involve the direct evaluation of features from the raw EEG signals. The empirical wavelet transform (EWT) is capable of building an adaptive wavelet to extract the AM–FM components of a signal. The adaptive selection of the wavelet can capture useful hidden information from non-stationary EEG signals. In this chapter, a wavelet based decomposition method is employed to decompose the signal into AM–FM components. Dominant time domain features evaluated from the AM–FM compo- nents are selected using the Kruskal–Wallis test. The selected features are given as the input for the classifiers to distinguish patients with schizophrenia from control patients. The performance of the system is tested by evaluating four performance parameters and the receiver operating characteristics curve. The remainder of the chapter is organized as follows: section 1.2 presents the methodology, the results and discussion are provided in section 1.3, and section 1.4 concludes the chapter. 1.2 Methodology This section includes descriptions of the dataset, empirical wavelet transform, features and classification techniques. The EEG signals are decomposed into AM–FM components using the EWT. Multiple time domain features are extracted from the obtained AM–FM components. Highly discriminant features are selected using the Kruskal–Wallis test and are classified using different classification techniques. The flowchart of the proposed methodology is shown in figure 1.1. 1.2.1 Dataset The dataset used in this chapter contains EEG recordings of 14 female and 67 male patients. The average age and years of education are 39 years and 14.5 years, respectively. The details of the dataset are available online [102]. Three press button tasks were performed by the subjects, namely (1) pressing a button to immediately generate a tone, (2) passively listening to the same tone and (3) pressing a button without generating a tone to study the corollary discharge in people with Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-5
- 32. schizophrenia and comparison controls. The healthy controls generated a press button tone while the schizophrenia patients did not. Hence, only condition one is tested to classify healthy and schizophrenia patients. The data are acquired from 64 sites on the scalp. The EEG data are sampled at a rate of 1024 Hz. EEG recordings of control and schizophrenic patients are shown in figure 1.2. 1.2.2 Empirical wavelet transform To extract information from the highly complex EEG signal, the signal is split into multiple components. The empirical wavelet transform (EWT) is one such adaptive mechanism to split the signal into multiple components. The EWT is capable of extracting some components from the signal by building adaptive wavelets. Each component obtained by the EWT has a compact support Fourier spectrum. The Figure 1.1. Flowchart of the proposed methodology. Figure 1.2. EEG signals of a healthy control and a schizophrenia patient. Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-6
- 33. separation of these modes is similar to Fourier spectrum segmentation and filtering. The EWT is defined in the same manner as the traditional wavelet transform. The EWT has detailed and approximation coefficients [103]. The detailed coefficients ( α W n t ( , ) f ) are defined as the inner products with the empirical wavelet given as ∫ ϕ τ ϕ τ τ ω ϕ ω = ⟨ ⟩ = − = ˆ ˆ α ( ) W n t f f t f ( , ) , ( ) ( )d ( ) ( ) . (1.1) f n n n v The approximation coefficients ( α W t (0, ) f ), defined as the inner product with a scaling function, can be written as ∫ φ τ φ τ τ ω φ ω = ⟨ ⟩ = − = ˆ ˆ α ν W t f f t f (0, ) , ( ) ( )d ( ( ) ( )) , (1.2) f 1 1 1 where f is the input signal in the time domain, and φ and ϕ are the wavelet and scaling functions, respectively. The reconstruction of the signal can be denoted as ∑ ∑ φ ω ϕ ω φ ω ω ϕ ω = × + × = ˆ × ˆ + ˆ × ˆ = = α α α α f t W t t W t t W W n ( ) (0, ) ( ) ( , ) ( ) (0, ) ( ) ( , ) ( ) . (1.3) n N n N 1 1 f f n f f n 1 1 ⎛ ⎝ ⎜ ⎜ ⎞ ⎠ ⎟ ⎟ The AM–FM components obtained from the signal are shown in figure 1.3. 1.2.3 Feature extraction Features are the statistical measures evaluated from the AM–FM components of signals. These statistical measures play an important role in the dimensionality Figure 1.3. Modes obtained from EWT. Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-7
- 34. reduction and classification of signals. In this chapter, various statistical measures have been evaluated. Based on the Kruskal–Wallis analysis, five statistical measures are selected as features, which are kurtosis, covariance, root mean square, minimum and mean. 1.2.3.1 Kurtosis Kurtosis measures the thickness along the tail of a given distribution for a given random variable. Kurtosis can be mathematically expressed as ∑ σ = − ¯ = f f N Kurtosis ( ) / , (1.4) n N 1 n 4 4 where N is the number of signals, ¯ f is the mean and σ is the standard deviation. 1.2.3.2 Variance Variance measures the spread of numbers from its mean value. It is the expectation of a squared deviation from the mean. The variance can be expressed as ∑ = − ¯ = N f f Variance 1 ( ) . (1.5) n N 1 n 2 1.2.3.3 Root mean square The root mean square (RMS) is the quadratic mean of the variable that measures the magnitude of varying quantity: ∑ = = N f RMS 1 ( ) . (1.6) n N 1 n 2 1.2.3.4 Mean The mean is the average value of all the samples in the variable and is expressed as ∑ = = N f Mean 1 . (1.7) n N 1 n 1.2.3.5 Minimum The minimum value of the variable is expressed as = = f Minimum min( ). (1.8) n N 1 n Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-8
- 35. 1.2.4 Classification techniques The purpose of classifiers is to classify input data into two or more classes. The feature matrix is given as an input for classifiers. In this chapter, five different classification techniques are employed to classify binary class data. The k-NN, DA, ensemble method, SVM and decision tree based classifiers are used. To classify the input signals different kernels are employed. In the case of k-NN, six kernels are used, namely the fine, medium, coarse, cosine, cubic and weighted kernels. Linear and quadratic kernels are employed with discriminant analysis. Four kernels, namely the bagged tree, boosted tree, subspace k-NN (SS-k-NN) and subspace discriminant (SS-D) kernels are used with an ensemble based classifier. Linear, medium Gaussian and coarse Gaussian kernels are used with the SVM. For the decision tree based classifiers, simple tree, complex tree and medium tree are used. The details of the classification methods can be found in [104–108]. The process of k- NN is denoted as ∑ = − + − + … + − = − = − − = d y y y y y y y y y y V V A A A ( , ) ( ) ( ) ( ) ( ) min max min , (1.9) i t i t i t ip tp i n i i 1 1 2 2 2 2 2 1 1 2 2 1 where d is the distance, yi is an input with p features, n is the total inputs and p is the total number of features. V1 is the max–min normalization matrix. In this chapter the total number of neighbors is selected as 5. The mathematical modeling of the SVM is formed by minimizing the objective function K(w), by taking the constraint + ⩾ = … z w y b i N ( ) 1( 1, 2, , ) i i T : = ∣∣ ∣∣ K w w ( ) min 1 2 . (1.10) 2 ⎛ ⎝ ⎜ ⎞ ⎠ ⎟ By augmenting the objective function, the Lagrangian function for the SVM thus formed is denoted by ∑ λ λ Ψ = − + − = w b w w z w y b ( , , ) 1 2 [ ( ) 1], (1.11) i N 1 t i i i T where, K(w) is the kernel, w is the weight matrix, b is the bias and y is the input. The ensemble method for classification is mathematically represented by ∑ ˆ • = • = G c G ( ) ( ), (1.12) i N 1 i i ens where ˆ • G ( ) ens is the ensemble based function estimator, • G ( ) i is the reweighted original data and ci is the averaging weights. Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-9
- 36. The discriminant analysis classifier can be represented as ∑ μ μ = − − = ∈ − S y y W S S ( )( ) , (1.13) y w j j W i i T i wi 1 B i where SB and Swi are the variance between the classes and the variance within the class, respectively. 1.2.5 Performance parameters The performance of the system is tested by evaluating four performance parameters. In this chapter, accuracy, sensitivity, specificity and precision are measured. In the following, true positive (TP) is the number of true positives correctly identified from the positive class, true negative (TN) is the number of true negatives correctly identified from the negative class, false positive (FP) is the number of data points classified into the positive class that actually belong to the negative class and false negative (FN) is the number of data points classified into the negative class that belong to the positive class. ACC, SEN, SPE and PRE denote the accuracy, sensitivity, specificity and precision, respectively. Accuracy is defined as the ratio of the total number of correctly identified instances to the total number of instances. The mathematical formulation of accuracy is given as = + + + + ACC TP TN TP FP TN FN . (1.14) The sensitivity or probability of detection is defined as the ability to correctly identify positive results. Sensitivity is represented by = + SEN TP TP FN . (1.15) The specificity or true negative rate is the ability to correctly identify actual negatives. The specificity is denoted by = + SPE TN TN FP . (1.16) The precision is the ratio of the total number of true positives to the total number of true positives and false positives. The precision is represented by = + PRE TP TP FP . (1.17) 1.3 Results and discussion This methodology uses the empirical wavelet transform and different classification techniques to separate schizophrenic patients from normal patients. There are 4108 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-10
