optimal selection of mother wavelet for accurate infant cry classification
TRANSCRIPT
TECHNICAL PAPER
Optimal selection of mother wavelet for accurate infant cryclassification
J. Saraswathy • M. Hariharan • Thiyagar Nadarajaw •
Wan Khairunizam • Sazali Yaacob
Received: 31 July 2013 / Accepted: 19 March 2014 / Published online: 2 April 2014
� Australasian College of Physical Scientists and Engineers in Medicine 2014
Abstract Wavelet theory is emerging as one of the pre-
valent tool in signal and image processing applications.
However, the most suitable mother wavelet for these
applications is still a relative question mark amongst
researchers. Selection of best mother wavelet through
parameterization leads to better findings for the analysis in
comparison to random selection. The objective of this
article is to compare the performance of the existing
members of mother wavelets and to select the most suitable
mother wavelet for accurate infant cry classification.
Optimal wavelet is found using three different criteria
namely the degree of similarity of mother wavelets, regu-
larity of mother wavelets and accuracy of correct recog-
nition during classification processes. Recorded normal and
pathological infant cry signals are decomposed into five
levels using wavelet packet transform. Energy and entropy
features are extracted at different sub bands of cry signals
and their effectiveness are tested with four supervised
neural network architectures. Findings of this study
expound that, the Finite impulse response based approxi-
mation of Meyer is the best wavelet candidate for accurate
infant cry classification analysis.
Keywords Infant cries � Mother wavelets � Similarity �Regularity � Classification accuracy
Introduction
Crying is a form of biological magnetic siren for an infant.
It is their only means of communication and infants gen-
erally attract the attention of their external vicinity by
crying to express their needs. Naturally, it is highly non
deterministic and carries numerous levels of information
about an infant as shown in Fig. 1 [1]. Consequently, it is
really a confusing task to identify the exact purpose of the
cry signals. Investigations on the newborn cry signals are
previously performed mainly to detect the pathological
status of the recently born infants by using various types of
conventional methods namely auditory analysis—one of
the more common method in infant cry recognition ana-
lysis and the main tool of this analysis is the human ear
which could distinguish different types of signals after
some repetitions and experiences, time domain analysis—a
discrimination method which requires the time domain
based features of signals such as latency and amplitude of
the signals, frequency domain analysis—classification
based on the frequency information of signals and spec-
trographic analysis—an amalgamation of time and fre-
quency domain analysis and has been an imperative tool in
acoustic analysis of infant cry [2]. Although these existing
methods have drawn noteworthy impacts in the infant cry
classification area, they are totally based on the subjective
evaluation, require good expertise, intangible and time
consuming. Figure 1 briefly illustrates the drawbacks and
limitations of the fore-mentioned conventional methods.
Hence, for immaculate diagnostic the needs for the
automatic classification of infant cry signals are emerging
rapidly due to its significant perks. Being a fully automated
system, the diagnosis judgment and results will be accu-
rate, fast and not limited to the quantity of infant cry signal
which are under diagnosis. Manual inspection of experts
J. Saraswathy (&) � M. Hariharan � W. Khairunizam �S. Yaacob
School of Mechatronic Engineering, University Malaysia Perlis
(UniMAP), Campus Pauh Putra, 02600 Arau, Perlis, Malaysia
e-mail: [email protected]
T. Nadarajaw
Department of Pediatrics, Hospital Sultanah Bahiyah,
05460 Alor Setar, Kedah, Malaysia
123
Australas Phys Eng Sci Med (2014) 37:439–456
DOI 10.1007/s13246-014-0264-y
will no longer be required. Moreover, it will be easily
reckonable, completely harmless and not unbearable to the
infant. This non-invasive method has been widely used in
infant cry signal analysis and has shown very promising
results. In the development of automated infant cry clas-
sification systems, momentous researches have been car-
ried out on this infant cry classification analysis and
successfully detected certain pathological conditions
among recently newborn babies such as brain damage [3],
cleft palate [4], hydrocephalus [5], sudden infant death
syndrome [6] and others [7, 8]. Recently, the classification
of two or three classes of infant cry signals, especially
using the normal, asphyxia and deaf cries is a detour. This
is because, asphyxia is a type of respiratory disorder which
may cause some peril long-lasting problems such as cere-
bral palsy, mental retardation, speaking, hearing, visual and
learning disabilities and even fatality if not subjected to
early diagnosis and treatments. According to the World
Health Organization forecast, in worldwide 4 to 9 million
cases of newborn asphyxia are reported annually and 20 %
of all newborn deaths are due to this mess [9, 10]. In
addition, deafness or ‘hypo-acoustic’ which defined as the
insufficiency of hearing ability may deter the performance
of child’s learning and development stages, especially in
school life if not subjected to early diagnosis and treat-
ments [11].
Rosales et al. [12] investigated the application of fuzzy
relational neural network (FRNN) in discriminating the
extracted Mel frequency cepstral coefficients (MFCCs) of
normal and asphyxia cry signals and achieved best accu-
racy of 88.67 %. Zabidi et al. [13] presented an analysis on
binary particle swarm optimization for selection of MFCCs
Infant Cry
Identity
Emotions
Weight Health
First Cry PretermVs fullterm
Gender
Pathology
Conventional classification methods
Auditory
Not reproducible
Provides only a meager or a tiny proportion of information
The rate of correct classification is highly depends on the experience and expertise
Time domain
Provides limited information (only time based details are available)
Frequency based information are not provided
Requires analysis by an expert
Frequency domain
Provides a vulgar representation of the frequency spectrum characteristics
Time based information are not provided
Requires analysis by an expert
Spectrographic
Requires manual and wary inspection
Restricted to large quantity of signals under analysis
Requires analysis by an expert
Fig. 1 Diverse levels of
information conveyed in infant
cry and the existing methods in
infant cry classification
440 Australas Phys Eng Sci Med (2014) 37:439–456
123
in the recognition of infant cries with asphyxia. The highest
correct recognition rate of 95.07 % was reported by using
Multi layer perceptron (MLP) neural network which was
trained with scaled conjugate gradient algorithm. Wavelet
packet transform (WPT) based features used for charac-
terizing the normal and pathological infant cry (asphyxia)
signals. This study reported an optimal recognition rate of
99 % using Probabilistic neural network (PNN) [14]. Ro-
sales et al. [15] analyzed the effectiveness of their proposed
genetic selection of fuzzy model (GSFM) which was
modeled with an optimized combination of feature selec-
tion method, type of fuzzy processing and learning algo-
rithm using genetic algorithm technique, in discrimination
of the extracted MFCCs from normal and asphyxia cries.
The best diagnosis accuracy of their proposed method was
90.68 %. PNN and General regression neural network
(GRNN) classifiers used to classify normal and asphyxia
cries using time frequency based statistical features and
reported 99 % as the best achieved accuracy, employing
principal component analysis (PCA) as feature reduction
method [16]. Maximum accuracy of 97.55 % successfully
presented in discrimination of normal and deaf infants
using MFCCs and FRNN [12]. Time frequency based
statistical features proposed for automatic classification of
normal and deaf cry signals, and the best performance of
the proposed features reported as 99 % using GRNN
classifier [17]. MFCCs and GSFM which designed with an
optimal combination of feature selection method, fuzzy
processing type and learning algorithm implemented to
distinguish normal and deaf cries, resulted with optimum
accuracy of 99.42 % [15]. Hariharan et al. [9] developed a
method based weighted linear prediction cepstal coefficient
(WLPCC) and PNN for the detection of normal and path-
ological (asphyxia and deaf) status from infant cry signals.
Due to the highly non-stationary characteristic of infant
cry signals, the time–frequency analysis is an excellent
approach for analyzing them, in time and frequency scale
simultaneously without loss of any prominent information
[17]. To the best of our knowledge, there is no research on
selection of suitable mother wavelet for classification of
different classes of infant cry signals with high accuracy by
focusing on the time frequency analysis. The proposed
research work’s aim is to select the best mother wavelet for
infant cry classification by investigating the effectiveness
of different mother wavelets (haar, daubechies, symlet,
coiflet, biorthogonal, reverse biorthogonal and finite
impulse response (FIR) based approximation of Meyer) in
three different criteria such as: degree of similarity of
mother wavelet with cry signal by assessing the cross
correlation coefficient, regularity of mother wavelet in
terms of the distribution of significant extracted wavelet
packet based features using respective mother wavelets and
classification accuracy of binary (experiment 1: normal vs
asphyxia and experiment 2: normal vs deaf) and multi class
problems (experiment 3: normal vs asphyxia vs deaf) using
the wavelet packet based features from different mother
wavelets as inputs for various supervised classifiers.
The rest of the paper is organized as follows. ‘‘Infant cry
database’’ section deals with a brief explanation of the
infant cry database used in this work. ‘‘Proposed Method-
ology for Selection of Best Mother Wavelet’’ section deals
with the proposed methodology of this present work,
including introduction to mother wavelet, WPTand the
feature extraction of energy and Shannon entropy features
with the employed classifiers. The results and discussion
from the three different selection methods of this study are
briefly presented in ‘‘Results and discussion’’ section.
Finally, this work concluded in ‘‘Conclusion’’ section with
some future directions.
Infant cry database
The infant cry signals under investigation are obtained
from a standard Mexican database which is a property of
the Instituto Nacional de Astrofisica Optica y Electronica
(INAOE)–CONACYT, Mexico [18]. It consists of 507 of
normal cry signals, 340 of asphyxia cry signals and 879 of
deaf cry signals with the length of 1 s. The infant cry
samples are recorded directly by specialized physicians
from just born up to 6 month old of babies. The samples
are labeled in the moment of their recording. Labels con-
tain information about the cause of the cry or the pathology
presented. Asphyxia is determined by the presence of
metabolic acidosis (pH 7.00), apgar of 0–3 to 5 min and
neurological manifestations as convulsions, coma or
hypotonic, as well as evidence of multi-organic dysfunc-
tion, with cellular and biochemical damage and circulatory
alterations. The collection of deaf samples is carried out
from babies who already diagnosed as deaf by a group of
doctors specialized in communication disorders [19, 20].
All the cry signals which used for our analysis are re-
sampled to 16 kHz [14]. Table 1 tabulates the character-
istics of the database and the samples used for the three
different experiments (experiment 1, experiment 2 and
experiment 3) of our analysis.
