Artificial
Intelligence
in
Medicine
53 (2011) 83–
95
Contents
lists
available
at
ScienceDirect
Artificial
Intelligence
in
Medicine
jou
rn
al
h
om
epage:
www.elsevier.com/locate/aiim
Kernel
machines
for
epilepsy
diagnosis
via
EEG
signal
classification:
A
comparative
study
Clodoaldo
A.M.
Lima
a,∗
,
André
L.V.
Coelho
b,∗
a
Information
Systems
Program,
School
of
Arts,
Sciences
and
Humanities,
University
of
São
Paulo,
Av.
Arlindo
Bettio,
1000,
03828-000
São
Paulo,
SP,
Brazil
b
Graduate
Program
in
Applied
Informatics,
Center
of
Technological
Sciences,
University
of
Fortaleza,
Av.
Washington
Soares,
1321/J30,
60811-905
Fortaleza,
CE,
Brazil
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
22
December
2010
Received
in
revised
form
24
May
2011
Accepted
18
July
2011
Keywords:
Kernel
machines
Feature
extraction
EEG
signal
classification
Epilepsy
a
b
s
t
r
a
c
t
Objective:
We
carry
out
a
systematic
assessment
on
a
suite
of
kernel-based
learning
machines
while
coping
with
the
task
of
epilepsy
diagnosis
through
automatic
electroencephalogram
(EEG)
signal
classification.
Methods
and
materials:
The
kernel
machines
investigated
include
the
standard
support
vector
machine
(SVM),
the
least
squares
SVM,
the
Lagrangian
SVM,
the
smooth
SVM,
the
proximal
SVM,
and
the
rel-
evance
vector
machine.
An
extensive
series
of
experiments
was
conducted
on
publicly
available
data,
whose
clinical
EEG
recordings
were
obtained
from
five
normal
subjects
and
five
epileptic
patients.
The
performance
levels
delivered
by
the
different
kernel
machines
are
contrasted
in
terms
of
the
criteria
of
predictive
accuracy,
sensitivity
to
the
kernel
function/parameter
value,
and
sensitivity
to
the
type
of
features
extracted
from
the
signal.
For
this
purpose,
26
values
for
the
kernel
parameter
(radius)
of
two
well-known
kernel
functions
(namely,
Gaussian
and
exponential
radial
basis
functions)
were
considered
as
well
as
21
types
of
features
extracted
from
the
EEG
signal,
including
statistical
values
derived
from
the
discrete
wavelet
transform,
Lyapunov
exponents,
and
combinations
thereof.
Results:
We
first
quantitatively
assess
the
impact
of
the
choice
of
the
wavelet
basis
on
the
quality
of
the
features
extracted.
Four
wavelet
basis
functions
were
considered
in
this
study.
Then,
we
provide
the
average
accuracy
(i.e.,
cross-validation
error)
values
delivered
by
252
kernel
machine
configurations;
in
particular,
40%/35%
of
the
best-calibrated
models
of
the
standard
and
least
squares
SVMs
reached
100%
accuracy
rate
for
the
two
kernel
functions
considered.
Moreover,
we
show
the
sensitivity
profiles
exhib-
ited
by
a
large
sample
of
the
configurations
whereby
one
can
visually
inspect
their
levels
of
sensitiveness
to
the
type
of
feature
and
to
the
kernel
function/parameter
value.
Conclusions:
Overall,
the
results
evidence
that
all
kernel
machines
are
competitive
in
terms
of
accuracy,
with
the
standard
and
least
squares
SVMs
prevailing
more
consistently.
Moreover,
the
choice
of
the
kernel
function
and
parameter
value
as
well
as
the
choice
of
the
feature
extractor
are
critical
decisions
to
be
taken,
albeit
the
choice
of
the
wavelet
family
seems
not
to
be
so
relevant.
Also,
the
statistical
values
calculated
over
the
Lyapunov
exponents
were
good
sources
of
signal
representation,
but
not
as
informative
as
their
wavelet
counterparts.
Finally,
a
typical
sensitivity
profile
has
emerged
among
all
types
of
machines,
involving
some
regions
of
stability
separated
by
zones
of
sharp
variation,
with
some
kernel
parameter
values
clearly
associated
with
better
accuracy
rates
(zones
of
optimality).
© 2011 Elsevier B.V. All rights reserved.
1.
Introduction
In
recent
years,
there
has
been
an
increasing
interest
in
the
application
of
kernel-based
learning
machines
to
real-world
data
analysis
problems,
like
those
involving
pattern
recognition,
func-
tional
regression,
density
estimation,
and/or
time
series
forecasting
[1,2].
This
type
of
non-parametric
learning
machine
is
different
from
earlier
techniques
in
many
respects,
particularly
in
the
way
it
utilizes
and
combines
results
from
optimization,
statistics,
and
∗
Corresponding
author.
Tel.:
+55
11
3091
1008,
fax:
+55
11
3091
8800.
E-mail
addresses:
c.lima@usp.br
(C.A.M.
Lima),
acoelho@unifor.br
(A.L.V.
Coelho).
functional
analysis
theory
to
achieve
maximum
generality,
flexibil-
ity,
and
performance.
Arguably,
the
most
representative
of
this
new
class
of
methods
are
the
support
vector
machines
(SVMs)
[3–5],
which
exhibit
structural
properties
that
resemble
those
shown
by
well-known
artificial
neural
network
(ANN)
models.
SVMs
operate
within
the
framework
of
regularization
theory
by
minimizing
an
empirical
risk
in
a
well-posed
and
consistent
way
[4].
In
particular,
when
dealing
with
the
task
of
pattern
classifica-
tion,
SVMs
are
considered
as
large-margin
binary
classifiers,
since
they
are
designed
to
separate,
with
the
widest
margin
possible,
the
patterns
belonging
to
the
different
groups
of
data
[3].
A
clear
advan-
tage
presented
by
the
SVM
approach
is
that
sparse
solutions
are
usually
obtained
as
a
result
of
its
associated
training
process.
Only
0933-3657/$
–
see
front
matter ©
2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.artmed.2011.07.003