Please
cite
this
article
in
press
as:
Lu,
S.,
et
al.,
Early
identification
of
mild
cognitive
impairment
using
incom-
plete
random
forest-robust
support
vector
machine
and
FDG-PET
imaging.
Comput
Med
Imaging
Graph
(2017),
http://dx.doi.org/10.1016/j.compmedimag.2017.01.001
ARTICLE IN PRESS
G Model
CMIG-1487;
No.
of
Pages
7
Computerized
Medical
Imaging
and
Graphics
xxx
(2017)
xxx–xxx
Contents
lists
available
at
ScienceDirect
Computerized
Medical
Imaging
and
Graphics
j
ourna
l
h
om
epa
ge
:
www.elsevier.com/locate/compmedimag
Early
identification
of
mild
cognitive
impairment
using
incomplete
random
forest-robust
support
vector
machine
and
FDG-PET
imaging
夽
Shen
Lu
a
,
Yong
Xia
b,∗
,
Weidong
Cai
a,∗
,
Michael
Fulham
c,d
,
David
Dagan
Feng
a,e
,
Alzheimer’s
Disease
Neuroimaging
Initiative
a
Biomedical
and
Multimedia
Information
Technology
(BMIT)
Research
Group,
School
of
Information
Technologies,
University
of
Sydney,
NSW
2006,
Australia
b
Shaanxi
Key
Lab
of
Speech
&
Image
Information
Processing
(SAIIP),
School
of
Computer
Science,
Northwestern
Polytechnical
University,
Xi’an
710072,
China
c
Department
of
Molecular
Imaging,
Royal
Prince
Alfred
Hospital,
NSW
2050,
Australia
d
Sydney
Medical
School,
University
of
Sydney,
NSW
2006,
Australia
e
Med-X
Research
Institute,
Shanghai
Jiaotong
University,
Shanghai
200030,
China
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
26
May
2016
Received
in
revised
form
9
January
2017
Accepted
10
January
2017
Keywords:
FDG-PET
Alzheimer’s
disease
Mild
cognitive
impairment
Robust
optimization
a
b
s
t
r
a
c
t
Alzheimer’s
disease
(AD)
is
the
most
common
type
of
dementia
and
will
be
an
increasing
health
prob-
lem
in
society
as
the
population
ages.
Mild
cognitive
impairment
(MCI)
is
considered
to
be
a
prodromal
stage
of
AD.
The
ability
to
identify
subjects
with
MCI
will
be
increasingly
important
as
disease
modifying
therapies
for
AD
are
developed.
We
propose
a
semi-supervised
learning
method
based
on
robust
opti-
mization
for
the
identification
of
MCI
from
[18F]Fluorodeoxyglucose
PET
scans.
We
extracted
three
groups
of
spatial
features
from
the
cortical
and
subcortical
regions
of
each
FDG-PET
image
volume.
We
mea-
sured
the
statistical
uncertainty
related
to
these
spatial
features
via
transformation
using
an
incomplete
random
forest
and
formulated
the
MCI
identification
problem
under
a
robust
optimization
framework.
We
compared
our
approach
to
other
state-of-the-art
methods
in
different
learning
schemas.
Our
method
outperformed
the
other
techniques
in
the
ability
to
separate
MCI
from
normal
controls.
©
2017
Elsevier
Ltd.
All
rights
reserved.
1.
Introduction
Alzheimer’s
disease
(AD)
is
a
neurodegenerative
brain
disorder
that
is
characterized
by
progressive
memory
loss,
cognitive
impair-
ment
and
the
inability
to
perform
usual
daily
activities
(Teune
et
al.,
2010
).
It
is
the
most
common
type
of
dementia,
accounting
for
about
65%
of
all
dementia
cases
globally
and
the
number
of
patients
is
increasing
every
year
as
people
live
longer
(Devous,
2002).
Mild
cognitive
impairment
(MCI)
is
considered
as
the
prodromal
phase
of
AD
(Albert
et
al.,
2011).
Individuals
with
MCI
show
greater
cogni-
tive
impairment
than
expected
for
their
age,
but
they
do
not
meet
the
criteria
for
dementia
(McKhann
et
al.,
2011).
The
conversion
夽
Data
used
in
preparation
of
this
article
were
obtained
from
the
Alzheimer’s
Disease
Neuroimaging
Initiative
(ADNI)
database
(adni.loni.usc.edu).
As
such,
the
investigators
within
the
ADNI
contributed
to
the
design
and
implementation
of
ADNI
and/or
provided
data
but
did
not
participate
in
analysis
or
writing
of
this
report.
A
complete
listing
of
ADNI
investigators
can
be
found
at:
http://adni.loni.usc.edu/wp-
content/uploads/how
to
apply/ADNI
Ma-nuscript
Citations.pdf.
∗
Corresponding
authors.
E-mail
addresses:
yxia@nwpu.edu.cn
(Y.
Xia),
tom.cai@sydney.edu.au
(W.
Cai).
rate
of
MCI
to
AD
is
estimated
to
be
between
10%–25%
per
year
(
Grand
et
al.,
2011).
Although
there
are
no
current
disease
modi-
fying
agents
to
halt
the
progression
of
AD
there
are
a
number
of
clinical
trials
underway
in
patients
with
pre-symptomatic
disease
(
Morris
et
al.,
2012).
Thus
as
effective
therapies
become
available
the
early
identification
of
patients
with
MCI
will
be
of
tremendous
benefit
to
patients
and
their
families.
The
pathology
of
AD
includes
cortical
and
subcortical
atro-
phy
together
with
the
deposition
of
-amyloid.
Two
widely
used
AD
biomarkers
are
structural
imaging
with
magnetic
resonance
(MR)
imaging
(Fox
and
Schott,
2004)
and
functional
imaging
with
[
18
F]Fluorodeoxyglucose
positron
emission
tomography
(FDG-PET)
(
Devous,
2002).
The
advantage
of
FDG-PET
over
MR
imaging
is
that
PET
can
detect
reduced
cerebral
glucose
metabolism
before
struc-
tural
change
is
evident
on
MR
imaging.
The
separation
of
patients
with
MCI
from
normal
controls
(NCs)
by
the
visual
analysis
of
FDG-
PET
images,
however,
is
difficult.
Visual
interpretation
of
these
studies
is
also
operator-dependent
and
related
to
the
skill
and
expe-
rience
of
the
reader.
A
reliable
and
robust
computer-aided
method
could
improve
this
situation.
http://dx.doi.org/10.1016/j.compmedimag.2017.01.001
0895-6111/©
2017
Elsevier
Ltd.
All
rights
reserved.