Hindawi Publishing Corporation

Computational Intelligence and Neuroscience

Volume 2011, Article ID 406391, 7 pages
doi:10.1155/2011/406391

Research Article
PyEEG: An Open Source Python Module for
EEG/MEG Feature Extraction

Forrest Sheng Bao,1 Xin Liu,2 and Christina Zhang3

1
Department of Computer Science, Department of Electrical Engineering, Texas Tech University, Lubbock TX 79409-3104, USA
2 ECHO Labs, Nanjing University of Posts and Telecommunications, Nanjing 210003, China
3
Department of Physiology, McGill University, Montreal, QC, Canada H3G 1Y6

Correspondence should be addressed to Forrest Sheng Bao, forrest.bao@gmail.com

Received 31 August 2010; Revised 26 October 2010; Accepted 31 December 2010

Academic Editor: Sylvain Baillet

Copyright © 2011 Forrest Sheng Bao et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.

Computer-aided diagnosis of neural diseases from EEG signals (or other physiological signals that can be treated as time series,
e.g., MEG) is an emerging field that has gained much attention in past years. Extracting features is a key component in the analysis
of EEG signals. In our previous works, we have implemented many EEG feature extraction functions in the Python programming
language. As Python is gaining more ground in scientific computing, an open source Python module for extracting EEG features
has the potential to save much time for computational neuroscientists. In this paper, we introduce PyEEG, an open source Python
module for EEG feature extraction.

1. Introduction fractal dimensions from computational geometry, entropies

from information theory, and so forth. An open source tool
Computer-aided diagnosis based on EEG has become possi- that can extract EEG features would benefit the computa-
ble in the last decade for several neurological diseases such as tional neuroscience community since feature extraction is
Alzheimer’s disease [1, 2] and epilepsy [3, 4]. Implemented repeatedly invoked in the analysis of EEG signals. Because
systems can be very useful in the early diagnosis of those of Python’s increasing popularity in scientific computing,
diseases. For example, traditional epilepsy diagnosis may and especially in computational neuroscience, a Python
require trained physicians to visually screen lengthy EEG module for EEG feature extraction would be highly useful.
records whereas computer-aided systems can shorten this In response, we have developed PyEEG, a Python module for
time-consuming procedure by detecting and picking out EEG feature extraction, and have tested it in our previous
EEG segments of interest to physicians [5, 6]. On top of that, epileptic EEG research [3, 8, 11].
computers can extend our ability to analyze signals. Recently, Compared to other popular programming languages in
researchers have developed systems [3, 4, 7, 8] that can scientific computing such as C++ or MATLAB, Python is
hopefully use (any) random interictal (i.e., non-seizure) EEG an open source scripting language of simple syntax and
records for epilepsy diagnosis in instances that are difficult various high-level libraries (for detailed advantages of
for physicians to make diagnostic decisions with their naked Python, read http://www.python.org/about/), such as Scipy
eyes. In addition to analyzing existing signals, this computer- (http://www.scipy.org/) which allows users to run MATLAB
based approach can help us model the brain and predict codes after slight modification. There have been several pop-
future signals, for example, seizure prediction [9, 10]. ular open source Python projects in the neuroimaging com-
All the above systems rely on characterizing the EEG munity already, such as NIPY (http://nipy.sourceforge.net/).
signal into certain features, a step known as feature extrac- However, in neural physiology community, Python is not yet
tion. EEG features can come from different fields that study quite popular. As we are not aware of any open source tools in
time series: power spectral density from signal processing, Python (or other programming languages) that can extract
2 Computational Intelligence and Neuroscience

EEG features as mentioned above, we introduce and release

PyEEG in this paper. PyEEG
Though originally designed for EEG, PyEEG can also be
used to analyze other physiological signals that can be treated Non feature
as time series, especially MEG signals that represent the extraction functions
magnetic fields induced by currents of neural electrical
activities. EEG series
The rest of the paper is organized as follows. In Section 2,
we introduce the framework of PyEEG. Section 3 gives the Feature extraction
definitions to compute EEG features. A tutorial of applying functions
PyEEG onto a public real EEG dataset is given in Section 4.
Section 5 concludes the paper.