Figure 2 demonstrates the estimated energy spectrum of
infant cry signals (normal, deaf and asphyxia). By visually
inspecting the Fig. 2, one may distinguish the different
patterns of cry signals. Nevertheless, it may lead to
incorrect elucidation from the spectrum plot or misclassi-
fication as well since there are higher degrees of overlaps
between the spectrums of cry signals for certain frequency
bands and will strictly requires good knowledge and
expertise to analyze. Hence, automatic recognition of
infant cry signals is desired by using advanced signal
Australas Phys Eng Sci Med (2014) 37:439–456 441
123
processing techniques which are necessary for mining the
useful information of cry signals for quantification and
efficient discrimination of cry signals.
Proposed methodology for selection of best mother
wavelet
Due to the highly non stationary characteristics of infant
cry signals, the performance of the newborn signals with
different mother wavelets is investigated in different cir-
cumstances or manner to enhance the selection result. In
the current study, the best mother wavelet for infant cry
classification is selected through evaluating the perfor-
mance of the different mother wavelets based on the three
distinguishable criteria: degree of similarity of mother
wavelets with cry signals, regularity of mother wavelets
and experimental results. Figure 3 illustrates entirely the
overall block diagram of the proposed methodology of the
analysis which incorporates the respective methods of the
selected criteria. The methodologies used in the present
study were described briefly in the following sections.
Method 1: similarity of mother wavelet with cry signals
One of the most paramount elements that must be con-
sidered in wavelet domain studies is the similarity of the
signal under investigation with the wavelet to be analyzed.
Good similarity between different waveforms is necessary
for better analysis and consistent results. A mother wavelet
is said to be similar with a signal, if the wavelet is able to
divulge its own frequency spectrums when correlated with
a signal, which are also contained in the signal under
analysis [21, 22]. Cross correlation is a superb tool to
measure the similarity of two waveforms as a function of a
time as it is insensitive to noise, simple and versatile.
Hence, cross correlation technique is used to evaluate and
asses the degree of similarity between mother wavelets and
different (normal and pathological) cry signals. In this
study, the low pass wavelet filter from wavelet filter bank
MATLAB [23] and one unit sample of infant cry signal
from different classes are cross correlated. All the signals
are normalized between the range of 0 and 1 before cross
correlation. Hence, the co-efficient value for each cross
correlation would possess highest value of 1 and minimum
value of 0. Cross coefficient value which is nearer to ‘1’
indicates the good similarity whereas ‘0’ refer to worst
similarity of two waveforms. Accordingly after passing
through cross correlation, the coefficients’ values amongst
all cross correlated coefficients are considered for selection
of best mother wavelet [21, 22].
Thus the following steps are followed to select the best
wavelet in cross correlation coefficient:
1. A specific mother wavelet is selected, low pass,
decomposed from wavelet filter bank MATLAB library.
2. The cross correlation coefficient is computed between
normalized cry signal and normalized selected mother
wavelet filter.
3. The best mother wavelet which maximizes the cross
correlation coefficient is selected.
Method 2: regularity of mother wavelets
Regularity is one of the most vital properties of wavelet basis
because it is responsible for a number of key wavelet prop-
erties such as vanishing moments, an order of approxima-
tion, smoothness of the mother wavelets and reproduction of
polynomials. It is also useful for getting nice and significant
features, like smoothness of the reconstructed signal, and for
the estimated function in nonlinear regression analysis [21–
24]. Normally in image processing applications, the regu-
larity of mother wavelets is determined by analyzing the
smoothness of the reconstructed signal, and by calculating
some significant parameters such as compression ratio, dis-
tortion, root mean square error and cross correlation [24, 25].
Theoretically, the decomposed wavelet coefficients are used
to reconstruct back the original signal, good wavelet coef-
ficients that are retained the maximal originality of the signal
with minimum distortions of artifacts or unwanted noises
which may originated from the decomposition algorithm will
reproduce a smoother signal. In the study, in order to identify
and asses the regularity level of different mother wavelets the
significance of wavelet packet based features of different
datasets (normal vs asphyxia and normal vs deaf) which are
computed from wavelet coefficients are considered. An
Table 1 Characteristics of database
Features Original database Experiment 1 Experiment 2 Experiment 3
Normal Asphyxia Deaf Normal Asphyxia Normal Deaf Normal Asphyxia Deaf
Number of samples 507 340 879 340 340 507 507 340 340 340
Sampling frequency, fs (Hz) 22,050 11,025 8,000 16,000 16,000 16,000 16,000 16,000 16,000 16,000
Sample length (s) 1 1 1 1 1 1 1 1 1 1
Experiment 1, normal vs asphyxia; experiment 2, normal vs deaf; experiment 3, normal versus asphyxia vs deaf
442 Australas Phys Eng Sci Med (2014) 37:439–456
123
0 1000 2000 3000 4000 5000 6000 7000 8000-20
-10
0
10
20
30
40
50
60
70
80
Frequency (Hz)
dB
DeafNormalAsphyxia
Fig. 2 Estimated spectrum of
the corresponding cry signals
Feature extraction using: haar,db2,db3,db4,db5,db6,db7,db8,db9,db10,db20,
sym2,sym3,sym4,sym5,sym6,sym7,sym8, sym9, sym10, coif1,coif2,coif3,coif4,coif5,bior1.1,
bior1.3, bior1.5,bior2.2,bior2.4,bior2.6,bior2.8,bior3.1,bior3.3,bior3.5,bior3.7,
bior3.9,bior4.4,bior5.5, bior6.8,rbio1.1,rbio1.3,rbio1.5,rbio2.2,rbio2.4,rbio2.6,rbio2.8,rbio3.1,
rbio3.3,rbio3.5,rbio3.7,rbio3.9,rbio4.4,rbio5.5,rbio6.8 and dmey at 5th level decomposition
Normal versus Asphyxia
versus Deaf
Normal versus Deaf
Normal versus
Asphyxia
Infant cry signal
Wavelet packet transform (Convolution of cry signal with mother
wavelet)
Extracted wavelet coefficients (Energy & Entropy)
Classification of infant cries (PNN, GRNN, MLP and TDNN)
Method 1: Similarity of
mother wavelets with
cry signals
Method 2: Regularity of
mother wavelets
Method 3: Classification
results
Feature extraction using: haar,db2,db3,db4,db5,db6,db7,db8,db9,db10,db20,
sym2,sym3,sym4,sym5,sym6,sym7,sym8, sym9, sym10, coif1,coif2,coif3,coif4,coif5,bior1.1,
bior1.3, bior1.5,bior2.2,bior2.4,bior2.6,bior2.8,bior3.1,bior3.3,bior3.5,bior3.7,
bior3.9,bior4.4,bior5.5, bior6.8,rbio1.1,rbio1.3,rbio1.5,rbio2.2,rbio2.4,rbio2.6,rbio2.8,rbio3.1,
rbio3.3,rbio3.5,rbio3.7,rbio3.9,rbio4.4,rbio5.5,rbio6.8 and dmey at 5th level decomposition
Normal versus Asphyxia
versus Deaf
Normal versus Deaf
Normal versus
Asphyxia
Infant cry signal
Wavelet packet transform (Convolution of cry signal with mother
wavelet)
Extracted wavelet coefficients (Energy & Entropy)
Classification of infant cries (PNN, GRNN, MLP and TDNN)
Fig. 3 Block diagram of the
proposed best mother wavelet
selection methodology
Australas Phys Eng Sci Med (2014) 37:439–456 443
123
independent sample t test (p \ 0.0001 and 99.99 % of con-
fidence interval) is performed to evaluate the number of
significant wavelet based features extracted from each
mother wavelets. The features (energy and entropy) are
extracted at different sub bands using wavelet packet with
different mother wavelets namely haar, daubechies, symlet,
coiflet, biorthogonal, reverse biorthogonal and FIR based
approximation of Meyer (dmey). Number of decomposition
level is chosen as five based on the previous work by Ha-
riharan et al. [14] since they have reported that the maximum
accuracies are obtained from the fifth level of wavelet packet
decomposition using PNN in classifying normal and
asphyxia cry signals (Please refer to the ’’Extracted wavelet
coefficients’’ section for further information on extraction of
energy and entropy features).
Thus the following steps are followed to select the
optimal wavelet in number of significant features:
1. Cry signal is decomposed into fifth level using
WPTand with a specific mother wavelet.
2. Energy and entropy features are computed at different
sub bands of cry signals.
3. The independent t test (p \ 0.0001) is performed
among the extracted wavelet packet based features of
the datasets (normal vs asphyxia and normal vs deaf).
4. The best mother wavelet which maximizes the number
of significant features is selected.
Method 3: classification results
In medical diagnostic area, it is necessary to discriminate
different patterns of samples effectively with higher rate
of correct classification or accuracy. Hence, the classifi-
cation result is used as one of the selection method
for identifying the best mother wavelet for infant cry
classification. Four different types of artificial neural
networks (PNN, GRNN, MLP and TDNN - Please refer
to’’Classifiers’’ section for further information on these
classifiers) are trained to classify the different wavelet
packet based cry features which are extracted from the
fifth level of decomposition into three different classes
(experiment 1: normal vs asphyxia, experiment 2: normal
vs deaf and experiment 3: normal vs asphyxia vs deaf).
Two classification validation schemes are (conventional
and 10-fold cross validation) are used to prove the
steadfastness of the classification results.
Thus the following steps are followed to select the
optimal wavelet in empirical accuracy:
1. Cry signal is decomposed into fifth level using wavelet
packet transform with a specific mother wavelet.
2. Energy and entropy features are computed at different
sub bands of cry signals.
3. Extracted feature vectors are discriminated using PNN,
GRNN, MLP and TDNN classifiers through conven-
tional and 10-fold cross validation schemes.
4. The best mother wavelet which maximizes the classi-
fication accuracy of the three different experiments is
selected.
Mother wavelet and wavelet packets transform
Mother wavelet is a basic wave shaped signal which is
associated with translation and dilation activities when
involve with a signal decomposition algorithm. If w(t) is a
mother wavelet, the basis function at discrete scale a and
discrete dilation b is as shown in Eq. 1.
wa;bðtÞ ¼ 2�a=2wð2�at � bÞ ð1Þ
where a and b are the discrete dilation and discrete transla-
tion respectively. The inner product of the basis function
with the signal at different scales and translations may endow
with the complete spectrum of wavelet coefficients [26].