- 37. signals available for schizophrenia patients and 3608 control signals. Every signal of each class has a total of 3072 samples. N-100 channels play an important role [102] in the detection of schizophrenia. Based on this, the first ten N-100 channels are considered for the evaluation. To maintain uniformity, a common experimental platform is used for both classes. Each signal is given as an input to the EWT. The boundary conditions are kept the same for both classes. The boundaries are chosen as {4 8 12 30}. Different numbers of AM–FM components are obtained from the signals using EWT. The minimum number of AM–FM components is eight and the maximum is 28. However, to maintain the synchronism between all signals for further computation, the number of AM–FM components is considered to be eight. Various statistical parameters are evaluated from the AM–FM components. Highly discriminable features are selected based on the results of the Kruskal–Wallis (KW) test. The KW test is a non-parameterized analysis of variance. It is used to find the discrimination ability of features by evaluating the probability of χ. A probability value ⩽0.05 is considered to be significant for classification. A total of five features are selected based on the KW test. These are kurtosis, variance, root mean square, minima and maxima, respectively. The probabilistic values of all the features are shown in tables 1.1–1.5, respectively. It is evident from these tables that most of the AM–FM components and channels are highly discriminable. Inspired by the obtained results presented in tables 1.1–1.5, the selected features are given as the input for different classifiers. All the channels of every feature of each AM–FM component are combined. For every AM–FM component, the feature matrices obtained for schizophrenia and normal patients are 4108 × 50 and 3608 × 50, respectively. In this methodology, the ten-fold cross-validation method is employed for classification. Here, the input data are partitioned randomly into ten disjoint sets. Nine sets are used for training the input data and the remaining set is utilized for testing. The patients with schizophrenia are separated from the normal patients using five types of classification techniques. Table 1.6 shows the classification accuracy obtained by the k-NN classifier. Six kernels are used for the classification. The classification accuracies obtained with the fine, medium, coarse, cosine, cubic and weighted kernels are, respectively, 81%, 84.1%, 82.1%, 84%, 83.3% and 84.3% for M-1, 72.1%, 75.6%, 71.5%, 75.2%, 72.4% and 75.9% for M-2, 66%, 70.2%, 68.2%, 69.7%, 67.5% and 71.1% for M-3, 66%, 69.5%, 67.6%, 68.3%, 66.2% and 70% for M-4, 65%, 69.6%, 70%, 68.6%, 65.7% and 70.7% for M-5, 61.3%, 65.4%, 67.9%, 65%, 62.8% and 66.4% for M-6, 59.3%, 61.8%, 64.8%, 61.5%, 61.2% and 63.8% for M-7, and 60.4%, 63.4%, 65.6%, 63.9%, 62.4% and 64.6% for M-8. The maximum accuracies obtained with the fine, medium, coarse, cosine, cubic and weighted kernel are, respectively, 81%, 84.1%, 82.1%, 84%, 83.3% and 84.3% for M-1. Table 1.7 shows the accuracy of four classifiers, namely the discriminant analysis, SVM, ensemble and decision tree classifiers. The classification accuracies with the linear kernel are 74.9%, 59.3%, 56.9%, 56.4%, 55.8%, 55.7%, 56.3% and 57.1% for, respectively, M-1, M-2, M-3, M-4, M-5, M-6, M-7 and M-8, and with QDA the accuracies are 59%, 60.4%, 58.6%, 58.2%, 58.2%, 58.3%, 58.6% and 58.8%, respectively. The maximum accuracies for LDA and QDA are 74.9% and 60.4% Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-11
- 43. for M-1 and M-2, respectively. The accuracies obtained with the ensemble classifier for M-1, M-2, M-3, M-4, M-5, M-6, M-7 and M-8, respectively, are 87.5%, 87.7%, 83.4%, 82.2%, 83%, 76.5%, 66.8% and 68.2% for bagged tree (Bag-T), 85.6%, 85.5%, 80.8%, 80.3%, 81.4%, 74.9%, 65.1% and 67.4% for boosted tree (BT), 72.2%, 69.9%, 66.1%, 65.7%, 63.8%, 63.7%, 63.4% and 63.1% for SS-k-NN and 81.4%, 62%, 60.8%, 61%, 60.9%, 61%, 61.2% and 61.2% for SS-D. The maximum accuracies obtained with Bag-T, BT, SS-k-NN and SS-D are 87.7%, 85.6%, 72.2% and 81.4% for M-1 and M-2. Linear, medium and coarse kernels of SVM are used to test the accuracy. For M-1, M-2, M-3, M-4, M-5, M-6, M-7 and M-8, respectively, accuracies are achieved of 86.4%, 69%, 63.4%, 61.6%, 61.2%, 61.1%, 61.4% and 63.9% for the linear kernel, 88.7%, 88.2%, 86.1%, 82.9%, 82.3, 69.5%, 61.6% and 67.5% for the medium kernel, and 83.4%, 62.8%, 61.5%, 61.3%, 61.3%, 61.4%, 61.4% and 61.4% for the coarse kernel. M-1 provides maximum accuracies of 86.4%, 88.7% and 83.4% for the linear, medium and coarse kernels, respectively. For M-1, Table 1.6. Classification accuracy of k-NN. k-nearest neighbors SB Fine Medium Coarse Cosine Cubic Weighted M-1 81 84.1 82.1 84 83.3 84.3 M-2 72.1 75.6 71.5 75.2 72.4 75.9 M-3 66 70.2 68.2 69.7 67.5 71.1 M-4 66 69.5 67.6 68.3 66.2 70 M-5 65 69.6 70 68.6 65.7 70.7 M-6 61.3 65.4 67.9 65 62.8 66.4 M-7 59.3 61.8 64.8 61.5 61.2 63.8 M-8 60.4 63.4 65.6 63.9 62.4 64.6 Table 1.7. The classification accuracy of the DA, ensemble, SVM and decision tree classifiers. Discriminant analysis Ensemble Support vector machine Decision tree classifier SB L Q Bag-T BT SS-k-NN SS-D Linear Medium Coarse CT ST MT M-1 74.9 59 87.5 85.6 72.2 81.4 86.4 88.7 83.4 78.9 75.6 68.7 M-2 59.3 60.4 87.7 85.5 69.9 62 69 88.2 62.8 80.1 78.8 76.7 M-3 56.9 58.6 83.4 80.8 66.1 60.8 63.4 86.1 61.5 76.3 73.9 72.2 M-4 56.4 58.2 82.2 80.3 65.7 61 61.6 82.9 61.3 73.5 70.9 69.6 M-5 55.8 58.2 83 81.4 63.8 60.9 61.2 82.3 61.3 74.2 71.6 68.5 M-6 55.7 58.3 76.5 74.9 63.7 61 61.1 69.5 61.4 68.1 66.7 62.5 M-7 56.3 58.6 66.8 65.1 63.4 61.2 61.4 61.6 61.4 61.4 62.6 60.8 M-8 57.1 58.8 68.2 67.4 63.1 61.2 63.9 67.5 61.4 62.6 62.5 60.7 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-17
- 44. M-2, M-3, M-4, M-5, M-6, M-7 and M-8, respectively, the accuracies are 78.9%, 80.1%, 76.3%, 73.5%, 74.2%, 68.1%, 61.4% and 62.6% for the complex tree (CT), 75.6%, 78.8%, 73.9%, 70.9%, 71.6%, 66.7%, 62.6% and 62.5% for the simple tree (ST), and 68.7%, 76.7%, 72.2%, 69.6%, 68.5%, 62.5%, 60.8% and 60.7% for the medium tree (MT). The decision tree type classifier provides the maximum accuracy for M-2 for CT, ST and MT with values of 80.1%, 78.8% and 76.7%, respectively. Among all the classifiers, the SVM produces the highest accuracy with the medium kernel. Hence, the performance parameters of the medium kernel are shown in table 1.8. Four performance parameters are evaluated, namely accuracy (ACC), sensitivity (SEN), specificity (SPE) and precision (PRE). For M-1, M-2, M-3, M-4, M-5, M-6, M-7 and M-8, respectively, we find a sensitivity of 91.3%, 86.51%, 85.87%, 83.44%, 83.73%, 76.44%, 72.19% and 63.49%, a specificity 7.37%, 89.29%, 86.24%, 82.54%, 8165%, 67.89%, 61.10% and 69.16%, and a precision of 79.7%, 83.78%, 78.41%, 71.65%, 69.72%, 34.97%, 7.43% and 45.59%. The highest accuracy and sensitivity are obtained as 88.70% and 91.13%, respectively, for M-1. Maximum specificity and precision are obtained as 89.29% and 83.78%, respectively, for M-2. The receiver operating characteristic (ROC) curve shows the performance of a classification model for all classification thresholds. The ROC curve of all the AM– FM components for a medium Gaussian SVM is shown in figure 1.4. As evident from figures 1.4(a) and (b), the area under curve (AUC) is 94%. The change in classifier characteristics is identified at a true positive rate (TPR) of 80% and a false positive rate (FPR) of 5% for M-1. The change in classifier characteristics is identified at 84% TPR and 9% FPR for M-2. Figures 1.4(c) and (d) represent the ROC curves of M-3 and M-4. The AUC for M-3 is 93% while for M-4 it is 91%. The change in classifier characteristics for M-3 and M-4 are obtained at FPRs of 9% and 10% and TPRs of 78% and 72%, respectively. The ROCs of M-5, M-6, M-7 and M-8 are shown in figures 1.4(e), (f), (g) and (h). The AUCs are 91%, 84%, 69% and 72% for M-5, M-6, M-7 and M-8, respectively. The change is observed at TPRs of 70% and 35% and FPRs of 9% and 7% for M-5 and M-6, respectively, while for M-7 and M-8 it is observed at TPRs of 2% and 18% and FPRs of 7% and 46%. Table 1.8. The performance parameters of the medium kernel SVM. Performance parameters SB CC SEN SPE PRE M-1 88.70 91.13 87.37 79.70 M-2 88.20 86.51 89.29 83.78 M-3 86.10 85.87 86.24 78.41 M-4 82.90 83.44 82.54 71.65 M-5 82.30 83.73 81.65 69.72 M-6 69.50 76.44 67.89 34.97 M-7 61.60 72.19 61.10 7.43 M-8 67.50 63.49 69.16 45.59 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-18
- 45. Figure 1.4. Receiver operator characteristics for a medium SVM of the first eight subbands M-1–M-8, (a) to (h), respectively. Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-19
- 46. 1.4 Conclusion Patients with schizophrenia cannot be easily identified through visual inspection. Doctors recommend a number of neurological tests to identify the symptoms of schizophrenia. However, these tests are not always effective. Electroencephalogram signals provide vital information about the neurological changes that happen in the schizophrenic state. In this chapter, a novel method based on the empirical wavelet transform is proposed for the identification of schizophrenia. The majority of the information resides in the first two AM–FM components, as these provide the highest correct classifications of schizophrenic patients and normal patients. The classification abilities of different classification techniques are tested. It is found that SVM is the best method, followed by the ensemble, k-nearest neighbors, decision tree and, finally, discriminant analysis classifier. The medium kernel of the SVM provides the best performance parameters with an accuracy of 88.7%, a sensitivity of 91.13%, a specificity of 89.29% and a precision of 83.78%. References [1] Samuel B G 1995 Diagnostic and statistical manual of mental disorders 4th edn (DSM-IV) Am. J. Psychiatry 152 1228 [2] McGlashan T H 1998 Early detection and intervention of schizophrenia: rationale and research Br. J. Psychiatry 172 3–6 [3] Lett T A, Voineskos A N, Kennedy J L, Levine B and Daskalakis Z J 2014 Treating working memory deficits in schizophrenia: a review of the neurobiology Biol. Psychiatry 75 361–70 [4] Hee N S, Jin S H, Kim S Y and Ham B J 2002 EEG in schizophrenic patients: mutual information analysis Clin. Neurophysiol. 113 1954–60 [5] Knox H F and Macfie C C 1941 Electroencephalography in schizophrenia Am. J. Psychiatry 98 374–81 [6] Koukkou M, Lehmann D, Wackermann J, Dvorak I and Henggeler B 1993 Dimensional complexity of EEG brain mechanisms in untreated schizophrenia Biol. Psychiatry 33 397– 407 [7] Karson C N, Coppola R, Daniel D G and Weinberger D R 1988 Computerized EEG in schizophrenia Schizophr. Bull. 14 193–7 [8] Merrin E L, Floyd T C and Fein G 1989 EEG coherence in unmedicated schizophrenic patients Biol. Psychiatry 25 60–6 [9] Davis P A 1940 Evaluation of the electroencephalograms of schizophrenic patients Am. J. Psychiatry 96 851–60 [10] Lehmann D, Faber P L, Galderisi S, Herrmann W M and Koenig T 2005 EEG microstate duration and syntax in acute, medication-naïve, first-episode schizophrenia: a multi-center study Psychiatry Res.: Neuroimaging 138 141–56 [11] Kornetsky C and Mirsky A F 1965 On certain psychopharmacological and physiological differences between schizophrenic and normal persons Psychopharmacologia 8 309–18 [12] Sponheim S R, Clementz B A, Iacono W G and Beiser M 2000 Clinical and biological concomitants of resting state EEG power abnormalities in schizophrenia Biol. Psychiatry 48 1088–97 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-20