2. Main Framework
PyEEG’s target users are programmers (anyone who writes Feature values
programs) working on computational neuroscience. Figure 1
shows its framework. PyEEG is a Python module that focuses
only on extracting features from EEG/MEG segments. There- Figure 1: PyEEG framework.
fore, it does not contain functions to import data of various
formats or export features to a classifier. This is due to the
modularity and composition principles of building open 3. Supported Feature Extraction
source software which indicate that small programs that can
work well together via simple interfaces are better than big In this section, we detail the definitions and computation
monolithic programs. Since open source tools like EEG/MEG procedures to extract EEG features (as shown in Table 1)
data importers (e.g., EEGLab, Biosig, etc.) and classifier in PyEEG. Since there are many parameters and various
front-ends are already available, there is no need for us to algorithms for one feature, the numerical value of a feature
reinvent the wheel. Users can easily hook PyEEG up with extracted by PyEEG may be different from that extracted by
various existing open source software to build toolchains for other toolboxes. Users may need to adjust our code or use
their EEG/MEG research. non-default values for the parameters in order to meet their
PyEEG consists of two sets of functions. needs. Please note that the index of an array or a vector starts
from 1 rather than 0 in this section.
(1) Preprocessing functions, which do not return any
feature values. Only two such functions have been 3.1. Power Spectral Intensity and Relative Intensity Ratio.
implemented so far.     builds embedding To a time series [x1 , x2 , . . . , xN ], denote its Fast Fourier
sequence (from given lag and embedding dimen- Transform (FFT) result as [X1 , X2 , . . . , XN ]. A continuous
sion) and
   
 computes first- frequency band from flow to fup is sliced into K bins,
order differential sequence. One can build differential which can be of equal width or not. Boundaries of bins are
sequences of higher orders by repeatedly applying specified by a vector band = [ f1 , f2 , . . . , fK ], such that the
first-order differential computing. lower and upper frequencies of the ith bin are fi and fi+1 ,
respectively. Commonly used unequal bins are EEG/MEG
(2) Feature extraction functions, that return feature val- rhythms, which are, δ (0.5–4 Hz), θ (4–7 Hz), α (8–12 Hz),
ues. These are listed in Table 1. β (12–30 Hz), and γ (30–100 Hz). For these bins, we have
band = [0.5, 4, 7, 12, 30, 100].
The Power Spectral Intensity (PSI) [12] of the kth bin is
PyEEG only uses functions in standard Python library
evaluated as
and SciPy, the de facto Python module for scientific com-
puting. PyEEG does not define any new data structure, k+1 / fs )
N( f
but instead uses only standard Python and NumPy data PSIk = |Xi |, k = 1, 2, . . . , K − 1, (1)
structures. The reason is that we want to simplify the use i= N( fk / fs )
of PyEEG, especially for users without much program-
ming background. The inputs of all functions are a time where fs is the sampling rate, and N is the series length.
sequence as a list of floating-point numbers and a set Relative Intensity Ratio (RIR) [12] is defined on top of
of optional feature extraction parameters. Parameters have PSI
default values. The output of a feature extraction function is
a floating-point number if the feature is a scalar or a list of PSI j
RIR j = K −1 , j = 1, 2, . . . , K − 1. (2)
floating-point numbers (a vector) otherwise. Details about k=1 PSIk
functions are available in the PyEEG reference guide at
http://PyEEG.SourceForge.net/. PSI and RIR are both vector features.
Computational Intelligence and Neuroscience 3

Table 1: PyEEG-supported features and extraction functions with their return types.

Feature name Function name Return type

Power Spectral Intensity (PSI) and Relative Intensity Ratio (RIR)  
Two 1-D vectors
Petrosian Fractal Dimension (PFD) 
A scalar
Higuchi Fractal Dimension (HFD) 
A scalar
Hjorth mobility and complexity 
Two scalars
Spectral Entropy (Shannon’s entropy of RIRs)  
A scalar
SVD Entropy  
A scalar
Fisher Information   
A scalar
Approximate Entropy (ApEn)  
A scalar
Detrended Fluctuation Analysis (DFA) 
A scalar
Hurst Exponent (Hurst) 
A scalar