For the present investigation, a set of different types of
mother wavelets (haar, daubechies, symlet, coiflet, bior-
thogonal, reverse biorthogonal and FIR based approximation
Table 2 Characteristics of different mother wavelets
Wavelet Surname Biorthogonal Symmetry Orthogonality Compact
support
Vanishing
order
Filter length
Haar ‘haar’ Yes Yes Yes Yes 1 2
Daubechies ‘db’ Yes Far from Yes Yes N 2 N
Symlet ‘sym’ Yes Near from Yes Yes N 2 N
Coiflet ‘coif’ Yes Near from Yes Yes N 6 N
Biorthogonal ‘bior’ Yes Yes No Yes Nr, Nd Max(2Nr, 2Nd) ? 2
Reverse Biorthogonal ‘rbio’ Yes Yes No Yes Nr, Nd Max(2Nr, 2Nd) ? 2
Finite impulse response (FIR)
based approximation of Meyer
‘dmey’ Yes Yes Yes Yes – 62
N, Order of wavelet; recon, reconstruction; dec, decomposition
444 Australas Phys Eng Sci Med (2014) 37:439–456
123
of Meyer) was considered. These wavelet families are suit-
able for both continuous and discrete wavelet transform
(DWT), however they differ in characteristics. Table 2,
presents the crucial characteristics of mother wavelets
namely symmetry (useful in avoiding de-phasing), compact
support (allow efficient implementation), orthogonality
(allow fast algorithm), filter length (determine degree of
smoothness), biorthogonal (provides phase linearity) and
vanishing order [24, 25]. The further information regarding
these wavelet functions can be reviewed from earlier
research works [27–29].
WPT is an extension of wavelet transform (WT) which
requires a mother wavelet for its algorithm function [30]. It
has been widely and successfully applied in different
applications [30–32], since WPT splits the original signals
into both low and high frequency bands as well as provides
more and better frequency resolution features about the
original signal of analysis. Furthermore, the multi resolu-
tion property of WPT is very useful in voice signal pro-
cessing areas [32]. The major difference between WT and
WPT is the structure of the binary tree, where the WT gives
a left recursive binary tree structure by decomposing the
lower frequency band whereas WPT gives a balanced
binary tree structure by decomposing both the lower
(approximation coefficients) and higher frequency bands
(detail coefficients) [14].
Extracted wavelet coefficients
In this present work, the normal and pathological infant cry
signals are decomposed into five levels by different mother
wavelets: haar, daubechies (order2–10 & 20), symlet (order
2–10), coiflet (order 1–5), biorthogonal (order 1.1, 1.3, 1.5,
2.2, 2.4, 2.6, 2.8, 3.1, 3.3, 3.5, 3.7, 3.9, 4.4, 5.5 and 6.8),
reverse biorthogonal (order 1.1, 1.3, 1.5, 2.2, 2.4, 2.6, 2.8,
3.1, 3.3, 3.5, 3.7, 3.9, 4.4, 5.5 and 6.8) and FIR based
approximation of Meyer (dmey). Energy and Shannon
entropy are computed using the extracted wavelet packet
coefficients as shown in Fig. 4 The Eqs. 2 and 3 are used to
extract the sub band energy and entropy features
respectively.
Energy ¼ log 10
Pmi¼1 C
p5;k
���
���2
L
2
64
3
75
m ¼ 1; 2; 3. . .5; k ¼ 0; 1; 2. . .25 � 1
ð2Þ
where k is the wavelet packet node, m represents the
number of decomposition level, p is the scale index and L
is the number of wavelet coefficients of corresponding sub
bands.
Infant Cry Signal
f1 f2 f3 f4 f5 f6 f7 f8 f9 f10 f11 f12 f13 f14 f15 f16 f17 f18 f19 f20 f21 f22 f23 f24 f25 f26 f27 f28 f29 f30 f31 f32
Energy Entropy…
1st
3rd
2nd
4th
5th
Fig. 4 Wavelet packet based features (energy and entropy) extraction
Table 3 The learning parameters of the MLP and TDNN neural
networks
Network function MLP TDNN
Number of layers 3 3 with input
delay (0,1)
Number of input neurons 32 32
Number of hidden neurons 23/24 23/24
Number of output neurons 2* and 3* 2* and 3*
Performance goal 0.001 0.001
Learning rate 0.1 0.1
Momentum factor 0.9 0.9
Training algorithm Scaled conjugate
algorithm
Scaled conjugate
algorithm
Activation function ‘logsig’, ‘logsig’ ‘logsig’, ‘logsig’
2*, For normal versus asphyxia & normal versus deaf; 3*, For normal
versus asphyxia vs deaf
Australas Phys Eng Sci Med (2014) 37:439–456 445
123
Entropy ¼ �Xm
i¼1
Cp5;k
���
���2
log Cp5;k
���
���2
;m ¼ 1; 2; 3. . . 5;
k ¼ 0; 1; 2. . .25 � 1
ð3Þ
where k is the wavelet packet node, m represents the
number of decomposition level and p is the scale index.
Through this process, 32 energy and 32 entropy features
are extracted from different cry signals.
Classifiers
Artificial neural networks are artificially designed decision
making tools with many interconnected neurons. Recently,
the importance or usage of artificial neural networks in
multidisciplinary areas is cannot be denied [33–35]. In the
present study, two radial basis neural networks namely
PNN and GRNN are used for the classification of normal
and pathological cries, since they have some advanta-
ges such as being relatively robust with any external
disturbances and extra as reported in [36–39]. The net-
works comprise of 4 different layers such as input layer,
patter layer which is activated by exponential function for
this analysis, summation layer and output layer. Smoothing
parameter (r) is a key element of these radial basis net-
works because the performance of these networks is highly
rely on that [16]. Seeing that, the smoothing parameter for
PNN and GRNN is varied between 0.04 and 0.085 in steps
of 0.005 based on the experimental investigations. The
detailed mathematical derivations about these radial basis
neural networks can be found in these papers [36–39].
To compare the reliability of classification results of
radial basis neural networks (PNN and GRNN), commonly
used neural network models in previous times namely
multilayer perceptron and time-delay neural network are
also used as classifiers. The number of hidden neurons for
MLP and TDNN structures are chosen based on a criteria,
that the number of hidden neurons should be 2/3 the size of
the input layer, plus the size of the output layer (23 hidden
neurons for two class & 24 hidden neurons for three class
problems) [40]. The other learning parameters of MLP and
Fig. 5 Comparative plots of
correlation coefficients with
different mother wavelet filters
for normal and pathological cry
signals
Fig. 6 Number of the
significant features of different
datasets from chosen mother
wavelets selection through
classification accuracy
446 Australas Phys Eng Sci Med (2014) 37:439–456
123
Ta
ble
4S
um
mar
yo
fst
atis
tics
of
extr
acte
den
erg
yfe
atu
res
(dm
ey)
Fea
ture
sN
orm
alA
sph
yx
iaP
val
ue
No
rmal
Dea
fP
val
uea
No
rmal
Dea
fP
val
ueb
Asp
hy
xia
Dea
fP
val
ue
1-
2.8
0±
0.1
8-
2.0
2±
0.9
1\
0.0
01
-2
.60
±0
.49
-3
.11
±0
.49
\0
.00
1-
2.8
0±
0.1
8-
3.1
2±
0.5
2\
0.0
01
-2
.02
±0
.91
-3
.12
±0
.52
\0
.00
1
2-
2.1
8±
0.6
7-
2.6
9±
1.2
0\
0.0
01
-1
.98
±0
.71
-2
.70
±0
.55
\0
.00
1-
2.1
8±
0.6
7-
2.6
3±