- 47. [13] Roach B J and Mathalon D H 2008 Event-related EEG time-frequency analysis: an overview of measures and an analysis of early gamma band phase locking in schizophrenia Schizophr. Bull. 34 907–26 [14] Moran L and Hong L 2011 High vs low frequency neural oscillations in schizophrenia Schizophr. Bull. 37 659–63 [15] Giannitrapani D and Kayton L 1974 Schizophrenia and EEG spectral analysis Electroencephalogr. Clin. Neurophysiol. 36 377–86 [16] Boutros N N, Arfken C, Galderisi S, Warrick J, Pratt G and Iacono W 2008 The status of spectral EEG abnormality as a diagnostic test for schizophrenia Schizophr. Res. 99 225–37 [17] Clementez B A, Sponheim S R, Iacono W G and Beiser M 1994 Resting EEG in first‐ episode schizophrenia patients, bipolar psychosis patients, and their first‐degree relatives Psychophysiology 31 486–94 [18] Johannesen J, Bi J, Jiang R, Kenney J and Chen C M 2016 Machine learning identification of EEG features predicting working memory performance in schizophrenia and healthy adults Neuropsychiatr. Electrophysiol. 2 3 [19] Li Y and Fan F 2005 Classification of schizophrenia and depression by EEG with ANNs 2005 IEEE Engineering in Medicine and Biology 27th Annual Conf. pp 2679–82 [20] Shim M, Hwang H, Kim D, Lee S and Im C 2016 Machine-learning-based diagnosis of schizophrenia using combined sensor-level and source-level EEG features Schizophr. Res. 176 314–9 [21] Winterer G, Ziller M, Dor H, Frick K, Mulert C, Wuebben Y and Herrmann W M 2000 Frontal dysfunction in schizophrenia—a new electrophysiological classifier for research and clinical applications Eur. Arch. Psychiatry Clin. Neurosci. 250 207–14 [22] Chu L, Qiu R, Liu H, Ling Z and Shi X 2017 Individual recognition in schizophrenia using deep learning methods with random forest and voting classifiers: insights from resting state EEG streams, arXiv:1707.03467 [23] Bose T, Sivakumar S and Kesavamurthy B 2016 Identification of schizophrenia using EEG alpha band power during hyperventilation and post-hyperventilation J. Med. Biol. Eng. 36 901–11 [24] Ince N F, Pellizzer G, Tewfik A H, Nelson K, Leuthold A, McClannahan K and Stephane M 2009 Classification of schizophrenia with spectro-temporo-spatial MEG patterns in working memory Clin. Neurophysiol. 120 1123–34 [25] Flor H P, Lind J C and Koles Z J 2004 A source-imaging (low-resolution electromagnetic tomography) study of the EEGs from unmedicated males with depression Psychiatry Res.: Neuroimaging 130 191–207 [26] Begić D, Hotujac L and Jokić‐Begić N 2000 Quantitative EEG in ‘positive’ and ‘negative’ schizophrenia Acta Psychiatr. Scand. 101 307–11 [27] Morstyn R, Duffy F H and McCarley R W 1983 Altered topography of EEG spectral content in schizophrenia Electroencephalogr. Clin. Neurophysiol. 56 263–71 [28] Michelogiannis S, Paritsis N and Trikas P 1991 EEG coherence during hemispheric activation in schizophrenics Eur. Arch. Psychiatry Clin. Neurosci. 241 31–4 [29] Dierks T K, Maurer R and Ihl A 1989 Schmidtke Topographic Brain Mapping of EEG and Evoked Potentials (Berlin: Springer) pp 507–17 [30] Michel C M, Koukkou M and Lehmann D 1993 EEG reactivity in high and low symptomatic schizophrenics, using source modelling in the frequency domain Brain Topogr. 5 389–94 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-21
- 48. [31] Umesh S, Nizamie S H, Goyal N, Tikka S and Bose S 2016 Social anhedonia and gamma band abnormalities as a composite/multivariate endophenotype for schizophrenia: a dense array EEG study Early Interv. Psychiatry 12 362–71 [32] Alfimova M and Uvarova L 2003 Cognitive peculiarities in relatives of schizophrenic and schizoaffective patients: heritability and resting EEG-correlates Int. J. Psychophysiol. 49 201–16 [33] Lacroix D, Chaput Y, Rodriguez J P, Filion M, Morrison D, St. Denis P and Albert J-M 1995 Quantified EEG changes associated with a positive clinical response to clozapine in schizophrenia Progr. Neuro-Psychopharmacol. Biol. Psychiatry 19 861–76 [34] White P T, Demyer W and Demyer M K 1964 EEG abnormalities in early childhood schizophrenia: a double-blind study of psychiatrically disturbed and normal children during promazine sedation Am. J. Psychiatry 120 950–8 [35] Verleger R, Wascher E, Arolt V, Daase C, Strohm A and Kömpf D 1999 Slow EEG potentials (contingent negative variation and post-imperative negative variation) in schizophrenia: their association to the present state and to Parkinsonian medication effects Clin. Neurophysiol. 110 1175–92 [36] Alimardani F, Cho J H, Boostani R and Hwang H J 2018 Classification of bipolar disorder and schizophrenia using steady-state visual evoked potential based features IEEE Access 6 40379–88 [37] Sabeti M, Katebi S and Boostani R 2009 Entropy and complexity measures for EEG signal classification of schizophrenic and control participants Artif. Intell. Med. 47 263–74 [38] Sabeti M, Boostani R, Katebi S D and Price G W 2007 Selection of relevant features for EEG signals classification of schizophrenic patients Biomed. Signal Process. Control 2 122–34 [39] Sabeti M, Katebi S D, Boostani R and Price G W 2011 A new approach for EEG signal classification of schizophrenic and control participants Expert Syst. Appl. 38 2063–71 [40] Taghavi M, Boostani R, Sabeti M and Taghavi S M 2011 Usefulness of approximate entropy in the diagnosis of schizophrenia Iran. J. Psychiatry Behav. Sci. 5 [41] Akbar Y, Khotimah S N and Haryanto F 2016 Spectral and brain mapping analysis of EEG based on Pwelch in schizophrenic patients J. Phys. Conf. Ser. 694 012070 [42] Sabeti M, Boostani R and Katebi S D 2007 A new approach to classify the schizophrenic and normal subjects by finding the best channels and frequency bands 2007 15th Int. Conf. on Digital Signal Processing [43] Prasad P D, Halahalli H N, John J P and Majumdar K K 2014 Single-trial EEG classification using logistic regression based on ensemble synchronization IEEE J. Biomed. Health Inform. 18 1074–80 [44] Azlan W A and Low Y F 2014 Feature extraction of electroencephalogram (EEG) signal— a review 2014 IEEE Conf. on Biomedical Engineering and Sciences (IECBES) [45] Li Y, Tong S, Liu D, Gai Y, Wang X, Wang J, Qiu Y and Zhu Y 2008 Abnormal EEG complexity in patients with schizophrenia and depression Clin. Neurophysiol. 119 1232–41 [46] Fenton G W, Fenwick P B, Dollimore J, Dunn T L and Hirsch S R 1980 EEG spectral analysis in schizophrenia Br. J. Psychiatry 136 445–55 [47] Boostani R, Khadijeh S and Sabeti M 2009 An efficient classifier to diagnose of schizophrenia based on the EEG signals Expert Syst. Appl. 36 6492–9 [48] Parvinnia E, Sabeti M, Zolghadri M J and Boostani R 2014 Classification of EEG signals using adaptive weighted distance nearest neighbor algorithm J. King Saud Univ. Comput. Inf. Sci. 26 1–6 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-22
- 49. [49] Abrams R and Taylor M A 1979 Differential EEG patterns in affective disorder and schizophrenia Arch. Gen. Psychiatry 36 1355–8 [50] Saletu B, Anderer P, Saletu Z G M, Arnold O and Pascual-Marqui R D 2002 Classification and evaluation of the pharmacodynamics of psychotropic drugs by single-lead pharmaco- EEG, EEG mapping and tomography (LORETA) Methods Find. Exp. Clin. Pharmacol. 24 97–120 [51] Streit M, Wölwer W, Brinkmeyer J, Ihl R and Gaebel W 2001 EEG-correlates of facial affect recognition and categorisation of blurred faces in schizophrenic patients and healthy volunteers Schizophr. Res. 49 145–55 [52] Jeong J H, Kim D J, Chae J H, Kim S Y, Ko H J and Paik I H 1998 Nonlinear analysis of the EEG of schizophrenics with optimal embedding dimension Med. Eng. Phys. 20 669–76 [53] Bob P 2012 Consciousness, schizophrenia and complexity Cogn. Syst. Res. 13 87–94 [54] Vasios C E, Papageorgiou C, Matsopoulos G K, Nikita K S and Uzunoglu N 2003 A decision support system of evoked potentials for the classification of patients with first- episode schizophrenia Ger. J. Psychiatry 5 05 [55] Cerquera A, Gjini K, Bowyer S M and Boutros N 2017 Comparing EEG nonlinearity in deficit and nondeficit schizophrenia patients: preliminary data Clin. EEG Neurosci. 48 376–82 [56] Sui J, Castro E, He H, Bridwell D A, Du Y, Pearlson G D, Jiang T and Calhoun V D 2014 Combination of FMRI-SMRI-EEG data improves discrimination of schizophrenia patients by ensemble feature selection 2014 36th Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society pp 3889–92 [57] Aydın S, Arica N, Ergul E and Tan O 2015 Classification of obsessive compulsive disorder by EEG complexity and hemispheric dependency measurements Int. J. Neural Syst. 25 1550010 [58] Akar S A, Kara S, Latifoğlu F and Bilgiç V 2016 Analysis of the complexity measures in the EEG of schizophrenia patients Int. J. Neural Syst. 26 1650008 [59] Fan F-Y, Li Y-J, Qiu Y-H and Zhu Y-S 2005 Use of ANN and complexity measures in cognitive EEG discrimination 2005 IEEE Engineering in Medicine and Biology 27th Annu. Conf. [60] Abhishekh H A, Thirthalli J, Manjegowda A, Phutane V H, Muralidharan K and Gangadhar B N 2013 Ictal EEG fractal dimension in ECT predicts outcome at 2 weeks in schizophrenia Psychiatry Res. 209 155–9 [61] Abhishekh H A, Thirthalli J, Manjegowda A, Phutane V H, Muralidharan K and Gangadhar B N 2013 Ictal EEG fractal dimension in ECT predicts outcome at 2 weeks in schizophrenia Psychiatry Res. 209 155–9 [62] Khodayari-Rostamabad A, Reilly J P, Hasey G, deBruin H and MacCrimmon D 2010 Diagnosis of psychiatric disorders using EEG data and employing a statistical decision model 2010 Annual Int. Conf. of the IEEE Engineering in Medicine and Biology pp 4006–9 [63] Julieta R, Luis F, Cerdán M A, Guevara C A and Araceli S 2001 Abnormal EEG patterns in treatment-resistant schizophrenic patients Int. J. Neurosci. 109 47–59 [64] Kaplan A Y, Borisov S V and Zheligovskiy V A 2005 Classification of the adolescent EEG by the spectral and segmental characteristics for normals Zh. Vyssh. Nerv. Deiat. Im. I P Pavlova 55 478–86 [65] AliMardani F, Boostani R and Blankertz B 2016 Presenting a spatial-geometric EEG feature to classify BMD and schizophrenic patients Int. J. Adv. Telecommun. Electrotech. Signals Syst. 5 79–85 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-23