3.2. Petrosian Fractal Dimension (PFD). To a time series, 3.6. SVD Entropy. Reference [17] defines an entropy mea-
PFD is defined as sure using Singular Value Decomposition (SVD). Let the
log10 N input signal be [x1 , x2 , . . . , xN ]. We construct delay vectors as
PFD = , (3)  
log10 N + log10 (N/(N + 0.4Nδ )) y(i) = xi , xi+τ , . . . , xi+(dE −1)τ , (7)
where N is the series length, and Nδ is the number of sign
where τ is the delay and dE is the embedding dimension. In
changes in the signal derivative [13]. PFD is a scalar feature.
this paper, dE = 20 and τ = 2. The embedding space is then
constructed by
3.3. Higuchi Fractal Dimension (HFD). Higuchi’s algorithm
[14] constructs k new series from the original series  T
Y = y(1), y(2), . . . , y(N − (dE − 1)τ) . (8)
[x1 , x2 , . . . , xN ] by
The SVD is then performed on matrix Y to produce M
xm , xm+k , xm+2k , . . . , xm+(N −m)/kk , (4)
singular values, σ1 , . . . , σM , known as the singular spectrum.
where m = 1, 2, . . . , k. The SVD entropy is then defined as
For each time series constructed from (4), the length
L(m, k) is computed by 
M
HSVD = − σ i log2 σ i , (9)
 (N −m)/k 

 i=1
i=2 xm+ik − xm+(i−1)k (N − 1)
L(m, k) = . (5)
(N − m)/kk where M is the number of singular values and σ 1 , . . . , σ M are

 normalized singular values such that σ i = σi / M j =1 σ j . SVD
The average length is computed as L(k) = [ ki=1 L(i, k)]/k. entropy is a scalar feature.
This procedure repeats kmax times for each k from 1 to
kmax , and then uses a least-square method to determine the
slope of the line that best fits the curve of ln(L(k)) versus 3.7. Fisher Information. The Fisher information [18] can be
ln(1/k). The slope is the Higuchi Fractal Dimension. HFD is defined in normalized singular spectrum used in (9)
a scalar feature. −1

M
(σ i+1 − σ i )2
I= . (10)
3.4. Hjorth Parameters. To a time series [x1 , x2 , . . . , xN ], i=1
σi
the Hjorth mobility
√
and complexity

[15] are, respectively,
Fisher information is a scalar feature.
defined as M2/TP and (M4 · TP)/(M2 · M2), where
  
TP = xi /N , M2 = di /N , M4 = (di − di−1 )2 /N , and
3.8. Approximate Entropy. Approximate entropy (ApEn) is
di = xi − xi−1 . Hjorth mobility and complexity are both scalar a statistical parameter to quantify the regularity of a time
features. series [19].
ApEn is computed by the following steps.
3.5. Spectral Entropy. The spectral entropy [16] is defined as
follows (1) Let the input signal be [x1 , x2 , . . . , xN ].
(2) Build subsequence x(i, m) = [xi , xi+1 , . . . , xi+m−1 ] for
1 
K
H=− RIRi log RIRi , (6) 1 ≤ i ≤ N − m, where m is the length of the
log(K) i=1 subsequence. In [7], m = 1, 2, or 3.
where RIRi and K are defined in (2). Spectral entropy is (3) Let r represent the noise filter level, defined as r =
a scalar feature. k × SD for k = 0, 0.1, 0.2, . . . , 0.9.
4 Computational Intelligence and Neuroscience

(4) Build a set of subsequences {x( j, m)} = {x( j, m) | 4. Using PyEEG on Real Data
j ∈ [1..N − m]}, where x( j, m) is defined in step 2.
In this section, we use PyEEG on a real EEG dataset to dem-
(5) For each x(i, m) ∈ {x( j, m)}, compute
onstrate its use in everyday research.
N −m The dataset (http://epileptologie-bonn.de/cms/front
kj
j =1
C(i, m) = , (11) content.php?idcat=193&lang=3), from Klinik für Epilep-
N −m tologie, Universität Bonn, Germany [22], has been widely
where used in previous epilepsy research. In total, there are five
⎧    sets, each containing 100 single-channel EEG segments.
⎨1 if x(i, m) − x j, m  < r,
kj = ⎩ (12) Each segment has 4096 samples. Data in sets A and B are
0 otherwise. extracranial EEGs from 5 healthy volunteers with eyes open
and eyes closed, respectively. Sets C and D are intracranial
(6) data over interictal periods while Set E over ictal periods.
⎡ ⎤ Segments in D are from within the epileptogenic zone,
−m
N
1 ⎣ C(i, m) ⎦ while those in C are from the hippocampal formation of
ApEn(m, r, N ) = ln . (13)
N −M i=1
C(i, m + 1) the opposite hemisphere of the brain. Sets C, D, and E are
composed from EEGs of 5 patients. The data had a spectral
ApEn is a scalar feature. bandwidth of 0.5–85 Hz. Please refer to [22] for more details.
Using PyEEG is like using any other Python module.
3.9. Detrended Fluctuation Analysis. Detrended Fluctuation Users simply need to import PyEEG and then call its
Analysis (DFA) is proposed in [20]. functions as needed. PyEEG is provided as a single Python
The procedures to compute DFA of a time series file. Therefore, it only needs to be downloaded and placed
[x1 , x2 , . . . , xN ] are as follows. under a directory on Python module search paths, such as the
working directory. Alternatively,  environment
(1) First integrate x into a new series y = [y(1), . . . , variable can be set to point to the location of PyEEG.