0.5
4\
0.0
01
-2
.69
±1
.20
-2
.63
±0
.54
[0
.00
1*
3-
1.9
8±
0.6
0-
2.4
1±
1.2
7\
0.0
01
-1
.81
±0
.63
-1
.48
±0
.75
\0
.00
1-
1.9
8±
0.6
0-
1.3
3±
0.7
0\
0.0
01
-2
.41
±1
.27
-1
.33
±0
.70
\0
.00
1
4-
2.1
3±
0.6
8-
2.6
8±
1.1
8\
0.0
01
-1
.94
±0
.70
-2
.30
±0
.66
\0
.00
1-
2.1
3±
0.6
8-
2.1
7±
0.6
4[
0.0
01
*-
2.6
8±
1.1
8-
2.1
7±
0.6
4\
0.0
01
5-
3.0
0±
0.7
6-
3.3
8±
1.1
9\
0.0
01
-2
.97
±0
.75
-1
.64
±0
.50
\0
.00
1-
3.0
0±
0.7
6-
1.6
5±
0.4
8\
0.0
01
-3
.38
±1
.19
-1
.65
±0
.48
\0
.00
1
6-
3.2
4±
0.6
1-
3.1
4±
1.3
1[
0.0
01
*-
3.1
2±
0.7
6-
1.7
5±
0.6
2\
0.0
01
-3
.24
±0
.61
-1
.84
±0
.60
\0
.00
1-
3.1
4±
1.3
1-
1.8
4±
0.6
0\
0.0
01
7-
2.7
0±
0.6
0-
2.4
5±
1.3
4[
0.0
01
*-
2.5
3±
0.7
5-
1.5
3±
0.7
1\
0.0
01
-2
.70
±0
.60
-1
.47
±0
.70
\0
.00
1-
2.4
5±
1.3
4-
1.4
7±
0.7
0\
0.0
01
8-
3.0
8±
0.5
7-
2.7
4±
1.4
7\
0.0
01
-2
.9±
0.7
7-
2.0
5±
0.6
4\
0.0
01
-3
.08
±0
.57
-2
.12
±0
.63
\0
.00
1-
2.7
4±
1.4
7-
2.1
2±
0.6
3\
0.0
01
9-
4.0
9±
0.6
4-
4.3
7±
1.0
2\
0.0
01
-4
.15
±0
.74
-1
.45
±0
.48
\0
.00
1-
4.0
9±
0.6
4-
1.5
5±
0.4
5\
0.0
01
-4
.37
±1
.02
-1
.55
±0
.45
\0
.00
1
10
-3
.78
±0
.68
-4
.32
±1
.12
\0
.00
1-
3.9
0±
0.7
8-
1.5
2±
0.4
7\
0.0
01
-3
.78
±0
.68
-1
.61
±0
.42
\0
.00
1-
4.3
2±
1.1
2-
1.6
1±
0.4
2\
0.0
01
11
-3
.24
±0
.88
-4
.07
±1
.31
\0
.00
1-
3.4
0±
0.9
0-
1.3
3±
0.5
7\
0.0
01
-3
.24
±0
.88
-1
.42
±0
.50
\0
.00
1-
4.0
7±
1.3
1-
1.4
2±
0.5
0\
0.0
01
12
-3
.61
±0
.78
-4
.25
±1
.23
\0
.00
1-
3.7
4±
0.8
2-
1.4
6±
0.5
3\
0.0
01
-3
.61
±0
.78
-1
.56
±0
.46
\0
.00
1-
4.2
5±
1.2
3-
1.5
6±
0.4
6\
0.0
01
13
-2
.96
±0
.82
-3
.48
±1
.24
\0
.00
1-
2.9
4±
0.8
1-
1.4
7±
0.4
6\
0.0
01
-2
.96
±0
.82
-1
.44
±0
.46
\0
.00
1-
3.4
8±
1.2
4-
1.4
4±
0.4
6\
0.0
01
14
-2
.89
±0
.89
-3
.63
±1
.36
\0
.00
1-
2.9
4±
0.8
7-
1.2
6±
0.5
0\
0.0
01
-2
.89
±0
.89
-1
.23
±0
.52
\0
.00
1-
3.6
3±
1.3
6-
1.2
3±
0.5
2\
0.0
01
15
-3
.01
±0
.91
-3
.90
±1
.32
\0
.00
1-
3.1
8±
0.9
2-
1.2
4±
0.5
7\
0.0
01
-3
.01
±0
.91
-1
.29
±0
.57
\0
.00
1-
3.9
0±
1.3
2-
1.2
9±
0.5
7\
0.0
01
16
-2
.88
±0
.88
-3
.76
±1
.37
\0
.00
1-
2.9
8±
0.8
8-
1.3
4±
0.5
5\
0.0
01
-2
.88
±0
.88
-1
.32
±0
.56
\0
.00
1-
3.7
6±
1.3
7-
1.3
2±
0.5
6\
0.0
01
17
-7
.85
±1
.01
-8
.14
±1
.30
\0
.00
1-
7.8
7±
1.0
2-
5.6
3±
0.9
3\
0.0
01
-7
.85
±1
.01
-5
.72
±0
.95
\0
.00
1-
8.1
4±
1.3
0-
5.7
2±
0.9
5\
0.0
01
18
-7
.64
±0
.91
-7
.95
±1
.25
\0
.00
1-
7.5
5±
0.8
6-
5.6
1±
0.9
0\
0.0
01
-7
.64
±0
.91
-5
.70
±0
.93
\0
.00
1-
7.9
5±
1.2
5-
5.7
0±
0.9
3\
0.0
01
19
-7
.33
±0
.89
-7
.45
±1
.21
[0
.00
1*
-7
.26
±0
.86
-5
.42
±0
.84
\0
.00
1-
7.3
3±
0.8
9-
5.5
1±
0.8
8\
0.0
01
-7
.45
±1
.21
-5
.51
±0
.88
\0
.00
1
20
-7
.49
±0
.93
-7
.60
±1
.21
[0
.00
1*
-7
.41
±0
.90
-5
.50
±0
.86
\0
.00
1-
7.4
9±
0.9
3-
5.5
9±
0.8
9\
0.0
01
-7
.60
±1
.21
-5
.59
±0
.89
\0
.00
1
21
-6
.82
±0
.91
-6
.10
±1
.12
\0
.00
1-
6.6
4±
1.0
5-
4.4
7±
0.8
7\
0.0
01
-6
.82
±0
.91
-4
.60
±0
.98
\0
.00
1-
6.1
0±
1.1
2-
4.6
0±
0.9
8\
0.0
01
22
-6
.98
±0
.88
-6
.57
±1
.12
\0
.00
1-
6.8
4±
0.9
6-
4.8
0±
0.8
2\
0.0
01
-6
.98
±0
.88
-4
.90
±0
.91
\0
.00
1-
6.5
7±
1.1
2-
4.9
0±
0.9
1\
0.0
01
23
-7
.24
±0
.93
-7
.22
±1
.20
[0
.00
1*
-7
.16
±0
.94
-5
.31
±0
.83
\0
.00
1-
7.2
4±
0.9
3-
5.4
0±
0.8
8\
0.0
01
-7
.22
±1
.20
-5
.40
±0
.88
\0
.00
1
24
-7
.14
±0
.91
-7
.05
±1
.19
[0
.00
1*
-7
.05
±0
.92
-5
.21
±0
.82
\0
.00
1-
7.1
4±
0.9
1-
5.3
0±
0.8
7\
0.0
01
-7
.05
±1
.19
-5
.30
±0
.87
\0
.00
1
25
5.9
5±
-4
.93
-4
.37
±0
.99
\0
.00
1-
4.8
5±
0.8
2-
1.7
8±
0.7
7\
0.0
01
5.9
5±
-4
.93
-1
.99
±0
.80
\0
.00
1-
4.3
7±
0.9
9-
1.9
9±
0.8
0\
0.0
01
26
6.5
5±
-4
.77
-4
.35
±1
.02
\0
.00
1-
4.7
4±
0.7
8-
2.0
0±
0.6
4\
0.0
01
6.5
5±
-4
.77
-2
.18
±0
.63
\0
.00
1-
4.3
5±
1.0
2-
2.1
8±
0.6
3\
0.0
01
27
6.9
3±
-5
.58
-4
.55
±1
.17
\0
.00
1-
54
0±
1.1
0-
2.6
9±
0.8
6\
0.0
01
6.9
3±
-5
.58
-2
.87
±0
.93
\0
.00
1-
4.5
5±
1.1
7-
2.8
7±
0.9
3\
0.0
01
28
7.0
6±
-5
.01
-4
.40
±1
.08
\0
.00
1-
4.9
5±
0.8
7-
2.3
5±
0.6
7\
0.0
01
7.0
6±
-5
.01
-2
.52
±0
.66
\0
.00
1-
4.4
0±
1.0
8-
2.5
2±
0.6
6\
0.0
01
29
2.1
2±
-6
.72
-5
.77
±1
.15
\0
.00
1-
6.5
1±
1.1
3-
4.1
0±
1.0
0\
0.0
01
2.1
2±
-6
.72
-4
.24
±1
.13
\0
.00
1-
5.7
7±
1.1
5-
4.2
4±
1.1
3\
0.0
01
30
3.6
0±
-6
.49
-5
.29
±1
.20
\0
.00
1-
6.2
4±
1.2
5-
3.5
0±
1.1
9\
0.0
01
3.6
0±
-6
.49
-3
.67
±1
.36
\0
.00
1-
5.2
9±
1.2
0-
3.6
7±
1.3
6\
0.0
01
31
6.1
7±
6.1
1-
4.7
1±
1.2
1\
0.0
01
-5
.78
±1
.27
-3
.05
±1
.13
\0
.00
16
.17
±6
.11
-3
.26
±1
.27
\0
.00
1-
4.7
1±
1.2
1-
3.2
6±
1.2
7\
0.0
01
32
4.9
1±
5.0
8-
4.9
8±
1.2
2\
0.0
01
-6
.10
±1
.29
-3
.17
±1
.28
\0
.00
14
.91
±5
.08
-3
.39
±1
.46
\0
.00
1-
4.9
8±
1.2
2-
3.3
9±
1.4
6\
0.0
01
*In
sig
nifi
can
tfe
atu
res;
a,
(50
7n
orm
al?
50
7d
eaf)
;b,
(34
0n
orm
al?
34
0d
eaf)
Australas Phys Eng Sci Med (2014) 37:439–456 447
123
Ta
ble
5S
um
mar
yo
fst
atis
tics
of
extr
acte
den
tro
py
feat
ure
s(d
mey
)
Fea
ture
sN
orm
alA
sph
yx
iaP
val
ue
No
rmal
Dea
fP
val
uea
No
rmal
Dea
fP
val
ueb
Asp
hy
xia
Dea
fP
val
ue
11
3.0
0±
3.0
73
5.4
7±
18
.64
\0
.00
11
7.8
7±
11
.86
11
.38
±4
.75
\0
.00
11
3.0
0±
3.0
71
1.5
3±
4.6
8\
0.0
01
35
.47
±1
8.6
41
1.5
3±
4.6
8\
0.0
01
23
5.6
0±
14
.69
29
.99
±2
1.4
1\
0.0
01
39
.92
±1
5.0
72
2.9
9±
9.4
1\
0.0
01
35
.60
±1
4.6
92
3.6
3±
9.3
5\
0.0
01
29
.99
±2
1.4
12
3.6
3±
9.3
5\
0.0
01
33
7.8
4±
16
.67
27
.63
±1
7.7
2\
0.0
01
41
.56
±1
6.4
63
8.8
2±
18
.88
[0
.00
1*
37
.84
±1
6.6
73
8.6
8±
20
.39
[0
.00
1*
27
.63
±1
7.7
23
8.6
8±
20
.39
\0
.00
1
43
4.4
3±
15
.14
26
.25
±1
8.1
7\
0.0
01
37
.81
±1
6.0
72
7.6
6±
11
.88
\0
.00
13
4.4
3±
15
.14
29
.46
±1
2.0
9\
0.0
01
26
.25
±1
8.1
72
9.4
6±
12
.09
[0
.00
1*
52
0.0
6±
10
.46
16
.09
±1
1.5
8\
0.0
01
21
.06
±1
1.5
14
3.0
9±
13
.03
\0
.00
12
0.0
6±
10
.46
42
.49
±1
2.6
1\
0.0
01
16
.09
±1
1.5
84
2.4
9±
12
.61
\0
.00
1
61
6.3
6±
8.0
21
8.8
4±
14
.00
[0
.00
1*
18
.04
±1
0.4
53
7.8
3±
13
.58
\0
.00
11
6.3
6±
8.0
23
6.4
0±
12
.67
\0
.00
11
8.8
4±
14
.00
36
.40
±1
2.6
7\
0.0
01
72
4.5
3±
11
.60
25
.91
±1
7.4
8[
0.0
01
*2
7.5
8±
13
.72
39
.65
±1
4.8
3\
0.0
01
24
.53
±1
1.6
03
9.9
7±
13
.73
\0
.00
12
5.9
1±
17
.48
39
.97
±1
3.7
3\
0.0
01
81
7.6
9±
8.7
82
3.1
9±
17
.74
\0
.00
12
0.5
8±
11
.85
34
.23
±1
4.9
3\
0.0
01
17
.69
±8
.78
32
.38
±1
3.7
9\
0.0
01
23
.19
±1
7.7
43
2.3
8±
13
.79
\0
.00
1
98
.13
±4
.22
7.3
4±
5.9
9[
0.0
01
*8
.17
±6
.06
47
.68
±1
2.3
7\
0.0
01
8.1
3±
4.2
24
5.9
8±
11
.75
\0
.00
17
.34
±5
.99
45
.98
±1
1.7
5\
0.0
01
10
10
.72
±5
.53
7.8
7±
6.7
6\
0.0
01
10
.26
±7
.14
46
.10
±1
1.8
7\
0.0
01
10
.72
±5
.53
44
.16
±1
0.7
4\
0.0
01
7.8
7±
6.7
64
4.1
6±
10
.74
\0
.00
1
11
17
.34
±1
0.5
21
0.0
4±
8.6
9\
0.0
01
15
.93
±1
0.7
74
3.9
1±
16
.09
\0
.00
11
7.3
4±
10
.52
44
.52
±1
2.2
2\
0.0
01
10
.04
±8
.69
44
.52
±1
2.2
2\
0.0
01
12
12
.50
±6
.89
8.5
6±
7.5
8\
0.0
01
11
.70
±7
.65
44
.44
±1
2.1
4\
0.0
01
12
.50
±6
.89
43
.68
±1
0.8
7\
0.0
01
8.5
6±
7.5
84
3.6
8±
10
.87
\0
.00
1
13
20
.66
±1
0.2
01
5.5
5±
12
.51
\0