- 50. [66] Youn T, Park H-J, Kim J-J, Kim M S and Kwon J S 2003 Altered hemispheric asymmetry and positive symptoms in schizophrenia: equivalent current dipole of auditory mismatch negativity Schizophr. Res. 59 253–60 [67] Shagass C, Roemer R A, Straumanis J J and Josiassen R C 1984 Psychiatric diagnostic discriminations with combinations of quantitative EEG variables Br. J. Psychiatry 144 581–92 [68] Lou E, Zhang S and Qiao S 2009 Application of eigenvector estimation and SVM for EEG signals classification 2009 9th Int. Conf. on Electronic Measurement & Instruments [69] Raghavendra B S, Dutt D N, Halahalli H N and John J P 2009 Complexity analysis of EEG in patients with schizophrenia using fractal dimension Physiol. Meas. 30 795–808 [70] Zhang S, Qiao S and Wang W 2010 Classification of schizophrenia’s EEG based on high order pattern discovery 2010 IEEE Fifth Int. Conf. on Bio-Inspired Computing: Theories and Applications (BIC-TA) [71] Piryatinska A, Darkhovsky B and Kaplan A 2017 Binary classification of multichannel- EEG records based on the ϵ-complexity of continuous vector functions Comput. Methods Programs Biomed. 152 131–9 [72] Sabeti M, Mohammad H S and Pric G W 2007 Fuzzy accuracy-based classifier systems for EEG classification of schizophrenic patients First Joint Congress on Fuzzy and Intelligent Systems [73] Phang C, Ting S, Samdin B and Ombao H 2019 Classification of EEG-based effective brain connectivity in schizophrenia using deep neural networks 9th Int. IEEE/EMBS Conf. on Neural Engineering (NER) [74] Li F et al 2019 Differentiation of schizophrenia by combining the spatial EEG brain network patterns of rest and task P300 IEEE Trans. Neural Syst. Rehabil. Eng. 27 594–602 [75] Chu W, Huang M, Jian B and Cheng K 2017 Analysis of EEG entropy during visual evocation of emotion in schizophrenia Ann. Gener. Psychiatry 16 34 [76] Yin Z, Li J, Zhang Y, Ren A, Von Meneen K M and Huang L 2017 Functional brain network analysis of schizophrenic patients with positive and negative syndrome based on mutual information of EEG time series Biomed. Signal Process. Control 31 331–8 [77] Kirino E and Inoue R 1999 The relationship of mismatch negativity to quantitative EEG and morphological findings in schizophrenia J. Psychiatr. Res. 33 445–56 [78] Tikka S K, Yadav S, Nizamie S H, Das B, Tikka D L and Goyal N 2014 Schneiderian first rank symptoms and gamma oscillatory activity in neuroleptic naïve first episode schizo- phrenia: a 192 channel EEG study Psychiatry Invest. 11 465–75 [79] Sabeti M, Behroozi R and Moradi E 2016 Analysing complexity, variability and spectral measures of schizophrenic EEG signal Int. J. Biomed. Eng. Technol. 21 109 [80] Lifshitz K, Lee K L and Susswein S 1987 Long-term replicability of EEG spectra and auditory evoked potentials in schizophrenic and normal subjects Neuropsychobiology 18 205–11 [81] Liu H, Zhang T, Ye Y, Pan C, Yang G, Wang J and Qiu R C 2017 A data driven approach for resting-state EEG signal classification of schizophrenia with control participants using random matrix theory, arXiv:1712.05289 [82] Röschke J, Fell J and Beckmann P 1995 Nonlinear analysis of sleep EEG data in schizophrenia: calculation of the principal Lyapunov exponent Psychiatry Res. 56 257–69 [83] Merrin E L, Meek P, Floyd T C and Callaway E 1990 Topographic segmentation of waking EEG in medication-free schizophrenic patients Int. J. Psychophysiol. 9 231–6 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-24
- 51. [84] Thilakvathi B, Shenbagadevi S, Bhanu K S and Malaippan M 2017 EEG signal complexity analysis for schizophrenia during rest and mental activity Biomed. Res. 28 1–9 [85] Dvey-Aharon Z, Fogelson N, Peled A and Intrator N 2015 Schizophrenia detection and classification by advanced analysis of EEG recordings using a single electrode approach PLoS One 10 e0123033 [86] Kim J W, Lee Y S, Han D H, Min K J and Lee K 2015 Diagnostic utility of quantitative EEG in un-medicated schizophrenia Neurosci. Lett. 589 126–31 [87] Dvey-A Z, Fogelson N, Peled A and Intrator N 2017 Connectivity maps based analysis of EEG for the advanced diagnosis of schizophrenia attributes PLoS One 12 e0185852 [88] Buettner R, Beil D, Scholtz S and Djemai A 2020 Development of a machine learning based algorithm to accurately detect schizophrenia based on one-minute EEG recordings HICSS- 53 Proc.: 53nd Hawaii Int. Conf. on System Sciences [89] Etevenon P 1984 Intra and inter-hemispheric changes in alpha intensities in EEGs of schizophrenic patients versus matched controls Biol. Psychol. 19 247–56 [90] Al-Fahoum A S and Al-Fraihat A 2014 Methods of EEG signal features extraction using linear analysis in frequency and time-frequency domains ISRN Neurosci. 2014 730218 [91] Khaleghi A, Sheikhani A, Mohammadi M R, Nasrabadi A M, Vand S R, Zarafshan H and Moeini M 2015 EEG classification of adolescents with type I and type II of bipolar disorder Australas. Phys. Eng. Sci. Med. 38 551–9 [92] Bachiller A et al 2014 Decreased entropy modulation of EEG response to novelty and relevance in schizophrenia during a P300 task Eur. Arch. Psychiatry Clin. Neurosci. 265 525–35 [93] Hiesh M, Lam A Y, Shen C, Chen W, Lin F, Sung H, Lin J, Chiu M and Lai F 2013 Classification of schizophrenia using genetic algorithm-support vector machine (GA-SVM) 2013 35th Annual Int. Conf. of the IEEE Engineering in Medicine and Biology Society (EMBC) pp 6047–50 [94] Nasehi S and Pourghassem H 2011 A novel effective feature selection algorithm based on S- PCA and wavelet transform features in EEG signal classification 2011 IEEE 3rd Int. Conf. on Communication Software and Networks [95] Doege K, Bates A T, White T P, Das D, Boks M P and Liddle P F 2009 Reduced event- related low frequency EEG activity in schizophrenia during an auditory oddball task Psychophysiology 46 566–77 [96] Akar S A, Kara S, Latifoglu F and Bilgic V 2012 Wavelet–Welch methodology for analysis of EEG signals of schizophrenia patients 2012 Cairo Int. Biomedical Engineering Conf. (CIBEC) [97] Pauline A D 1942 Comparative study of the EEGs of schizophrenic and manic-depressive patients Am. J. Psychiatry 99 210–7 [98] Haring C, Neudorfer C, Schwitzer J, Hummer M, Saria A, Hinterhuber H and Fleischhacker W W 1994 EEG alterations in patients treated with clozapine in relation to plasma levels Psychopharmacology 114 97–100 [99] Zhao Q, Hu B J, Li Y, Peng H, Li L, Liu Q, Li Y D, Shi Q and Feng J 2013 An alpha resting EEG study on nonlinear dynamic analysis for schizophrenia 2013 6th Int. IEEE/ EMBS Conf. on Neural Engineering (NER) pp 484–8 [100] Ziqiang Z and Puthusserypady S 2007 Analysis of schizophrenic EEG synchrony using empirical mode decomposition 2007 15th Int. Conf. on Digital Signal Processing [101] Phang C R, Ting C M, Numan F and Ombao H 2019 Classification of EEG-based brain connectivity networks in schizophrenia using a multi-domain connectome convolutional neural network, arXiv:1903.08858 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-25
- 52. [102] Ford J M, Palzes V A, Roach B J and Mathalon D H 2014 Did I do that? Abnormal predictive processes in schizophrenia when button pressing to deliver a tone Schizophr. Bull. 40 804–12 [103] Gilles J 2013 Empirical wavelet transform IEEE Trans. Signal Process. 61 3999 [104] Song Y Y and Lu Y 2015 Decision tree methods: applications for classification and prediction Shanghai Arch. Psychiatry 27 130–5 [105] Awad M and Khanna R 2015 Support vector machines for classification Efficient Learning Machines (Berkeley, CA: Apress) [106] Adeniyi D A, Wei Z and Yong-quan Y 2016 Automated web usage data mining and recommendation system using k-nearest neighbor (KNN) classification method Appl. Comput. Inform. 12 1 [107] Bühlmann P 2012 Bagging, boosting and ensemble methods Handbook of Computational Statistics (Berlin: Springer) [108] Tharwat A 2016 Linear vs quadratic discriminant analysis classifier: a tutorial Int. J. Appl. Pattern Recogn. 3 145 Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 1-26
- 53. IOP Publishing Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 Varun Bajaj and G R Sinha Chapter 2 Fuzzy scale invariant feature transform phase locking value and its application to PTSD EEG data Zahra Ghanbari, Mohammad H Moradi Biomedical Department, Amirkabir University of Technology, Tehran, Iran In this study we introduce a novel powerful phase synchrony index called the scale invariant feature transform phase locking value (FSIFT-PLV). This index benefits from robust SIFT descriptors, a phase estimation approach based on reduced interference Rihaczek distribution and a proposed fuzzy framework. FSIFT-PLV can detect bivariate phase synchrony properly in the presence of noise, volume conductance, common reference and sample size bias, as well as small shifts in time. It is applied to electroencephalogram (EEG) signals recorded from combat related post-traumatic stress disorder (PTSD) veterans and two control groups, including trauma exposed non-PTSD veterans and healthy controls who have not experienced any trauma including war. PTSD is a chronic debilitating disorder which may occur as a result of life-threatening mental trauma. The EEG signals are recorded in two resting states (‘eyes-open’ and ‘eyes-closed’). FSIFT-PLV is used to generate functional connectivity matrices. ANOVA is applied to extracted features at a confidence level of 99%. Our study possesses some unique properties: (i) investigating patients who have experienced PTSD symptoms for more than 30 years, (ii) considering trauma exposed non-PTSD veterans as the second control group, (iii) studying the resting state in both the eyes-open and eyes-closed conditions, and (iv) introducing a novel powerful phase synchrony and applying it to EEG signals. doi:10.1088/978-0-7503-3279-8ch2 2-1 ª IOP Publishing Ltd 2020