y(N )], where y(k) = ki=1 (xi − x) and x is the average On Python interpreter, we first import PyEEG and load
of x1 , x2 , . . . , xN . the data
(2) The integrated series is then sliced into boxes of equal
>>>
   
length n. In each box of length n, a least-squares line is
fit to the data, representing the trend in that box. The >>>
    
y coordinate of the straight line segments is denoted >>>  
 !"
 
by yn (k).
>>> !!  # " !$  $
 %
(3) The root-mean-square fluctuation of the
integrated series is calculated by F(n)
 = where  is the first segment in set A. The data type
 of !! is "
. After loading EEG data, we can use PyEEG
(1/N ) Nk=1 [y(k) − yn (k)]2 , where the part
y(k) − yn (k) is called detrending. to extract features as follows (using all default parameters):
(4) The fluctuation can be defined as the slope of the line >>> &'    !!!
relating log F(n) to log n.
>>> &'
DFA is a scalar feature.
()**+,-)(+-.*)
3.10. Hurst Exponent. The hurst exponent (HURST) [21] is >>> / 0     1/!!
also called Rescaled Range statistics (R/S). To calculate the >>> / 0  
hurst exponent for time series X = [x1 , x2 , . . . , xN ], the first
,(*..+(++),2*
step is to calculate the accumulated deviation from the mean
of time series within range T >>> '&    !!


>>> '&
1
t T
X(t, T) = (xi − x), where x = xi , t ∈ [1..N ]. *(,*(.+2)(-.+
i=1
T i=1
(14) Due to space limitations, we are not able to print all
feature values of all EEG segments. Instead, we visualize
Then, R(T)/S(T) is calculated as
the averages of the features (except RIR and PSI) within
R(T) max(X(t, T)) − min(X(t, T)) each of the five sets in Figure 2. Error bars represent the
=   . (15)
S(T) (1/T) Tt=1 [x(t) − x]2 variances of features in each set. PSIs for five sets are
plotted in Figure 3. Users can replot these pictures and get
The Hurst Exponent is obtained by calculating the slope of averages of features on Python interpreter by a testing script
the line produced by ln(R(n)/S(n)) versus ln(n) for n ∈ (http://code.google.com/p/pyeeg/wiki/TestScript) from our
[2..N ]. Hurst Exponent is a scalar feature. project website.
Computational Intelligence and Neuroscience 5

Detrended fluctuation
Petrosian fractal dimension Higuchi fractal dimension Fisher information analysis

0.605 0.24 0.9

0.22 0.1
0.6 0.8
0.595 0.2 0.08 0.7
0.59 0.18
0.06 0.6
0.585 0.16
0.5
0.58 0.14 0.04
0.4
0.575 0.12
0.02
A B C D E A B C D E A B C D E A B C D E
(a) (b) (c) (d)
Hurst exponent Approximate entropy Spectral entropy SVD entropy
0.68 0.3 2.25
0.8 2.2
0.66
0.64 0.78
0.25 2.15
0.62 0.76
2.1
0.6 0.2 0.74
0.58 2.05
0.72
0.56 0.7 2
0.54 0.15
0.68 1.95
0.52
A B C D E A B C D E A B C D E A B C D E
(e) (f) (g) (h)
Hjorth mobility Hjorth complexity
0.006 0.25
0.0055
0.2
0.005
0.0045 0.15
0.004 0.1
0.0035
0.05
0.003
0.0025 0
A B C D E A B C D E
(i) (j)

Figure 2: Distributions of ten features extracted by PyEEG in each set.