.00
12
1.6
0±
11
.62
45
.57
±1
2.4
0\
0.0
01
20
.66
±1
0.2
04
4.8
8±
11
.65
\0
.00
11
5.5
5±
12
.51
44
.88
±1
1.6
5\
0.0
01
14
21
.66
±1
1.3
01
4.3
3±
12
.88
\0
.00
12
1.4
9±
12
.05
45
.03
±1
3.7
4\
0.0
01
21
.66
±1
1.3
04
3.5
7±
13
.67
\0
.00
11
4.3
3±
12
.88
43
.57
±1
3.6
7\
0.0
01
15
20
.23
±1
2.5
31
1.5
1±
10
.08
\0
.00
11
8.3
6±
12
.24
43
.44
±1
5.3
4\
0.0
01
20
.23
±1
2.5
34
1.9
5±
15
.47
\0
.00
11
1.5
1±
10
.08
41
.95
±1
5.4
7\
0.0
01
16
22
.05
±1
2.4
01
2.9
7±
11
.37
\0
.00
12
0.9
8±
12
.78
43
.45
±1
3.3
6\
0.0
01
22
.05
±1
2.4
04
2.2
1±
12
.41
\0
.00
11
2.9
7±
11
.37
42
.21
±1
2.4
1\
0.0
01
17
0.2
9±
0.6
10
.14
±0
.16
\0
.00
10
.25
±0
.52
1.1
9±
0.9
7\
0.0
01
0.2
9±
0.6
11
.09
±0
.95
\0
.00
10
.14
±0
.16
1.0
9±
0.9
5\
0.0
01
18
0.3
4±
0.6
20
.13
±0
.12
\0
.00
10
.30
±0
.52
0.8
5±
0.5
8\
0.0
01
0.3
4±
0.6
20
.78
±0
.57
\0
.00
10
.13
±0
.12
0.7
8±
0.5
7\
0.0
01
19
0.4
3±
0.8
00
.21
±0
.18
\0
.00
10
.38
±0
.67
1.0
7±
0.6
5\
0.0
01
0.4
3±
0.8
01
.00
±0
.65
\0
.00
10
.21
±0
.18
1.0
0±
0.6
5\
0.0
01
20
0.3
9±
0.7
50
.19
±0
.16
\0
.00
10
.35
±0
.63
1.0
3±
0.6
6\
0.0
01
0.3
9±
0.7
50
.94
±0
.65
\0
.00
10
.19
±0
.16
0.9
4±
0.6
5\
0.0
01
21
0.5
8±
0.9
11
.33
±1
.33
\0
.00
10
.86
±1
.51
5.5
0±
4.1
4\
0.0
01
0.5
8±
0.9
15
.01
±4
.37
\0
.00
11
.33
±1
.33
5.0
1±
4.3
7\
0.0
01
22
0.4
6±
0.7
60
.74
±0
.67
\0
.00
10
.57
±0
.87
3.5
7±
2.7
0\
0.0
01
0.4
6±
0.7
63
.32
±2
.90
\0
.00
10
.74
±0
.67
3.3
2±
2.9
0\
0.0
01
23
0.4
3±
0.8
40
.25
±0
.22
\0
.00
10
.38
±0
.71
1.2
3±
0.7
6\
0.0
01
0.4
3±
0.8
41
.15
±0
.76
\0
.00
10
.25
±0
.22
1.1
5±
0.7
6\
0.0
01
24
0.4
3±
0.8
20
.32
±0
.27
[0
.00
1*
0.4
0±
0.7
11
.36
±0
.84
\0
.00
10
.43
±0
.82
1.2
7±
0.8
6\
0.0
01
0.3
2±
0.2
71
.27
±0
.86
\0
.00
1
25
3.2
4±
2.5
67
.22
±5
.95
\0
.00
14
.12
±5
.37
42
.03
±1
7.0
7\
0.0
01
3.2
4±
2.5
63
7.4
6±
16
.98
\0
.00
17
.22
±5
.95
37
.46
±1
6.9
8\
0.0
01
26
4.3
2±
2.5
47
.45
±6
.55
\0
.00
15
.07
±5
.55
38
.27
±1
5.2
4\
0.0
01
4.3
2±
2.5
43
3.9
5±
14
.09
\0
.00
17
.45
±6
.55
33
.95
±1
4.0
9\
0.0
01
27
2.3
0±
2.9
36
.88
±6
.93
\0
.00
13
.82
±6
.72
25
.93
±1
3.8
9\
0.0
01
2.3
0±
2.9
32
2.9
8±
14
.02
\0
.00
16
.88
±6
.93
22
.98
±1
4.0
2\
0.0
01
28
3.5
6±
2.7
17
.44
±7
.06
\0
.00
14
.57
±5
.92
31
.21
±1
3.7
8\
0.0
01
3.5
6±
2.7
12
7.4
4±
12
.41
\0
.00
17
.44
±7
.06
27
.44
±1
2.4
1\
0.0
01
29
0.6
7±
1.0
82
.08
±2
.12
\0
.00
11
.18
±2
.20
8.5
4±
6.6
4\
0.0
01
0.6
7±
1.0
87
.80
±6
.97
\0
.00
12
.08
±2
.12
7.8
0±
6.9
7\
0.0
01
30
0.9
0±
1.5
43
.51
±3
.60
\0
.00
11
.88
±3
.92
15
.80
±1
2.2
0\
0.0
01
0.9
0±
1.5
41
4.5
5±
12
.91
\0
.00
13
.51
±3
.60
14
.55
±1
2.9
1\
0.0
01
31
1.6
1±
2.8
06
.17
±6
.11
\0
.00
13
.20
±6
.50
21
.97
±1
4.7
4\
0.0
01
1.6
1±
2.8
01
9.6
7±
15
.56
\0
.00
16
.17
±6
.11
19
.67
±1
5.5
6\
0.0
01
32
0.9
9±
1.7
84
.91
±5
.08
\0
.00
12
.39
±5
.26
20
.87
±1
5.0
9\
0.0
01
0.9
9±
1.7
81
8.8
2±
16
.05
\0
.00
14
.91
±5
.08
18
.82
±1
6.0
5\
0.0
01
*In
sig
nifi
can
tfe
atu
res;
a(5
07
no
rmal
?5
07
dea
f);
b(3
40
no
rmal
?3
40
dea
f)
448 Australas Phys Eng Sci Med (2014) 37:439–456
123
TDNN classifiers are tabulated in Table 3. The signal
processing and classification algorithms are developed
under MATLAB environment [23].
Results and discussion
The following section briefly provides the empirical results
and discussion of the analysis, together with the details on
training and testing dataset segregation and statistical
analysis of extracted feature vectors. Due to the high
computational complexity in feature extraction process by
using higher order mother wavelets, only the results
obtained by using lower order mother wavelets were
presented.
Selection through similarity of mother wavelet with cry
signals
Figure 5 shows the comparative plot of correlation coeffi-
cients with different low and their respective higher index
mother wavelets for normal and pathological (asphyxia and
deaf) infant cry signals. The cross correlation results shown
in Fig. 5 are MATLAB [23] generated cross correlation
coefficient of one unit sample of cry signals with various
mother wavelet filters (available in MATLAB library).
From the Fig. 5 it was observed that, the correlation
coefficient values for the infant cry signals (asphyxia, deaf
and normal) when cross correlated with the ‘dmey’ mother
wavelet were greater (0.9) compared to other mother
wavelets including db20 which was used in [14] to achieve
99 % of maximum accuracy in discrimination of normal
and pathological cry signals. Hence, it was inferred that,
the ‘dmey’ mother wavelet is intimately similar or well
matched with the infant cry signals that are highly non
linear in nature. Results consolidate the aptness of ‘dmey’
mother wavelet as the best wavelet for the infant cry
classification. This result will further support our findings
with respect to selection of best mother wavelet.
Selection through regularity of mother wavelet-number
of significant features
A comparative analysis was performed to determine the
dispersion of significant and useful cry features after vali-
dation through independent t test (p \ 0.0001) from two
different datasets (normal vs asphyxia and normal vs deaf)
which were extracted using different wavelet families
(Fig. 6). As seen in Fig. 6, the ‘dmey’ wavelet family was
reported the maximum number of useful and significant
features in both cases, normal vs asphyxia (24) and normal vs
deaf (32) compared to other mother wavelets. Results
attested that the ‘dmey’ mother wavelet retained and pre-
served the original features of the cry signals with less loss of
salient information even after the fifth level of wavelet
packet decomposition algorithm. The good regularity prop-
erty of ‘dmey’ mother wavelet is also demonstrated in [41].
Tables 4 and 5 present the discriminatory ability of the
wavelet packet features (energy and entropy) which were
extracted from ‘dmey’ mother wavelet, in terms of mean,
standard deviation and p values through independent-
sample t test. The statistics of ‘dmey’ was selected and
tabulated, since it was outperformed in all the selection
methods compared to other mother wavelets. The p values
of different datasets such as normal vs asphyxia, normal vs
deaf (507 normal ? 507 deaf), normal vs deaf (340 nor-
mal ? 340 deaf) and asphyxia vs deaf were analyzed by
choosing 99.90 % of confidence interval. From Tables 4
and 5, it has been signified that the features extracted from
normal and pathological (asphyxia and deaf) cry signals are
almost differentiable, showed greater variation between
different groups and most of the features were statistically
significant (p \ 0.001). In the current work, the k-fold
cross validation (10-fold) or sometimes known as rotation
estimation and conventional validation schemes were used
to prove the reliability of the classification accuracy [42].
