Diagnostics,
Journal Year:
2024,
Volume and Issue:
14(23), P. 2710 - 2710
Published: Nov. 30, 2024
Cancer
ranks
second
among
the
causes
of
mortality
worldwide,
following
cardiovascular
diseases.
Brain
cancer,
in
particular,
has
lowest
survival
rate
any
form
cancer.
tumors
vary
their
morphology,
texture,
and
location,
which
determine
classification.
The
accurate
diagnosis
enables
physicians
to
select
optimal
treatment
strategies
potentially
prolong
patients'
lives.
Researchers
who
have
implemented
deep
learning
models
for
diseases
recent
years
largely
focused
on
neural
network
optimization
enhance
performance.
This
involves
implementing
with
best
performance
incorporating
various
architectures
by
configuring
hyperparameters.
Brain
tumors
are
diseases
that
affect
the
most
vital
organs
of
human
body.
Abnormal
cell
development
causes
growth
lesions
in
brain.
In
visualizing
emergence
a
brain
tumor,
MRI
(Magnetic
Resonance
Imaging)
is
relatively
good
method
as
it
has
no
radiation
compared
to
other
methods.
Artificial
intelligence
expected
accelerate
radiologists
detecting
tumor's
emergence.
This
study
proposes
an
automatic
classification
using
deep
learning
architecture
with
eight
EfficientNet
models
(BO-B7)
variations
classify
results
into
normal
or
tumor.
The
perform
well,
which
EfficientNet-B7
achieves
highest
training
accuracy
99.71%
and
validation
99.67%.
Compared
conventional
CNN,
superior
performance
computation
time.
From
experimental
results,
level
CNN
less
than
EfficienNet.
indicates
architectural
modifications
presented
model,
by
combining
layer
numbers,
image
resolution
channels
can
improve
classifying
results.
Brain
tumor
classification
plays
a
prominent
rolein
accurate
identification
of
abnormal
brain
tissues
and
helps
in
clinical
diagnosis
patient.
This
work
presents
approach
based
on
deep
learning
framework.
Deep
learning-based
approaches
have
been
used
the
due
to
its
self-learning
capability
outperformance
problems.
In
this
work,
study
classification,
2D
MRI
data
are
used.
The
proposed
method
consists
three
stages:
i)
pre-processing,
ii)
design
architecture
for
iii)
integration
conventional
handcrafted
features
with
features.
A
detailed
has
done
by
training
architectures
raw
images,
Local
Binary
Pattern
(LBP)
coded
texture
features,
Discrete
Wavelet
Transform
(DWT)
coefficients
classification.
SoftMax
classifier
purpose.
To
authenticate
method,
publicly
available
dataset
(Br35H)
accuracies
achieved
81.11%
when
is
trained
LBP
feature,
94.11%
network
DWT
coefficient,
94%
image
ResNet
respectively
at
epochs
50.
Further,
effectiveness
demonstrated
comparing
results
other
existing
methods.
experimental
show
efficacy
over
methods
considered
comparison.
Diagnostics,
Journal Year:
2024,
Volume and Issue:
14(23), P. 2710 - 2710
Published: Nov. 30, 2024
Cancer
ranks
second
among
the
causes
of
mortality
worldwide,
following
cardiovascular
diseases.
Brain
cancer,
in
particular,
has
lowest
survival
rate
any
form
cancer.
tumors
vary
their
morphology,
texture,
and
location,
which
determine
classification.
The
accurate
diagnosis
enables
physicians
to
select
optimal
treatment
strategies
potentially
prolong
patients'
lives.
Researchers
who
have
implemented
deep
learning
models
for
diseases
recent
years
largely
focused
on
neural
network
optimization
enhance
performance.
This
involves
implementing
with
best
performance
incorporating
various
architectures
by
configuring
hyperparameters.