Table 2: Values of parameters used in our example. on their definitions precisely even though faster algorithms
may exist. There are many other EEG features, such as
Parameter name Value In feature(s)
Lyapunov Exponents, that have not been yet implemented
kmax 5 HFD
in PyEEG. More EEG features will be added into PyEEG
τ 4 SVD Entropy in the future while we finish unit testing and documen-
dE 10 Fisher Information tation for each function. In personal emails, some open
r 0.3σ 1 source projects, such as ConnectomeViewer (http://www.
ApEn
m 10 connectomeviewer.org/viewer) and NIPY/PBrain (http://
fs 173 Spectral Entropy nipy.sourceforge.net/pbrain/), have expressed the interest in
band [1, 3, 5, . . . , 85] PSI and RIR including PyEEG into their code. Therefore, we will keep
1σ
is the standard deviation of the EEG segment. maintaining PyEEG as long as it can benefit the entire
computational neuroscience community.
From Figures 2 and 3, we can see that healthy, interictal,
and ictal EEG signals have different distributions for most
features. Table 2 lists parameters used in this experiment. Availability

5. Discussion and Future Development The software is released under GNU GPL v.3 at Google Code:
http://code.google.com/p/pyeeg/. No commercial software is
So far, we have listed features that can be extracted by required to run PyEEG. Because Python is cross-platform,
PyEEG and their definitions. Our implementation sticks PyEEG can run on Linux, Mac OS, and Windows.
6 Computational Intelligence and Neuroscience

×104 A ×104 B ×104 C

35 50 60

30 50
40
25
40
20 30
30
15
20
20
10
10 10
5

0 0 0
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
(a) (b) (c)
×104 D ×104 E
250
70

60 200

50
150
40

30 100

20
50
10

0 0
0 10 20 30 40 0 1 20 30 40
(d) (e)

Figure 3: Average PSI of each set. Note that the scale in y-axis of set E is much larger than that of other sets.

References intracranial EEG,” Journal of Machine Learning Research, vol.

7, pp. 1025–1044, 2006.
[1] J. Dauwels, F. Vialatte, and A. Cichocki, “A comparative study [6] C. W. Ko and H. W. Chung, “Automatic spike detection via
of synchrony measures for the early detection of Alzheimer’s an artificial neural network using raw EEG data: effects of
disease based on EEG,” in Proceedings of the 14th International data preparation and implications in the limitations of online
Conference on Neural Information Processing (ICONIP ’07), recognition,” Clinical Neurophysiology, vol. 111, no. 3, pp. 477–
vol. 4984 of Lecture Notes in Computer Science, pp. 112–125, 481, 2000.
2008. [7] V. Srinivasan, C. Eswaran, and N. Sriraam, “Approximate
[2] A. A. Petrosian, D. V. Prokhorov, W. Lajara-Nanson, and entropy-based epileptic EEG detection using artificial neural
R. B. Schiffer, “Recurrent neural network-based approach networks,” IEEE Transactions on Information Technology in
for early recognition of Alzheimer’s disease in EEG,” Clinical Biomedicine, vol. 11, no. 3, pp. 288–295, 2007.
Neurophysiology, vol. 112, no. 8, pp. 1378–1387, 2001. [8] F. S. Bao, J. M. Gao, J. Hu, D. Y. C. Lie, Y. Zhang, and K.
[3] F. S. Bao, D. Y. C. Lie, and Y. Zhang, “A new approach to J. Oommen, “Automated epilepsy diagnosis using interictal
automated epileptic diagnosis using EEG and probabilistic scalp EEG,” in Proceedings of the 31st Annual International
neural network,” in Proceedings of the 20th IEEE International Conference of the IEEE Engineering in Medicine and Biology
Conference on Tools with Artificial Intelligence (ICTAI ’08), vol. Society (EMBC ’09), pp. 6603–6607, September 2009.
2, pp. 482–486, November 2008. [9] K. Lehnertz, F. Mormann, T. Kreuz et al., “Seizure prediction
[4] J. Dauwels, E. Eskandar, and S. Cash, “Localization of seizure by nonlinear EEG analysis,” IEEE Engineering in Medicine and
onset area from intracranial non-seizure EEG by exploiting Biology Magazine, vol. 22, no. 1, pp. 57–63, 2003.
locally enhanced synchrony,” in Proceedings of the 31st Annual [10] E. O’Sullivan-Greene, I. Mareels, D. Freestone, L. Kulhmann,
International Conference of the IEEE Engineering in Medicine and A. Burkitt, “A paradigm for epileptic seizure prediction
and Biology Society (EMBC ’09), pp. 2180–2183, September using a coupled oscillator model of the brain,” in Proceedings
2009. of the 31st Annual International Conference of the IEEE
[5] A. B. Gardner, A. M. Krieger, G. Vachtsevanos, and B. Engineering in Medicine and Biology Society (EMBC ’09), pp.
Litt, “One-class novelty detection for seizure analysis from 6428–6431, September 2009.
Computational Intelligence and Neuroscience 7