- 54. 2.1 Introduction This section provides a brief introduction to the background of post-traumatic stress disorder (PTSD), as well as resting state eyes-closed and eyes-open EEG signals. In the following, we will mention some points about phase synchrony calculation. According to the American Psychiatric Association (APA), post-traumatic stress disorder (PTSD) is defined as a psychiatric disorder which may follow a traumatic event [1]. PTSD is a chronic debilitating anxiety condition which is characterized by unremitting distressing repetition of the traumatic experience, avoidance, hyper- arousal, hyper-vigilance, dissociation, emotional numbing and negative alteration in cognition [2]. fMRI studies have reported functional abnormalities in the cortical and sub- cortical regions of PTSD patients’ brains [3, 4]. Functional connectivity is a powerful approach for investigating the brain as a sophisticated network. Many disorders have been studied using functional connectivity, including PTSD [5–8]. Although studying abnormal patterns in the presence of emotional elicitation is of great importance, dysregulated patterns of the resting state functional connectivity may provide valuable knowledge about PTSD pathology [9, 10]. Correlations embedded in haemodynamic activity levels among various brain regions are addressed by resting state functional connectivity. Functional connectivity uses synchronization of the neural activation of the aforementioned regions [11], which can be calculated using different modalities, based on various methods. EEG is a low cost, commonly available modality with high temporal resolution. It can provide optimal observation of brain activities [12, 13]. Only a few studies have focused on the resting state functional connectivity of PTSD patients based on EEG signals. Resting state EEG can be recorded in two forms, including resting state eyes-closed (REC), and resting state eyes-open (REO). REC and REO have different basic properties. For example, the alpha band EEG is defined using synchronization in REC but desynchronization in REO as the feature [14]. Comparing REC and REO demonstrates differences in the delta, theta, alpha and beta sub-bands, both in adult and young participants [15]. One hypothesis states that the differences between REC and REO can be translated as a signal which reflects the brain’s activity in response to visual stimuli. To test this hypothesis, REC and REO EEG signals have been recorded in a completely dark environment. The reported results imply significantly different spectral powers and coherence values in the delta, theta, alpha1, alpha2, beta1, beta2 and gamma sub-bands. These findings suggest that the differences between REC and REO are independent of external stimuli to the visual system. This study also proposes that such differences actually reflect externally directed attention and, in contrast, internally directed attention, specific to REO and REC, respectively [16]. In this study we aim to examine the REC and REO signals recorded from combat related PTSD participants, and two control groups including combat trauma exposed non-PTSD participants and healthy controls who have not experienced any serious trauma including war. For this purpose, we will use functional Modelling and Analysis of Active Biopotential Signals in Healthcare, Volume 1 2-2
- 55. Another Random Scribd Document with Unrelated Content
- 59. The Project Gutenberg eBook of Over the Ocean; or, Sights and Scenes in Foreign Lands
- 60. This ebook is for the use of anyone anywhere in the United States and most other parts of the world at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this ebook or online at www.gutenberg.org. If you are not located in the United States, you will have to check the laws of the country where you are located before using this eBook. Title: Over the Ocean; or, Sights and Scenes in Foreign Lands Author: Curtis Guild Release date: October 30, 2012 [eBook #41233] Most recently updated: October 23, 2024 Language: English Credits: Produced by K Nordquist, Charlie Howard, and the Online Distributed Proofreading Team at http://www.pgdp.net (This book was produced from scanned images of public domain material from the Google Print project.) *** START OF THE PROJECT GUTENBERG EBOOK OVER THE OCEAN; OR, SIGHTS AND SCENES IN FOREIGN LANDS ***
- 61. OVER THE OCEAN; OR, SIGHTS AND SCENES IN FOREIGN LANDS. BY CURTIS GUILD, EDITOR OF THE BOSTON COMMERCIAL BULLETIN. BOSTON: LEE AND SHEPARD, PUBLISHERS. NEW YORK: LEE, SHEPARD AND DILLINGHAM. 1871.
- 62. Entered, according to Act of Congress, in the year 1871, By L E E A N D S H E P A R D, In the Office of the Librarian of Congress, at Washington. Cambridge: Printed by Welch, Bigelow, & Co. Stereotyped at the Boston Stereotype Foundry, No. 10 Spring Lane.
- 63. PREFACE. The following pages are the record of the fruition of years of desire and anticipation; probably the same that fills the hearts of many who will read them—a tour in Europe. The habits of observation, acquired by many years' constant occupation as a journalist, were found by the author to have become almost second nature, even when the duties of that profession were thrown aside for simple gratification and enjoyment; consequently, during a journey of nearly seven months, which was enjoyed with all the zest of a first tour, the matter which composes this volume was prepared. Its original form was in a series of sketches in the columns of the Boston Commercial Bulletin. In these the writer attempted to give as vivid and exact an idea of the sights and scenes which he witnessed as could be conveyed to those who had never visited Europe. Whether describing Westminster Abbey, or York Minster, Stratford-on-Avon, or the streets of London; the wonders of the Louvre, or the gayeties and glitter of Paris; the grandeur of the Alpine passes; the quaintness of old continental cities; experiences of post travelling; the romantic beauties of the Italian lakes; the underground wonders of Adelsberg, or the aqueous highways of Venice,—the author aimed to give many minute particulars, which foreign letter-writers deem of too little importance to mention, but which, nevertheless, are of great interest to the reader.
- 64. That the effort was, in some measure, successful, has been evinced by a demand for the sketches in permanent form, sufficient to warrant the publication of this volume. In so presenting them, it is with the belief that it may be pleasant to those who have visited the same scenes to revisit them in fancy with the writer, and with a hope that the volume may, in some degree, serve as a guide to those who intend to go "over the ocean," as well as an agreeable entertainment to the stay-at-homes. C. G.
- 65. CONTENTS. CHAPTER I. PAGE Going Abroad.—What it costs.—Hints to Tourists.—Life on board Ship.—Land Ho!— Examining Luggage.—The Emerald Isle.— Blarney Castle.—Dublin.—Dublin Castle.— St. Patrick's Cathedral.—Cheap John's Paradise.—Phœnix Park.—Across the Irish Sea.—Railroad travelling in England.— Guard vs. Conductor.—Word to the Wise. —Railroad Stations.—An Old English City. —Chester Cathedral.—The City Walls. 1-28 CHAPTER II. Chester to Liverpool.—An English Breakfast.— A Trial of Patience.—Liverpool Docks.—St. George's Hall.—Poverty and Suffering.— The Lake District.—Home of the Poets.— Keswick.—An English Church.—The Druids' Temple.—Brougham Hall.—A Roadside Inn. 28-46 CHAPTER III. Edinburgh.—Historic Streets.—Edinburgh Castle.—Bonnie Dundee.—Rooms of 47-79
- 66. Historic Story.—The Scottish Regalia.— Curiosities of the Old City.—Holyrood Palace.—Relics of the Past.—Holyrood Abbey.—Antiquarian Museum.—Scott and Scotland.—Hawthornden.—Roslin Chapel. —Melrose Abbey.—The Abbey Hotel.— Abbotsford.—Stirling Castle.—The Tournament Field.—Field of Bannockburn. —Lady of the Lake Scenes.—Scotch Lakes and Hills. CHAPTER IV. Glasgow Cathedral.—Vestiges of Vandalism.— Bible Stories in Colored Glass.—The Actor's Epitaph.—Tam O'Shanter's Ride.— Burns's Cottage.—Kirk Alloway.—A Reminder from the Witches.—Bonnie Doon.—Newcastle-on-Tyne.—York.— Beauties of York Minster.—Old Saxon Relics.—Sheffield.—The Cutlery Works.— English Mechanics.—English Ale.— Chatsworth.—Interior of the Palace.— Sculpture Gallery.—Landscape Effects.— Grand Conservatory.—Haddon Hall. 80-115 CHAPTER V. Kenilworth.—Stratford on Avon.—Interesting Mementos.—Stratford Church.— Shakespeare's Safeguard.—Warwick Castle.—Dungeon and Hall.—Warder's Horn and Warwick Vase.—Leicester's Hospital.—Beauchamp Chapel.—Mugby Junction.—Oxford.—The Mitre Tavern.— Bodleian Library.—Literary Treasures.— 116-151
- 67. Curiosities and Rarities.—Story of an Old Portrait.—Queen Bess on Matrimony.— Addison's Walk.—Boating on the Isis.— Martyr's Memorial. CHAPTER VI. London.—Feeing Servants.—Railway Porters. —London Hotels.—Sights in London Streets.—Cabs and Cab-drivers.—London Shops.—Hints to Buyers.—A London Banking-house.—Routine vs. Courtesy.— Westminster Abbey.—Tombs of Kings and Warriors.—Poets' Corner.—Tributes to Genius.—Penny Steamboat Trip.—Kew Gardens.—The Star and Garter. 152-185 CHAPTER VII. The Original Wax Works.—London Theatres. —Full Dress at the Opera.—Play Bills.—A Palace for the People.—Parks of London.— Zoölogical Gardens.—The Tower of London.—The Silver Key.—Site of the Scaffold.—Knights in Armor.—Regalia of England.—St. Paul's.—The Whispering Gallery.—Up into the Ball.—Down into the Crypt.—Gog and Magog.—Bank of England.—Hampton Court Palace.—The Gardens and People.—Windsor Castle.— Windsor Parks.—London Newspapers.— The Times.—The British Museum.— Bibliographical Curiosities.—Egyptian Galleries.—A Wealth of Antiquities.— Original Magna Charta.—Priceless Manuscripts. 185-246
- 68. CHAPTER VIII. From London to Paris.—Grand Hotels.—The Arch of Triumph.—Paris by Gaslight.—Site of the Guillotine.—Improvements in Paris. —The Bastille.—The Old Guard.—The Louvre.—Gallery of Masterpieces.—Relics of Napoleon I.—Palais Royal.—Jewelry.— French Funeral.—Père La Chaise.—Millions in Marble.—Tomb of Bonaparte.— Versailles.—Halls of the Crusades.— Gallery of the Empire.—Gallery of Battles. —Theatre in the Palace.—Fountains at Versailles.—Notre Dame.—Sainte Chapelle.—The Madeleine.—The Pantheon.—Les Champs Elysées.—Cafés Chantants.—The Jardin Mabille.—The Luxembourg.—Palace of St. Cloud.— Shops in Paris.—Bargains. 246-309 CHAPTER IX. Good by to Paris.—Church of St. Gudule.— Field of Waterloo.—Brussels dash;Antwerp.—The Cathedral Spire.— Dusseldorf.—Cologne Cathedral.—Riches of the Church.—Up the Rhine.—Bridge of Boats.—Coblentz and Ehrenbreitstein.— Stolzenfels.—Legendary Castles.—Bingen on the Rhine.—Roman Remains.— Mayence.—Wiesbaden.—Gambling Halls. —Frankfort-on-the-Main.—Heidelberg Castle.—The Great Tun.—The King's Seat. —Baden-Baden.—Sabbath Amusement.— Satan's Snare baited.—Among the 309-375
- 69. Gamblers.—Scene at the Table.— Strasburg Cathedral.—Strasburg Clock.— Clock at Basle.—Swiss Railways.— Travelling in Switzerland.—Zurich and its Scenery. 309-375 CHAPTER X. The Righi.—Guides and Alpenstocks.— Climbing the Alps.—Night on the Mountain Top.—The Yodlyn.—Lucerne.— Wonderful Organ Playing.—A Sail on Lake Lucerne.—Scene of Tell's Archery.—The St. Gothard Pass.—The Devil's Bridge.— The Brunig Pass.—A Valley of Beauty.— Interlaken.—Staubbach Waterfall.— Glaciers and Avalanches.—An Illuminated Waterfall.—Berne.—The Freiburg Organ.— Lake Leman.—The Prison of Chillon.— Geneva.—Swiss Washerwomen.—Glaciers by Moonlight.—Sunrise on Mont Blanc.— Valley of Chamouny.—View from Flegère. —Climbing again.—Crossing the Sea of Ice.—The Mauvais Pass.—Under a Glacier. —The Tête Noir Pass.—Italian Post Drivers.—The Rhone Valley.—Simplon Pass.—Gorge of Gondo.—Fressinone Waterfall.—Domo d'Ossola.—An Italian Inn.—Lake Maggiore.—Milan Cathedral.— A Wonderful Statue.—Death and Dross.— The La Scala Theatre.—Lake Como.— Italian Monks.—Madesimo Waterfall. 376-450 CHAPTER XI.