Table 6 shows, the segregation of infant cry samples for
training and testing of classification phases for the two
classification validation schemes.
Tables 7, 8, 9 highlight the simulation results of the
three different experiments using different supervised
classifiers. The maximum accuracy of 99.11 ± 0.18 %
(conventional validation, entropy, PNN) and 99.10 ±
0.22 % (10-fold cross validation, entropy, PNN) was
obtained from the ‘dmey’ mother wavelet as seen in
Table 7. From Table 7, it was found that, the maximum
accuracy of 97.28 ± 0.47 % (conventional validation,
Table 6 Training and testing datasets of 10-fold and conventional
validation schemes for three different experiments
Experiments Validation schemes
10-fold cross validation Conventional
(60 % training,
40 % testing)
Experiment 1
(340 normal ?
340 asphyxia)
Samples were segregated
randomly into 10 sets and
training was repeated for
10 times
Training = 408
samples
Testing = 272
samples
Experiment 2
(507 normal ?
507 deaf)
Samples were segregated
randomly into 10 sets and
training was repeated for
10 times
Training = 608
samples
Testing = 406
samples
Experiment 3
(340 normal ?
340 asphyxia ?
340 deaf)
Samples were segregated
randomly into 10 sets and
training was repeated for
10 times
Training = 612
samples
Testing = 408
samples
Australas Phys Eng Sci Med (2014) 37:439–456 449
123
Ta
ble
7R
esu
lts
of
PN
N,
GR
NN
,M
LP
and
TD
NN
for
exp
erim
ent
1(c
on
ven
tio
nal
val
idat
ion
and
10
-fo
ldcr
oss
val
idat
ion
)
Fea
ture
sV
alid
atio
n
typ
e
Cla
ssifi
ers
Mo
ther
Wav
elet
s
haa
rd
b2
sym
2co
if1
bio
r1.1
rbio
1.1
dm
ey
En
erg
yC
on
VP
NN
94
.45
±0
.27
95
.97
±0
.40
96
.33
±0
.27
96
.72
±0
.28
94
.79
±0
.51
94
.54
±0
.41
98
.64
±0
.22
GR
NN
89
.13
±2
.12
92
.73
±1
.50
92
.80
±1
.52
94
.43
±1
.08
89
.00
±2
.37
89
.29
±2
.05
98
.37
±0
.23
ML
P9
0.5
5±
2.1
09
2.5
4±
1.3
89
2.2
8±
1.6
99
4.3
4±
0.8
59
0.2
6±
2.8
59
0.0
7±
1.4
19
5.5
9±
1.5
1
TD
NN
91
.10
±2
.30
92
.87
±2
.19
92
.65
±2
.04
94
.56
±1
.63
90
.95
±1
.67
90
.77
±2
.14
96
.43
±0
.81
Cro
ssV
PN
N9
4.9
9±
0.4
49
6.6
3±
0.2
59
6.5
7±
0.3
59
7.1
2±
0.2
69
5.1
9±
0.4
69
5.1
6±
0.2
69
8.9
5±
0.0
8
GR
NN
89
.06
±2
.47
93
.31
±1
.80
93
.26
±1
.78
95
.00
±1
.35
89
.09
±2
.51
89
.10
±2
.59
98
.60
±0
.14
ML
P9
1.4
6±
0.7
49
3.6
7±
0.8
19
3.5
0±
0.8
99
5.1
3±
0.3
99
1.3
1±
1.0
59
1.7
2±
0.8
79
6.2
5±
0.5
1
TD
NN
91
.72
±0
.78
93
.57
±0
.75
93
.82
±0
.77
95
.75
±0
.43
92
.12
±0
.78
91
.93
±0
.73
96
.54
±0
.30
En
tro
py
Co
nV
PN
N9
5.0
4±
0.7
49
6.6
1±
0.6
69
6.3
3±
0.4
99
7.1
0±
0.2
99
4.7
9±
0.7
09
4.7
9±
0.8
79
9.1
1–
0.1
8
GR
NN
87
.07
±3
.14
91
.17
±2
.74
90
.79
±2
.93
92
.50
±2
.73
87
.50
±3
.43
87
.30
±3
.54
98
.10
±0
.69
ML
P9
2.6
8±
1.6
59
4.7
1±
1.4
69
4.5
2±
1.6
99
6.7
3±
1.1
09
2.4
6±
1.1
79
2.3
2±
1.7
89
6.5
4±
0.8
8
TD
NN
91
.99
±1
.86
94
.41
±1
.48
95
.00
±1
.41
96
.25
±1
.43
91
.88
±2
.18
92
.46
±1
.16
97
.28
±0
.47
Cro
ssV
PN
N9
5.1
9±
0.8
79
6.8
4±
0.5
89
6.8
8±
0.4
39
7.4
0±
0.4
29
5.2
1±
0.8
29
5.3
5±
0.8
39
9.1
0±
0.2
2
GR
NN
87
.41
±3
.41
91
.19
±2
.98
91
.10
±2
.95
92
.49
±2
.74
87
.34
±3
.30
87
.34
±3
.10
98
.38
±0
.51
ML
P9
3.7
2±
0.6
49
5.6
0±
0.3
39
5.4
4±
0.5
79
6.5
4±
0.4
19
3.4
2±
0.6
09
4.4
3±
0.7
59
7.1
5±
0.3
6
TD
NN
93
.51
±0
.73
95
.34
±0
.97
95
.79
±0
.35
96
.73
±0
.48
93
.81
±0
.54
93
.51
±0
.48
97
.19
±0
.44
Th
ev
alu
esin
bo
ldh
igh
lig
ht
the
max
imu
mo
bta
ined
accu
racy
450 Australas Phys Eng Sci Med (2014) 37:439–456
123
Ta
ble
8R
esu
lts
of
PN
N,
GR
NN
,M
LP
and
TD
NN
for
exp
erim
ent
2(c
on
ven
tio
nal
val
idat
ion
and
10
-fo
ldcr
oss
val
idat
ion
)
Fea
ture
sV
alid
atio
n
typ
e
Cla
ssifi
ers
Mo
ther
Wav
elet
s
haa
rd
b2
sym
2co
if1
bio
r1.1
rbio
1.1
dm
ey
En
erg
yC
on
VP
NN
99
.46
±0
.08
99
.63
±0
.05
99
.64
±0
.12
99
.37
±0
.13
99
.39
±0
.14
99
.36
±0
.11
99
.56
±0
.12
GR
NN
98
.76
±0
.38
99
.20
±0
.19
99
.21
±0
.18
99
.08
±0
.16
98
.74
±0
.38
98
.71
±0
.35
99
.31
±0
.38
ML
P9
6.9
5±
0.6
09
7.2
9±
1.0
49
6.9
2±
0.7
69
6.8
5±
1.3
49
6.7
7±
0.9
59
6.7
7±
0.9
59
7.2
4±
0.7
7
TD
NN
97
.32
±0
.92
97
.29
±0
.98
97
.19
±0
.85
97
.39
±0
.73
97
.49
±0
.91
97
.49
±0
.91
97
.68
±0
.82
Cro
ssV
PN
N9
9.6
1±
0.0
59
9.6
5±
0.0
89
9.7
3±
0.1
19
9.4
6±
0.0
89
9.5
6±
0.0
89
9.5
3±
0.1
19
9.8
0–
0.0
6
GR
NN
98
.94
±0
.43
99
.56
±0
.19
99
.47
±0
.18
99
.03
±0
.14
99
.00
±0
.41
98
.95
±0
.41
99
.52
±0
.14
ML
P9
7.4
6±
0.3
19
7.2
3±
0.3
59
7.2
3±
0.3
59
7.2
0±
0.4
19
7.4
3±
0.2
29
7.3
8±
0.2
49
7.7
8±
0.2
3
TD
NN
97
.64
±0
.23
97
.63
±0
.21
97
.62
±0
.35
97
.55
±0
.35
97
.75
±0
.34
97
.57
±0
.28
97
.80
±0
.37
En
tro
py
Co
nV
PN
N9
9.0
2±
0.1
89
9.2
7±
0.1
39
9.2
1±
0.1
39
9.1
7±
0.1
09
8.8
6±
0.1
89
8.9
7±
0.2
19
9.6
6±
0.0
8
GR
NN
97
.99
±0
.39
98
.79
±0
.18
98
.75
±0
.26
98
.60
±0
.19
97
.96
±0
.35
97
.99
±0
.39
98
.83
±1
.02
ML
P9
7.0
4±
0.7
19
7.1
7±
0.7
49
7.2
7±
1.0
79
7.0
4±
0.7
39
6.8
0±
0.8
69
6.9
9±
1.6
99
7.5
6±
0.7
4
TD
NN
97
.29
±1
.37
97
.44
±0
.50
97
.17
±0
.75
97
.19
±1
.00
97
.07
±1
.03
97
.19
±0
.78
98
.00
±0
.89
Cro
ssV
PN
N9
9.2
2±
0.1
09
9.3
9±
0.1
29
9.5
6±
0.0
89
9.4
2±
0.1
09
9.2
8±
0.1
39
9.2
1±
0.1
79
9.7
9±
0.0
9
GR
NN
98
.09
±0
.43
99
.04
±0
.16
98
.99
±0
.12
98
.84
±0
.13
98
.12
±0
.39
98
.10
±0
.45
99
.07
±0
.70
ML
P9
7.3
9±
0.2
09
7.3
0±
0.4
19
7.3
0±
0.4
19
7.5
7±
0.3
19
7.2
9±
0.4
39
7.2
5±
0.3
89
7.7
7±
0.2
4
TD
NN
97
.54
±0
.24
97
.59
±0
.23
97
.76
±0
.26
97
.69
±0
.16
97
.60
±0
.32
97
.54
±0
.24
97
.85
±0
.24
Th
ev
alu
esin
bo
ldh
igh
lig
ht
the
max
imu
mo
bta
ined
accu
racy
Australas Phys Eng Sci Med (2014) 37:439–456 451
123
Ta
ble
9R
esu
lts
of
PN
N,
GR
NN
,M
LP
and
TD
NN
for
exp
erim
ent
3(c
on
ven
tio
nal
val
idat
ion
and
10
-fo
ldcr
oss
val
idat
ion
)
Fea
ture
sV
alid
atio
nty
pe
Cla
ssifi
ers
Mo
ther
wav
elet
haa
rd
b2
sym
2co
if1
bio
r1.1
rbio
1.1
dm
ey
En
erg
yC
on
VP
NN
96
.21
±0
.31
96
.89
±0
.25
97
.02
±0
.37
97
.36
±0
.32
96
.11
±0
.35
96
.07
±0
.30
98
.82
±0
.14
GR
NN
95
.79
±0
.22
96
.81
±0
.24
96
.95
±0
.23
97
.24
±0
.27
96
.09
±0
.18
95
.99
±0
.34
98
.76
±0
.25
ML
P9
6.9
5±
0.7
99
7.6
2±
0.8
79
7.6
7±
1.0
79
7.9
9±
0.4
79
7.2
8±
0.9
19
7.2
1±
0.7
99
8.7
5±
0.4
7
TD
NN
97
.13
±0
.83
97
.98
±0
.72
97
.96
±0
.80
98
.01
±0
.61
96
.96
±0
.70
96
.86
±0
.68
98
.71
±0
.39
Cro
ssV
PN
N9
6.5
4±
0.2
49
7.3
5±
0.1
29
7.3
6±
0.1
99
7.9
1±
0.1
79
6.5
5±
0.1
19
6.4
5±
0.1
49
9.0
6±
0.1
1
GR
NN
96
.56
±0
.15
97
.25
±0
.20
97
.33
±0
.21
97
.75
±0
.17
96
.56
±0
.24
96
.47
±0
.22
99
.09
±0
.17
ML
P9
7.2
3±
0.3
59
7.1
5±
1.5
59
7.2
2±
1.5
59
7.5
1±
1.6
49
6.7
7±
1.8
59
7.2
4±
0.2
89
8.5
3±
1.0
7
TD
NN
97
.02
±0
.24
97
.25
±2
.19
97
.31
±1
.31
98
.08
±0
.13
96
.73
±1
.30
96
.73
±1
.32
98
.87
±0
.16
En
tro
py
Co
nV
PN
N9
5.7
0±
0.4
79
7.1
2±
0.4
59
7.0
1±
0.3
29
7.3
6±
0.4
09
5.6
8±
0.5
39
5.6
8±
0.5
39
8.9
9±
0.1
1
GR
NN
95
.46
±0
.21
96
.63
±0
.56
96
.60
±0
.37
96
.86
±0
.61
95
.44
±0
.35
95
.44
±0
.35
98
.68
±0
.51
ML
P9
7.6
0±
0.7
49
8.4
6±
0.4
39
8.3
8±
0.8
39
8.6
7±
0.6
49
7.5
5±
0.8
99
7.6
0±
0.5
59
8.6
3±
0.5
1
TD
NN
97
.55
±0
.71
98
.26
±0
.61
98
.17
±0
.76
98
.65
±0
.37
97
.55
±0
.61
97
.75
±0
.78
98
.60
±0
.43
Cro
ssV
PN
N9
5.9
5±
0.4
79
7.4
5±
0.3
69
7.3
4±
0.5
39
7.8
5±
0.4
09
5.8
6±
0.3
99
5.8
6±
0.3
99
9.2
0–
0.1
1
GR
NN
95
.81
±0
.39
97
.10
±0
.27
97
.13
±0
.35
97
.38
±0
.25
95
.75
±0
.41
95
.75
±0
.41
98
.84
±1
.12
ML
P9
7.7
1±
0.1
19
7.0
8±
1.5
89
7.5
3±
2.2
59
8.3
4±
0.8
09
7.5
9±