[11] F. S. Bao, Y.-L. Li, J.-M. Gao, and J. Hu, “Performance of

dynamic features in classifying scalp epileptic interictal and
normal EEG,” in Proceedings of 32nd International Conference
of IEEE Engineering in Medicine and Biology Society (EMBC
’10), 2010.
[12] R. Q. Quiroga, S. Blanco, O. A. Rosso, H. Garcia, and
A. Rabinowicz, “Searching for hidden information with gabor
transform in generalized tonic-clonic seizures,” Electroen-
cephalography and Clinical Neurophysiology, vol. 103, no. 4, pp.
434–439, 1997.
[13] A. Petrosian, “Kolmogorov complexity of finite sequences and
recognition of different preictal EEG patterns,” in Proceedings
of the 8th IEEE Symposium on Computer-Based Medical
Systems, pp. 212–217, June 1995.
[14] T. Higuchi, “Approach to an irregular time series on the basis
of the fractal theory,” Physica D, vol. 31, no. 2, pp. 277–283,
1988.
[15] B. Hjorth, “EEG analysis based on time domain properties,”
Electroencephalography and Clinical Neurophysiology, vol. 29,
no. 3, pp. 306–310, 1970.
[16] T. Inouye, K. Shinosaki, H. Sakamoto et al., “Quantification of
EEG irregularity by use of the entropy of the power spectrum,”
Electroencephalography and Clinical Neurophysiology, vol. 79,
no. 3, pp. 204–210, 1991.
[17] S. J. Roberts, W. Penny, and I. Rezek, “Temporal and spatial
complexity measures for electroencephalogram based brain-
computer interfacing,” Medical and Biological Engineering and
Computing, vol. 37, no. 1, pp. 93–98, 1999.
[18] C. J. James and D. Lowe, “Extracting multisource brain activity
from a single electromagnetic channel,” Artificial Intelligence in
Medicine, vol. 28, no. 1, pp. 89–104, 2003.
[19] S. M. Pincus, I. M. Gladstone, and R. A. Ehrenkranz, “A
regularity statistic for medical data analysis,” Journal of Clinical
Monitoring and Computing, vol. 7, no. 4, pp. 335–345, 1991.
[20] C.-K. Peng, S. Havlin, H. E. Stanley, and A. L. Goldberger,
“Quantification of scaling exponents and crossover phenom-
ena in nonstationary heartbeat time series,” Chaos, vol. 5, no.
1, pp. 82–87, 1995.
[21] T. Balli and R. Palaniappan, “A combined linear & nonlinear
approach for classification of epileptic EEG signals,” in
Proceedings of the 4th International IEEE/EMBS Conference on
Neural Engineering (NER ’09), pp. 714–717, May 2009.
[22] R. G. Andrzejak, K. Lehnertz, F. Mormann, C. Rieke, P. David,
and C. E. Elger, “Indications of nonlinear deterministic and
finite-dimensional structures in time series of brain electrical
activity: dependence on recording region and brain state,”
Physical Review E, vol. 64, no. 6, Article ID 061907, 8 pages,
2001.
 Advances in Journal of
Industrial Engineering
Multimedia
Applied
Computational
Intelligence and Soft
Computing
The Scientific International Journal of
Distributed
Hindawi Publishing Corporation
World Journal
Hindawi Publishing Corporation
Sensor Networks
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Advances in

Fuzzy
Systems
Modelling &
Simulation
in Engineering
Hindawi Publishing Corporation
Hindawi Publishing Corporation Volume 2014 http://www.hindawi.com Volume 2014
http://www.hindawi.com

Submit your manuscripts at

Journal of
http://www.hindawi.com
Computer Networks
and Communications  Advances in 
Artificial
Intelligence
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014

International Journal of Advances in

Biomedical Imaging Artificial
Neural Systems

International Journal of
Advances in Computer Games Advances in
Computer Engineering Technology Software Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

International Journal of
Reconfigurable
Computing

Advances in Computational Journal of

Journal of Human-Computer Intelligence and Electrical and Computer
Robotics
Hindawi Publishing Corporation
Interaction
Hindawi Publishing Corporation
Neuroscience
Hindawi Publishing Corporation Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014