- 70. The Splügen Pass.—The Via Main.—Tamina Gorge.—Falls of Schaffhausen.—Munich.— Galleries of Paintings.—Grecian Sculpture restored.—A Bronze Giant.—Hall of the Colossi.—The Palace.—Basilica of St. Boniface.—Salzburg.—Aquarial Wonders. —Visiting Lilliput.—Vienna.—Judging by Appearances.—Royal Regalia.—Cabinet of Minerals.—The Ambras Museum. 450-475 CHAPTER XII. Superb Mausoleum.—The Strauss Band.— Summer Palace.—Imperial Gallery.— Vienna Leather Work.—Shops and Prices. —The Cave of Adelsberg.—Underground Wonders.—Nature's Imitation of Art. 476-487 CHAPTER XIII. Venice.—Gondolas and Gondoliers.—Shylock. —The Rialto.—The Giant's Staircase.—The Lion's Mouth.—Terrible Dungeons.— Square of St. Mark.—The Bronze Horses. —Church of St. Mark.—Titian's Monument. —Canova's Monument.—Cathedrals and Pictures.—Florence.—Art in the Streets.— The Uffizi Gallery.—Old Masters in Battalions.—Hall of Niobe.—Cabinet of Gems.—Michael Angelo's House.—The Duomo.—The Campanile.—Church of Santa Croce.—Michael Angelo's Statuary. —Florentine Mosaics.—Medicean Chapel. —Pitti Palace.—Halls of the Gods.—The Cascine.—Powers, the Sculptor. 487-530
- 71. CHAPTER XIV. Tower of Pisa.—The Duomo.—Galileo's Lamp. —The Baptistery.—Campo Santo.—Over the Apennines.—Genoa.—Streets of Genoa.—Pallavicini Gardens.—Water Jokes.—Turin to Susa.—Mt. Cenis Pass.— Paris again.—Down in the Sewers. 531-548 CHAPTER XV. Sic transit.—English Rudeness.—Wonders of London.—Looking towards Home.—Last Purchases.—English Conservatism.— Reunion of Tourists.—All aboard.—Home again. 549-558
- 72. OVER THE OCEAN.
- 73. CHAPTER I. Do you remember, dear reader, when you were a youngster, and studied a geography with pictures in it, or a "First" or "Second" Book of History, and wondered, as you looked upon the wood-cuts in them, if you should ever see St. Paul's Cathedral, or Westminster Abbey, or London Bridge, or go to the Tower of London, and into the very room in which the poor little princes were smothered by the order of their cruel uncle Richard, by the two rude fellows in a sort of undress armor suit, as depicted in the Child's History of England, or should ever see the Paris you had heard your elders talk so much of, or those curious old Rhine castles, of which we read so many startling legends of robber knights, and fair ladies, and tournaments, and gnomes, and enchanters? What a realm of enchantment to us, story-book readers, was beyond the great blue ocean! and how we resolved, when we grew to be a man, we would travel all over the world, and see every thing, and buy ever so many curious things in the countries where they grew or were made. Even that compound which produced "the finest jet black ever beheld," was to us invested with a sort of poetic interest in boyhood's day, for the very stone jug that we held in our hand had come from London,—"97 High Holborn,"—and there was the picture of the palatial-looking factory on the pink label. LONDON! There was something sonorous in the sound, and something solid in the very appearance of the word when written. When we were a man, didn't we mean to go to London! Years added to youth dissipated many of these air-built castles, and other barriers besides the watery plain intervene between the
- 74. goal of one's wishes, and Europe looks further away than ever. "Going to Europe! Everybody goes to Europe nowadays," says a friend. True, and in these days of steam it is not so much of an event as formerly; indeed, one would judge so from many of his countrymen that he meets abroad, who make him blush to think how they misrepresent Americans. The Great Expositions at London and Paris drew from our shores every American who could by any manner of means or excuse leave business, and obtain funds sufficient to get over and back, if only for a six weeks' visit. The Exposition brought out to Paris and to Europe, among the swarm of Americans who went over, many such, and some who had scarcely visited beyond the confines of their native cities before crossing the Atlantic. These people, by their utter inexperience as travellers, and by their application of the precept inculcated in their minds that money would answer for brains, was a substitute for experience, and the only passport that would be required anywhere and for anything, became a source of mortification to their countrymen, easy game for swindling landlords and sharp shop-keepers, and rendered all the great routes of travel more beset with extortions and annoyances than ever before. But about "going to Europe." When one decides to start on a pleasure trip to that country for the first time, how many very simple things he wishes to know, that correspondents and people who write for the papers have never said anything about. After having once or twice gone over in a steamship, it never seems to occur to these writers that anybody else will want to become acquainted with the little minutiæ of information respecting life on board ship during the trip, and which most people do not like to say they know nothing about; and novices, therefore, have to clumsily learn by experience, and sometimes at four times the usual cost. Speaking of cost, let me say that this is a matter upon which hardly any two tourists will agree. How much does it cost to go to Europe? Of course the cost is varied by the style of living and the
- 75. thoroughness with which one sees sights; by thoroughness I mean, besides expenditure of time, the use of extra shillings "pour boires," and the skilful dispensation of extra funds, which will gain admission to many a forbidden shrine, insure many an unexpected comfort, and shorten many a weary journey. There is one popular error which one quickly becomes disabused of, and that is, that everything abroad is dirt cheap, and it costs a mere song to live. Good articles always bring good prices. Many may be cheaper than at home, it is true, but they are by no means thrown away, and good living in Paris cannot be had, as some suppose, for three francs a day. If one is going abroad for pleasure, and has a taste for travelling, let him first decide what countries he wishes to visit, the routes and time he will take, and then from experienced tourists ascertain about what it would cost; after having learned this, add twenty per cent. to that amount, and he will be safe. Safe in the knowledge that you have enough; safe in being able to make many little purchases that you will never dream of till you reach Regent Street, the Boulevards, the "Piazza San Marco," the Florence mosaic stores, or the Naples coral shops. Safe in making little side excursions to noted places that you will find on your route, and safe from the annoying reflection that you might have done so much better, and seen so much more, if you had not limited the expenditure to that very amount which your friend said would take you through. These remarks of course apply only to those who feel that they can afford but a fixed sum for the journey, and who ought always to wait till they can allow a little margin to the fixed sum, the more completely to enjoy the trip. I have seen Americans in French restaurants actually calculating up the price of a dinner, and figuring out the price of exchange, to
- 76. see if they should order a franc's worth more or less. We may judge how much such men's enjoyment is abridged. On the other hand, the class that I refer to, who imagine that money will pass for everything, increase the cost of travel to all, by their paying without abatement the demands of landlords and shopkeepers. The latter class, on the continent, are so accustomed, as a matter of course, to being "beaten down" in the price, that it has now come to be a saying among them, that he who pays what is at first demanded must be a fool or an American. In Paris, during the Exposition, green Englishmen and freshly-arrived Americans were swindled without mercy. The jewelry shops of the Rue de la Paix, the Grand Hotel, the shops of the Palais Royal, and the very Boulevard cafés fleeced men unmercifully. The entrance of an American into a French store was always the occasion of adding from twenty to twenty-five per cent. to the regular price of the goods. It was a rich harvest to the cringing crew, who, with smirks, shrugs, bows, and pardonnez moi's in the oiliest tones, swindled and cheated without mercy, and then, over their half franc's worth of black coffee at the restaurant, or glass of absinthe, compared notes with each other, and boasted, not how much trade they had secured or business they had done, but how much beyond the legitimate price they had got from the foreign purchaser, whom they laughed at. All the guide-books and many tourists exclaim against baggage, and urge the travelling with a single small trunk, or, as they call it in England, portmanteau. This is very well for a bachelor, travelling entirely alone, and who expects to go into no company, and will save much time and expense at railway stations; but there is some comfort in having wardrobe enough and some space for small purchases, even if a little extra has to be paid. It is the price of convenience in one respect, although the continual weighing of and charging for baggage is annoying to an American, who is unused to that sort of thing; and one very curious circumstance is discovered in
- 77. this weighing, no two scales on the continent give the same weight of the same luggage. Passage tickets from America to Europe it is, of course, always best to secure some time in advance, and a previous visit to the steamer may aid the fresh tourist in getting a state-room near the centre of the ship, near the cabin stairs, and one having a dead- light, all of which are desirable things. Have some old clothes to wear on the voyage; remember it is cold at sea even in summer; and carry, besides your overcoat and warm under-clothing, some shawls and railway rugs, the latter to lie round on deck with when you are seasick. There is no cure for seasickness; keep on deck, and take as much exercise as possible; hot drinks, and a hot water bottle at the feet are reliefs. People's appetites come to them, after seasickness, for the most unaccountable things, and as soon as the patient 'hankers' for anything, by all means let him get it, if it is to be had on board; for it is a sure sign of returning vigor, and in nine cases out of ten, is the very thing that will bring the sufferer relief. I have known a delicate young lady, who had been unable to eat anything but gruel for three days, suddenly have an intense longing for corned beef and cabbage, and, after eating heartily of it, attend her meals regularly the remainder of the voyage. Some make no effort to get well from port to port, and live in their state-rooms on the various little messes they imagine may relieve them, and which are promptly brought either by the stewardess or bedroom steward of the section of state- rooms they occupy. The tickets on the Cunard line express, or did express, that the amount received includes "stewards' fees;" but any one who wants to be well served on the trip will find that a sovereign to the table steward, and one to the bedroom steward,—the first paid the last