0.1
99
7.8
6±
0.5
39
8.5
0±
0.2
4
TD
NN
97
.26
±0
.13
97
.36
±2
.34
98
.43
±0
.30
98
.44
±1
.05
96
.71
±1
.48
97
.25
±1
.08
98
.17
±1
.15
Th
ev
alu
esin
bo
ldh
igh
lig
ht
the
max
imu
mo
bta
ined
accu
racy
452 Australas Phys Eng Sci Med (2014) 37:439–456
123
entropy, TDNN) and 97.19 ± 0.44 % (10-fold cross vali-
dation, entropy, TDNN) from the ‘dmey’ was indexed. As
seen in Table 8, the highest accuracy of 99.66 ± 0.08 %
(conventional validation, entropy, PNN) and 99.80 ±
0.06 % (10-fold cross validation, energy, PNN) was
obtained from the Meyer’s mother wavelet. From Table 8,
the maximum accuracy of 98.00 ± 0.89 % (conventional
validation, entropy, TDNN) and 97.85 ± 0.24 % (10-fold
cross validation, entropy, TDNN) from ‘dmey’ mother
wavelet was registered. From the Table 9, it was perceived
that the best results by using PNN, GRNN, MLP and
TDNN classifiers were attained from the ‘dmey’ mother
wavelet. The maximum classification accuracy of 98.99 ±
0.11 % (conventional validation, entropy, PNN) and
99.20 ± 0.11 % (10-fold cross validation, entropy, PNN)
was accounted. The highest discrimination accuracy of
98.75 ± 0.47 % (conventional validation, energy, MLP)
and 98.87 ± 0.16 % (10-fold cross validation, energy,
TDNN) was obtained as seen in Table 9.
The performance of the PNN, GRNN, MLP and TDNN
classifiers proven the robustness and significance of
wavelet packet based features which were extracted from
fifth level decomposition for maximum discrimination of
different types of infant cry signals. Figures 7 and 8,
present the comparative performance of classifiers for the
three different experiments using ‘dmey’ mother wavelet
and it has been observed that, PNN and GRNN were out-
performed MLP and TDNN networks. Based on the sim-
ulation results above, it has been proven that the ‘dmey’
mother wavelet was the most appropriate mother wavelet
compared to other tested types of mother wavelets.
Table 10 presents the performance comparison of the
proposed study and other existing classification works.
Hariharan et al. [14] discriminated the normal and asphyxia
cry signals with best accuracy of 99 % by implementing
WPT and PNN. However in their work only the perfor-
mance of different orders of Daubechies mother wavelet
was investigated and the best accuracy was from a higher
order of wavelet (db20). In [15] the authors have shown the
effectiveness of their proposed approaches, MFCC and
GSFM which designed with an optimal combination of
feature selection method, fuzzy processing type and
learning algorithm by reporting maximum accuracy of
90.68 % using 10-fold cross validation scheme. However,
in our work, we achieved a maximum accuracy of 99 %
using both conventional and 10-fold cross validation
schemes. Best recognition rate of 99 % reported by using
time frequency based statistical features which were
derived from normal and asphyxia cry signals, PCA and
PNN [16]. Hariharan et al. [17] investigated the use of
Fig. 7 Comparative
performance of different
classifiers through conventional
validation scheme
Fig. 8 Comparative
performance of different
classifiers through 10-fold cross
validation scheme
Australas Phys Eng Sci Med (2014) 37:439–456 453
123
GRNN to discriminate the normal and deaf and achieved
99 % of diagnosis accuracy using statistical features which
derived from time frequency plots. Proposed GSFM which
trained and tested with an optimal combination of feature
selection method, fuzzy processing type and learning
algorithm evaluated using extracted MFCC feature vectors
of normal and deaf cries and acquired 99.42 % of maxi-
mum classification accuracy [15]. Hariharan et al. [9]
developed an infant cry based multi class automated sys-
tem using WLPCC for signal processing and PNN for
classification process, with optimal classification result of
99 %.
In this proposed study, by emphasizing the time fre-
quency based approach a best mother wavelet (‘dmey’) for
maximum infant cry recognition was selected using sta-
tistically validated wavelet packet based feature vectors
and supervised neural networks through a combination of
different selection criteria. Maximum recognition accuracy
of above 99 % was reported for all the experiments which
are comparable or similar with most of the literature works
(Table 10). However, our proposed system seem superior
compared to other literature works, since it was designed as
a single unit of block system to tackle the binary as well as
the multi class recognition tasks mutually and successfully
which not attempted yet in other existing systems espe-
cially using the normal, asphyxia and deaf infant cry sig-
nals. Though, recently numerous automated classification
systems proposed and developed in infant cry classification
area only a few multiclass based recognition systems was
documented with higher successful recognition rates
around 99 % which is sufficient enough for implementation
in clinical trials even in the cases up to three class domain
problems (Table 10). Hence, it can be deduced from the
study results, that our proposed system maybe significant in
terms of clinical applications for the improved diagnostic
results. In addition, the proposed methodologies especially
Table 10 A performance
comparison of the proposed
methodology and other
automated infant cry
classification studies
Studies Signal processing
method
Classifier Accuracy (%)
Normal and asphyxia cry (experiment 1)
Hariharan et al. [14] WPT (only Daubechies
mother wavelet was
considered)
PNN 99 (60 % training, 40 %
testing)
Rosales-Perez et al. [15] MFCC Genetic selection of
fuzzy model
90.68 (10-fold)
Hariharan et al. [16] Time–frequency
analysis based
statistical
features ? PCA
PNN and GRNN 99.19 (10-fold)
98.88 (60 % training,
40 % testing)
Proposed methodology WPT (7 types of
different mother
wavelet was
considered)
PNN and GRNN 99.10 (10-fold)
99.11 (60 % training,
40 % testing)
Normal and deaf cry (experiment 2)
Hariharan et al. [17] Time–frequency
analysis based
statistical features
GRNN 99.31 (10-fold)
93.90 (data
independent, 670
segments for training
and 344 segments for
testing)
Rosales-Perez et al. [15] MFCC Genetic selection of
fuzzy model
99.42 (10-fold)
Proposed methodology WPT (7 types of
different mother
wavelet was
considered)
PNN and GRNN 99.80 (10-fold)
99.66 (60 % training,
40 % testing)
Normal, asphyxia and deaf cry (experiment 3)
Hariharan et al. [9] WLPCC PNN 99 (70 % training, 30 %
testing)
Proposed methodology WPT (7 types of
different mother
wavelet was
considered)
PNN and GRNN 99.20 (10-fold)
98.99 (60 % training,
40 % testing)
454 Australas Phys Eng Sci Med (2014) 37:439–456
123
cry features (sub band energies and entropies) which con-
tributed for the maximum efficacy of our proposed system
due to their good discriminatory ability maybe used or
adopted by the medical professionals for diagnosing the
pathological status of infants based on their experience
using cry signals.
Conclusion
Infant cry is a good indicator of expressing infants’ phys-
ical and physiological status. Recently, infant cry has
attracted great intention of research community to explore
and move towards for evolving cry based automatic clas-
sification algorithms by adopting different digital signal
processing and artificial intelligent techniques.
In the development of this automated classification
systems, this study concentrated on the time frequency
based technique, mainly emphasizing on the selection of
best mother wavelet among 56 different basis of wavelets
for infant cry classification by incorporating the wavelet
packet transform, decomposition of infant cry signals at
best decomposition level and classification using super-
vised neural networks. Similarity of mother wavelets,
regularity of mother wavelets and simulation results were
considered as the selection criteria to select the best
mother wavelet. Based on these different selection criteria
results, it was inferred that the Meyer’s wavelet (‘dmey’)
is the best candidate among other mother wavelets (haar,
daubechies, symlet, coiflet, biorthogonal and reverse bi-
orthogonal) for the accurate neonates cry signal classifi-
cation, since it was harmonized well with the normal and
pathological cry signals mutually with the highest cross
coefficients, exhibited higher regularity by reporting
maximum number of significant features for two different
datasets and yielded maximum empirical results in all
three experiments.
In future, the proposed study will be extended to
investigate the other types of infant cry signals, to test with
different pathological cry databases, and to study the
severity levels of the disordered cry signals (mild, mod-
erate and severe). A comparison of the present work with
other time frequency based approaches for example Wig-
ner-ville, Choi William and extra will be tackled out.
Acknowledgments The Baby Chillanto Data Base is a property of
the Instituto Nacional de Astrofisica Optica y Electronica –CONA-
CYT, Mexico. We like to thank Dr. Carlos A. Reyes-Garcia, Dr.
Emilio Arch-Tirado and his INR-Mexico group, and Dr. Edgar M.
Garcia-Tamayo for their dedication of the collection of the Infant Cry
data base. The authors would like to thank Dr. Carlos Alberto Reyes-
Garcia, Researcher, CCC-Inaoep, Mexico for providing infant cry
database. All authors declare that they have no financial or any
commercial conflicts of interest.
References
1. Saraswathy J, Hariharan M, Yaacob S, Khairunizam W (2012)
Automatic classification of infant cry: a review. In: Proceedings
of the 2012 international conference on biomedical engineering
(ICOBE-2012), pp 543–548
2. Lederman D (2002) Automatic classification of infants’ cry.
Dissertation, University of Negev, Department of Electrical and
Computer Engineering
3. Boukydis CFZ, Lester BM (1985) Infant crying. Plenum press,
New York
4. Lederman D, Cohen A, Zmora E, Wermke K, Hauschildt S,
Stellzig-Eisenhauer A (2008) Classification of cries of infants
with cleft-palate using parallel hidden Markov models. Med Biol
Eng Comput 46:965–975
5. Michelsson K, Kaskinn H, Aulanko R, Rinne A (1984) Sound
spectrographic cry analysis of infants with hydrocephalus. Acta
Paediatr Scand 73:65–68
6. Colton RH, Steinschneider A (1981) The cry characteristics of an
infant who died of sudden infant death syndrome. J Speech Hear
Dis 46:359–363
7. Ruiz Diaz MA, Reyes Garcia CA, Altamirano Robles LC, Xal-
tena Altamirano JE (2012) Automatic infant cry analysis for the
identification of qualitative features to help opportune diagnosis.
Biomed Signal Process Control 7:43–49
8. Verduzco-Mendoza A, Arch-Tirado E, Reyes-Garcia CA, Ley-
bon-Ibarra J, Licona-Bonilla J (2012) Spectrographic infant cry
analysis in newborns with profound hearing loss and perinatal
high-risk newborns. Cir Cir 80:3–10
9. Hariharan M, Sin Chee L, Yaacob S (2012) Analysis of infant cry
through weighted linear prediction cepstral coefficients and
probabilistic neural network. J Med Syst 36:1309–1315
10. Office of Health and Nutrition CFP, Health and Nutrition, Bureau
for Global Programs Field Support and Research, U.S. (2004)
Detecting and treating newborn asphyxia, maternal neonatal &
health, pp 1–2. http://pdf.usaid.gov/pdf_docs/Pnacy993.pdf
Accessed 18 June 2012
11. Cunningham M, Edward O (2003) Hearing assessment in infants
and children: recommendations beyond neonatal screening. Am
Acad Pediatr Clin Rep 111:436–440
12. Rosales-Perez A, Reyes-Garcia CA, Gomez-Gil P (2011) Genetic
fuzzy relational neural network for infant cry classification. Lect
Notes Comput Sci 6718:288–296
13. Zabidi A, Mansor W, Khuan LY, Yassin IM, Sahak R (2011)
Binary particle swarm optimization for selection of features in the
recognition of infants cries with asphyxia. In: Proceedings of the
IEEE international colloquium on signal processing and its
applications, pp 272–276
14. Hariharan M, Yaacob S, Awang SA (2011) Pathological infant
cry analysis using wavelet packet transform and probabilistic
neural network. Exp Syst Appl 38:15377–15382
15. Rosales-Perez A, Reyes-Garcia CA, Gonzalez JA, Arch-Tirado E
(2012) Infant cry classification using genetic selection of a fuzzy
model. Lect Notes Comput Sci 7441:212–219
16. Hariharan M, Saraswathy J, Sindhu R, Wan Khairunizam, Yaa-
cob S (2012) Infant cry classification to identify asphyxia using
time-frequency analysis and radial basis neural networks. Exp
Syst Appl 39:9515–9523
17. Hariharan M, Sindhu R, Sazali Yaacob (2012) Normal and
hypoacoustic infant cry signal using time-frequency analysis and
general regression neural network. Comput Methods Programs
Biomed 108:559–569
18. http://ingenieria.uatx.mx/orionfrg/cry/. Accessed 7 May 2010
19. Reyes-Galaviz OF, Cano-Ortiz S, Reyes-Garcia C, y Electronica
CO, Puebla M (2009) Evolutionary-neural system to classify
Australas Phys Eng Sci Med (2014) 37:439–456 455
123
infant cry units for pathologies identification in recently born
babies. In: Proceedings of the 8th Mexican international confer-
ence on artificial intelligent, pp 330–335
20. Verduzco-Mendoza A, Arch-Tirado E, Reyes-Garcia CA, Leybon-
Ibarra J, Licona-Bonilla J (2009) Qualitative and quantitative crying
analysis of new born babies delivered under high risk gestation.
Multimodal Signals Lect Notes Artif Intell 5398:320–327
21. Singh BN, Tiwari AK (2006) Optimal selection of wavelet basis
function applied to ECG signal denoising. Digit Signal Process
16:275–287
22. Kumari A, Bisht M (2013) Optimal wavelet filter maximizes the
cross correlation coefficient with an ECG signal. Int J Innov
Technol Res 1(2):191–193
23. Matlab� Documentation, version 7.0, Release 14, 2004. The
MathWorks, Inc
24. Kale VN, Khalsa NN (2010) Performance evaluation of various
wavelets for image compression of natural and artificial images.
Int J Comput Sci Commun 1(1):179–184
25. Goudarzi MM, Taheri A, Pooyan M, Mahboobi R (2006) Mul-
tiwavelet and biological signal processing. Int J Inf Technol
2(4):264–272
26. Martis RJ, Acharya UR, Ray AK, Chakraborty C (2011) Appli-
cation of higher order cumulants to ECG signals for the cardiac
health diagnosis. In: Proceedings of the 33rd international con-
ference on IEEE EMBS, pp 1697–1700
27. Gandhi T, Panigrahi BK, Anand S (2011) A comparative study of
wavelet families for EEG signal classification. Neurocomputing
74:3051–3057
28. Nagaria B, Hashmi F, Dhakad P (2011) Comparative analysis of fast
wavelet transform for image compression for optimal image quality
and higher compression ratio. Int J Eng Sci Technol 3:4014–4019
29. Chaudhari A, Chaudhary P, Cheeran AN, Aswani Y (2012)
Improving signal to noise ratio of low-dose CT image using
wavelet transform. Int J Comput Sci Eng 4:779–789
30. Hariharan M, Fook CY, Sindhu R, Ilias B, Yaacob S (2012) A
comparative study of wavelet families for classification of wrist
motions. Comput Electr Eng 38:1798–1807
31. Murugesapandian P, Yaacob S, Hariharan M (2008) Feature
extraction based on Mel-scaled wavelet packet transform for the
diagnosis of voice disorders. Proc IFMBE 2008 21:790–793
32. Burrus CS, Ramesh A, Gopinath, Guo H (1998) Introduction to
wavelets and wavelet transforms: a primer. Prentice Hall,
Englewood Cliffs
33. Gopinath B, Shanthi N (2013) Computer-aided diagnosis system
for classifying benign and malignant thyroid nodules in multi-
stained FNAB cytological images. Australas Phys Eng Sci Med.
doi:10.1007/s13246-013-0199-8
34. Wang Y, Yang Z, Hao D, Zhang S, Yang Y, Zeng Y (2013)
Measurement of subcutaneous adipose tissue thickness by near-
infrared. Australas Phys Eng Sci Med. doi:10.1007/s13246-013-
0196-y
35. Mahmoodian H (2012) Predicting the continuous values of breast
cancer relapse time by type-2 fuzzy logic system. Australas Phys
Eng Sci Med 35:193–204
36. Hariharan M, Paulraj MP, Yaccob S (2010) Time-domain fea-
tures and probabilistic neural network for the detection of vocal
fold pathology. Malays J Comput Sci 23:60–67
37. Hariharan M, Paulraj MP, Yaacob S (2011) Detection of vocal
fold paralysis and oedema using time-domain features and
probabilistic neural network. Int J Biomed Eng Technol 6:46–57
38. Specht DF (1990) Probabilistic neural networks. IEEE Trans
Neural Netw 3:109–118
39. Specht DF (1991) A general regression neural network. IEEE
Trans Neural Netw 2:568–576
40. Heaton Jeff (2008) Introduction to neural networks for Java, 2nd
edn. Heaton Research, St. Louis
41. Cohen I (2001) Enhancement of speech using bark-scaled
wavelet packet decomposition. In: Proceedings of the 7th Euro-
pean conference speech, communication and technology, 2nd
INTERSPEECH Event, Aalborg, Denmark, Sept 3–7
42. Kohavi R (1995) A study of cross-validation and bootstrap for
accuracy estimation and model selection. In: Proceedings of the
14th international joint conference on artificial intelligence, vol 2,
Montreal, Quebec, Canada, pp 1137–1143
456 Australas Phys Eng Sci Med (2014) 37:439–456
123