- 78. day before reaching port, and the second by instalments of half to commence with, and half just before leaving,—will have a marvellously good effect, and that it is, in fact, an expected fee. If it is your first voyage, and you expect to be sick, speak to the state- room steward, who has charge of the room you occupy, or the stewardess, if you have a lady with you; tell him you shall probably need his attention, and he must look out for you; hand him half a sovereign and your card, with the number of your room, and you will have occasion to experience most satisfactorily the value of British gold before the voyage is over. If a desirable seat at the table is required in the dining-saloon—that is, an outside or end seat, where one can get out and in easily,—or at the table at which the captain sometimes presides, a similar interview with the saloon steward, a day or two before sailing, may accomplish it. Besides these stewards, there are others, who are known as deck stewards, who wait upon seasick passengers, who lie about the decks in various nooks, in pleasant weather, and who have their meals brought to them by these attentive fellows from the cabin table. It is one phase of seasickness that some of the sufferers get well enough to lie languidly about in the fresh, bracing air, and can eat certain viands they may fancy for the nonce, but upon entering the enclosed saloon, are at once, from the confined air or the more perceptible motion of the ship, afflicted with a most irrepressible and disagreeable nausea. Well, the ticket for Liverpool is bought, your letter of credit prepared, and you are all ready for your first trip across the water. People that you know, who have been often, ask, in a nonchalant style, what "boat" you are going "over" in; you thought it was a steamer, and the easy style with which they talk of running over for a few weeks, or should have gone this month, if they hadn't been so busy, or they shall probably see you in Vienna, or Rome, or St. Petersburg, causes you to think that this, to you, tremendous undertaking of a first voyage over the Atlantic is to be but an insignificant excursion, after all, and that the entire romance of the
- 79. affair and the realizing of your imagination is to be dissolved like one of youth's castles in the air. So it seems as you ride down to the steamer, get on board, pushing amid the crowds of passengers and leave-taking friends; and not until a last, and perhaps, tearful leave- taking, and when the vessel fairly swings out into the stream, and you respond to the fluttering signal of dear ones on shore, till rapid receding renders face and form indistinguishable, do you realize that you are fairly launched on the great ocean, and friends and home are left behind, as they never have been before. One's first experience upon the great, awful Ocean is never to be forgotten. My esteem for that great navigator, Christopher Columbus, has risen one hundred per cent. since I have crossed it, to think of the amount of courage, strength of mind, and faith it must have required to sustain him in his venturesome voyage in the frail and imperfect crafts which those of his day must have been. Two days out, and the great broad sweep of the Atlantic makes its influence felt upon all who are in any degree susceptible. To the landsman, the steamship seems to have a regular gigantic see-saw motion, very much like that of the toy ships that used to rise and fall on mimic waves, moved by clock-work, on clocks that used to be displayed in the store windows of jewellers and fancy dealers. Now the bows rise with a grand sweep,—now they sink again as the vessel plunges into an advancing wave,—up and down, up and down, and forging ahead to the never-ceasing, tremulous jar of the machinery. In the calmest weather there is always one vast swell, and when wind or storm prevails, it is both grand and terrible. The great, vast ocean is something so much beyond anything I ever imagined,—the same vast expanse of dark-blue rolling waves as far as the eye can reach,—day after day, day after day,—the great ship a mere speck, an atom in the vast circle of water,—water everywhere. The very wind sounds differently than on land; a cheerful breeze is like the breath of a giant, and a playful wave will send a dozen hogsheads of water over the lofty bulwarks.
- 80. But in a stiff breeze, when a great wave strikes like an iron avalanche against the ship, she seems to pause and shudder, as it were, beneath the blow; then, gathering strength from the unceasing throb of the mighty power within, urges her way bravely on, while far as the eye can reach, as the ship sinks in the watery valleys, you see the great black tossing waves, all crested with spray and foam, like a huge squadron of white-plumed giant cavalry. The spray sometimes flies high over the smoke-stack, and a dash of saline drops, coming fiercely into the face, feels like a handful of pebbles. A look around on the vast expanse, and the ship which at the pier seemed so huge, so strong, so unyielding, becomes an atom in comparison,—is tossed, like a mere feather, upon old Ocean's bosom; and one realizes how little is between him and eternity. There seem to be no places that to my mind bring man so sensibly into the presence of Almighty God as in the midst of the ocean during a storm, or amid the grand and lofty peaks of the Alps; all other feelings are swallowed up in the mute acknowledgment of God's majesty and man's insignificance. If ever twelve days seem long to a man, it is during his first voyage across the Atlantic; and the real beauty of green grass is best appreciated by seeing it on the shores of Queenstown as the steamer sails into Cork harbor. Land again! How well we all are! A sea voyage,—it is nothing. Every one who is going ashore here is in the bustle of preparation. We agree to meet A and party in London; we will call on B in Paris,—yes, we shall come across C in Switzerland. How glib we are talking of the old country! for here it is,—no three thousand miles of ocean to cross now. A clear, bright Sunday morning, and we are going ashore in the little tug which we can see fuming down the harbor to meet us. We part with companions with a feeling of regret. Seated on the deck of the little tug, the steamer again looms up, huge and
- 81. gigantic, and we wonder that the ocean could have so tossed her about. But the bell rings, the ropes are cast off, the tug steams away, our late companions give us three parting cheers, and we respond as the distance rapidly widens between us. Custom-house officials examine your luggage on the tug. American tourists have but very little trouble, and the investigation is slight; cigars and fire-arms not forming a prominent feature in your luggage, but little, if any, inconvenience may be anticipated. This ordeal of the custom-house constitutes one of the most terrible bugbears of the inexperienced traveller. It is the common opinion that an inspection of your baggage means a general and reckless overhauling of the personal property in your trunks—a disclosure of the secrets of the toilet, perhaps of the meagreness of your wardrobe, and a laying of profane hands on things held especially sacred. Ladies naturally dread this experience, and gentlemen, too, who have been foolish enough to stow away some little articles that custom-house regulations have placed under the ban. But the examination is really a very trifling affair; it is conducted courteously and rapidly, and the traveller laughs to himself about his unfounded apprehensions. The tug is at the wharf; the very earth has a pleasant smell; let us get on terra firma. Now, then, a landsman finds out, after his first voyage, what "sea legs" on and sea legs off, that he has read of so much in books, mean. He cannot get used to the steadiness of the ground, or rather, get at once rid of the unsteadiness of the ship. I found myself reeling from side to side on the sidewalk, and on entering the Queen's Hotel, holding on to a desk with one hand, to steady myself, while I wrote with the other. The rolling motion of the ship, to which you have become accustomed, is once more perceptible; and I knew one friend, who did not have a sick day on board ship, who was taken landsick two hours after stepping on shore, and had as
- 82. thorough a casting up of accounts for an hour as any of us experienced on the steamer at sea. The Cunard steamers generally arrive at, or used to arrive at, Queenstown on Sunday mornings, and all who land are eager to get breakfast ashore. We tried the Queen's Hotel, where we got a very fair breakfast, and were charged six or eight shillings for the privilege of the ladies sitting in a room till the meal was ready for us—the first, and I think the only, positive swindle I experienced in Ireland. After breakfast the first ride on an English (or rather Irish) railway train took us to Cork. The road was through a lovely country, and, although it was the first of May, green with verdure as with us in June—no harsh New England east winds; and one can easily see in this country how May-day came to be celebrated with May-queens, dances, and May-poles. To us, just landed from the close steamer, how grateful was the fragrance of the fresh earth, the newly-blossomed trees, and the hedges all alive with twittering sparrows! The country roads were smooth, hard, and clear as a ball-room floor; the greensward, fresh and bright, rolled up in luxuriant waves to the very foot of the great brown-trunked trees; chapel bells were tolling, and we saw the Irish peasantry trudging along to church, for all the world as though they had just stepped out of the pictures in the story-books. There were the women with blue-gray cloaks, with hoods at the back, and broad white caps, men in short corduroys, brogues, bobtail coats, caubeens and shillalah; then there was an occasional little tip-cart of the costermonger and his wife, drawn by a donkey; the jaunting-car, with half a dozen merry occupants, all forming the moving figures in the rich landscape of living green in herbage, and the soft brown of the half moss-covered stone walls, or the corrugated stems of the great trees. We were on shore again; once more upon a footing that did not slide from beneath the very step, and the never-ending broad expanse of heaving blue was exchanged for the more grateful scene of pleasant fields and waving trees; the sufferings of a first voyage
- 83. had already begun to live in remembrance only as a hideous nightmare. A good hotel at Cork is the Imperial Hotel; the attendance prompt, the chamber linen fresh and clean, the viands well prepared. The scenery around Cork is very beautiful, especially on the eastern side, on what is known as the upper and lower Glanmere roads, which command fine views. The principal promenade is a fine raised avenue, or walk, over a mile in length, extending through the meadows midway between two branches of the River Lee, and shaded by a double row of lofty and flourishing elms. Our first walk in Ireland was from the Imperial Hotel to the Mardyke. Fifteen minutes brought us to the River Lee; and now, with the city proper behind us, did we enjoy the lovely scene spread out to view. In the month of May one realizes why Ireland is called the Emerald Isle—such lovely green turf, thick, luxurious, and velvety to the tread, and so lively a green; fancy New England grass varnished and polished, and you have it. The shade trees were all in full leaf, the fruit trees in full flower; sheep and lambs gamboling upon the greensward, birds piping in the hedges, and such hedges, and laburnums, and clambering ivy, and hawthorn, the air perfumed with blossoms, the blue sky in the background pierced by the turrets of an old edifice surrounded by tall trees, round which wheeled circles of cawing rooks; the little cottages we passed, half shrouded in beautiful clambering Irish ivy, that was peopled by the nests of the brisk little sparrows, filling the air with their twitterings; the soft spring breeze, and the beautiful reach of landscape—all seemed a realization of some of those scenes that poets write of, and which we sometimes fancy owe their existence to the luxuriance of imagination.
- 84. Returning, we passed through another portion of the city, which gave us a somewhat different view; it was nearly a mile of Irish cabins. Of course one prominent feature was dirt, and we witnessed Pat in all his national glory. A newly-arrived American cannot help noticing the deference paid to caste and position; we, who treat Irish servants and laborers so well as we do, are surprised to see how much better they treat their employers in Ireland, and how little kind treatment the working class receive from those immediately above them. The civil and deferential Pat who steps aside for a well-dressed couple to pass, and touches his hat, in Cork, is vastly different from the independent, voting Pat that elbows you off the sidewalk, or puffs his fragrant pipe into your very face in America. In Ireland he accepts a shilling with gratitude, and invocation of blessings on the donor; in America he condescends to receive two dollars a day! A fellow-passenger remarked that in the old country they were a race of Touch-hats, in the new one of Go to ——. I found them here obliging and civil, ready to earn an honest penny, and grateful for it, and much more inclined to "blarney" a little extra from the traveller than to swindle it out of him. I made an arrangement with a lively driver to take us to the celebrated Blarney Castle in a jaunting-car—a delightful vehicle to ride in of a pleasant spring day, as it was on that of our excursion. The cars for these rides are hung on springs, are nicely cushioned, and the four passengers sit back to back, facing to the side; and there being no cover or top to the vehicle, there is every opportunity of seeing the passing landscape. No American who has been interested in the beautiful descriptions of English and Irish scenery by the British poets can realize their truthfulness until he looks upon it, the characteristics of the scenery, and the very climate, are so different from our own. The ride to Blarney Castle is a delightfully romantic one, of about six miles; the road, which is smooth, hard, and kept in excellent order,
- 85. winds upon a side hill of the River Lee, which you see continually flashing in and out in its course through the valley below; every inch of ground appears to be beautifully cultivated. The road is lined with old brown stone walls, clad with ivy of every variety—dark-green, polished leaf, Irish ivy, small leaf, heart leaf, broad leaf, and lance leaf, such as we see cultivated in pots and green-houses at home, was here flourishing in wild luxuriance. The climate here is so moist that every rock and stone fence is clad with some kind of verdure; the whole seems to satisfy the eye. The old trees are circled round and round in the ivy clasp; the hedges are in their light-green livery of spring; there are long reaches of pretty rustic lanes, with fresh green turf underneath grand old trees, and there are whole banks of violets and primroses —yes, whole banks of such pretty, yellow primroses as we preserve singly in pots at home. There are grand entrances to avenues leading up to stately estates, pretty ivy-clad cottages, peasants' miserable, thatched cabins, great sweeps of green meadow, and the fields and woods are perfectly musical with singing birds, so unlike America: there are linnets, that pipe beautifully; finches, thrushes, and others, that fill the air with their warblings; skylarks, that rise in regular circles high into the air, singing beautifully, till lost to vision; rooks, that caw solemnly, and gather in conclaves on trees and roofs. Nature seems trying to cover the poverty and squalor that disfigures the land with a mantle of her own luxuriance and beauty. Blarney Castle is a good specimen of an old ruin of that description for the newly-arrived tourist to visit, as it will come up to his expectation in many respects, in appearance, as to what he imagined a ruined castle to be, from books and pictures. It is a fine old building, clad inside and out with ivy, situated near a river of the same name, and on a high limestone rock; it was built in the year 1300. In the reign of Elizabeth it was the strongest fortress in
- 86. Munster, and at different periods has withstood regular sieges; it was demolished, all but the central tower, in the year 1646. The celebrated Blarney Stone is about two feet below the summit of the tower, and held in its place by iron stanchions; and as one is obliged to lie at full length, and stretch over the verge of the parapet, having a friend to hold upon your lower limbs, for fear an accidental slip or giddiness may send you a hundred feet below, it may be imagined that the act of kissing the Blarney Stone is not without its perils. However, that duty performed, and a charming view enjoyed of the rich undulating country from the summit, and inspection made of some of the odd little turret chambers of the tower, and loopholes for archery, we descended, gratified the old woman who acts as key-bearer by crossing her palm with silver, strolled amid the beautiful groves of Blarney for a brief period, and finally rattled off again in our jaunting-cars over the romantic road. The Shelborne House, Dublin, is a hotel after the American style, a good Fifth Avenue sort of affair, clean, and well kept, and opposite a beautiful park (Stephens Green). Americans will find this to be a house that will suit their tastes and desires as well, if not better, than any other in Dublin. Sackville Street, in Dublin, is said to be one of the finest streets in Europe. I cannot agree with the guide-books in this opinion, although, standing on Carlisle Bridge, and looking down this broad avenue, with the Nelson Monument, one hundred and ten feet in height, in the centre, and its stately stores on each side, it certainly has a very fine appearance. Here I first visited shops on the other side of the water, and the very first thing that strikes an American is the promptness with which he is served, the civility with which he is treated, the immense assortment and variety of goods, and the effort of the salesmen to do everything to accommodate the purchaser. They seem to say, by their actions, "We are put here to attend to buyers' wants; to serve them, to wait upon them, to make the goods and the establishment attractive; to sell goods, and we want to sell goods." On the other hand, in our own country the style and manner of the clerks is too often that of "I'm
- 87. just as good, and a little better, than you—buy, if you want, or leave —we don't care whether we sell or not—it's a condescension to inform you of our prices; don't expect any attention." The variety of goods in the foreign shops is marvellous to an American; one pattern or color not suiting, dozens of others are shown, or anything will be made at a few hours' notice. Here in Dublin are the great Irish poplin manufactures; and in these days of high prices, hardly any American lady leaves Dublin without a dress pattern, at least, of this elegant material, which can be obtained in the original packages of the "Original Jacobs" of the trade, Richard Atkinson, in College Green, whose front store is a gallery of medals and appointments, as poplin manufacturer to members of royal families for years and years. The ladies of my party were crazy with delight over the exquisite hues, the splendid quality, the low prices—forgetting, dear creatures, the difference of exchange, and the then existing premium on gold, and sixty per cent. duty that had to be added to the rate before the goods were paid for in America. Notwithstanding the stock, the hue to match the pattern a lady had in her pocket was not to be had. "We can make you a dress, if you can wait, madam," said the polite shopman, "of exactly the same color as your sample." "How long will it take to make it?" "We can deliver it to you in eight or ten days." "O, I shall be in London then," said the lady. "That makes no difference, madam. We will deliver it to you anywhere in London, carriage free." And so, indeed, it was delivered. The order was left, sent to the factory by the shopman, and at the appointed time delivered in London, the lady paying on delivery the same rate as charged for
- 88. similar quality of goods at the store in Dublin, and having the enviable satisfaction of showing the double poplin that was "made expressly to her order"—one dress pattern—"in Dublin." I mention this transaction to show what pains are taken to suit the purchaser, and how any one can get what he wants abroad, if he has the means to pay. This is owing chiefly to the different way of doing business, and also to the sharper competition in the old countries. For instance, the Pacific Mills, of Lawrence, Mass., would never think of opening a retail store for the sale of their goods on Washington Street, Boston; and if an English lady failed to find a piece of goods of the color that suited her, of manufacturing sixteen or eighteen yards to her order, and then sending it, free of express charge, to New York. The quantity and variety of goods on hand are overwhelming; the prices, in comparison with ours, so very low that I wanted to buy a ship-load. Whole stores are devoted to specialities—the beautiful Irish linen in every variety, Irish bog-wood carving in every conceivable form, bracelets, rings, figures, necklaces, breast-pins, &c. I visited one large establishment, where every species of dry goods, fancy goods, haberdashery, and, I think, everything except eatables, were sold. Three hundred and fifty salesmen were employed, the proprietors boarding and lodging a large number of them on the premises. The shops in Dublin are very fine, the prices lower than in London, and the attendance excellent. "But Dublin—are you going to describe Dublin?" Not much, dear reader. Describing cities would only be copying the guide-book, or doing what every newspaper correspondent thinks it necessary to do. Now, if I can think of a few unconsidered trifles, which correspondents do not write about, but which tourists,
- 89. on their first visit, always wish information about, I shall think it doing a service to present them in these sketches. The Nelson Monument, a Doric column of one hundred and ten feet high, upon which is a statue eleven feet high of the hero of the Nile, always attracts the attention of visitors. The great bridges over the Liffey, and the quays, are splendid pieces of workmanship, and worth inspection, and of course you will go to see Dublin Castle. This castle was originally built by order of King John, about the year 1215. But little of it remains now, however, except what is known as the Wardrobe Tower, all the present structure having been built since the seventeenth century. Passing in through the great castle court-yard, a ring at a side door brought a courteous English housekeeper, who showed us through the state apartments. Among the most noteworthy of these was the presence-chamber, in which is a richly-carved and ornamental throne, frescoed ceilings, richly- upholstered furniture, &c., the whole most strikingly reminding one of those scenes at the theatre, where the "duke and attendants," or the "king and courtiers," come on. It is here the lord lieutenant holds his receptions, and where individuals are "presented" to him as the representative of royalty. The great ball-room is magnificent. It is eighty-two feet long, and forty-one wide, and thirty-eight in height, the ceiling being decorated with beautiful paintings. One represents George III., supported by Liberty and Justice, another the Conversion of the Irish by St. Patrick, and the third, a very spirited one, Henry II. receiving the Submission of the Native Irish Chiefs. Henry II. held his first court in Dublin in 1172. The Chapel Royal, immediately adjoining, is a fine Gothic edifice, with a most beautiful interior, the ceiling elegantly carved, and a beautiful stained-glass window, with a representation of Christ before Pilate, figures of the Evangelists, &c. Here, carved and displayed, are the coats-of-arms of the different lord lieutenants from the year 1172 to the present time. The throne of the lord lieutenant in one gallery, and that for the archbishop opposite, are
- 90. Welcome to our website – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookbell.com