Sensors,
Год журнала:
2022,
Номер
22(24), С. 9875 - 9875
Опубликована: Дек. 15, 2022
Of
the
various
tumour
types,
colorectal
cancer
and
brain
tumours
are
still
considered
among
most
serious
deadly
diseases
in
world.
Therefore,
many
researchers
interested
improving
accuracy
reliability
of
diagnostic
medical
machine
learning
models.
In
computer-aided
diagnosis,
self-supervised
has
been
proven
to
be
an
effective
solution
when
dealing
with
datasets
insufficient
data
annotations.
However,
image
often
suffer
from
irregularities,
making
recognition
task
even
more
challenging.
The
class
decomposition
approach
provided
a
robust
such
challenging
problem
by
simplifying
boundaries
dataset.
this
paper,
we
propose
model,
called
XDecompo,
improve
transferability
features
pretext
downstream
task.
XDecompo
designed
based
on
affinity
propagation-based
effectively
encourage
explainable
component
highlight
important
pixels
that
contribute
classification
explain
effect
speciality
extracted
features.
We
also
explore
generalisability
handling
different
datasets,
as
histopathology
for
images.
quantitative
results
demonstrate
robustness
high
96.16%
94.30%
CRC
images,
respectively.
demonstrated
its
generalization
capability
achieved
(both
quantitatively
qualitatively)
compared
other
Moreover,
post
hoc
method
used
validate
feature
transferability,
demonstrating
highly
accurate
representations.
International Journal of Molecular Sciences,
Год журнала:
2024,
Номер
25(3), С. 1789 - 1789
Опубликована: Фев. 1, 2024
Endometriosis
(E)
and
adenomyosis
(A)
are
associated
with
a
wide
spectrum
of
symptoms
may
present
various
histopathological
transformations,
such
as
the
presence
hyperplasia,
atypia,
malignant
transformation
occurring
under
influence
local
inflammatory,
vascular
hormonal
factors
by
alteration
tumor
suppressor
proteins
inhibition
cell
apoptosis,
an
increased
degree
lesion
proliferation.
Material
methods:
This
retrospective
study
included
243
patients
from
whom
tissue
E/A
or
normal
control
uterine
was
harvested
stained
histochemical
classical
immunohistochemical
staining.
We
assessed
symptomatology
patients,
structure
ectopic
epithelium
neovascularization,
hormone
receptors,
inflammatory
cells
oncoproteins
involved
in
development.
Atypical
areas
were
analyzed
using
multiple
immunolabeling
techniques.
Results:
The
cytokeratin
(CK)
CK7+/CK20−
expression
profile
E
foci
differentiated
them
digestive
metastases.
neovascularization
marker
cluster
differentiation
(CD)
34+
increased,
especially
A
foci.
T:CD3+
lymphocytes,
B:CD20+
CD68+
macrophages
tryptase+
mast
abundant,
cases
transformation,
being
markers
proinflammatory
microenvironment.
In
addition,
we
found
significantly
division
index
(Ki67+),
inactivation
genes
p53,
B-cell
lymphoma
2
(BCL-2)
Phosphatase
tensin
homolog
(PTEN)
E/A-transformed
malignancy.
Conclusions:
Proinflammatory/vascular/hormonal
changes
trigger
progression
onset
cellular
atypia
exacerbating
symptoms,
pain
vaginal
bleeding.
These
triggers
represent
future
therapeutic
targets.
Expert Systems with Applications,
Год журнала:
2022,
Номер
216, С. 119452 - 119452
Опубликована: Дек. 30, 2022
Transformer
models
have
recently
become
the
dominant
architecture
in
many
computer
vision
tasks,
including
image
classification,
object
detection,
and
segmentation.
The
main
reason
behind
their
success
is
ability
to
incorporate
global
context
information
into
learning
process.
By
utilising
self-attention,
recent
advancements
design
enable
consider
long-range
dependencies.
In
this
paper,
we
propose
a
novel
transformer,
named
Swin
with
Cascaded
UPsampling
(SwinCup)
model
for
segmentation
of
histopathology
images.
We
use
hierarchical
shifted
windows
as
an
encoder
extract
features.
multi-scale
feature
extraction
transformer
enables
attend
different
areas
at
scales.
A
cascaded
up-sampling
decoder
used
improve
its
aggregation.
Experiments
on
GLAS
CRAG
colorectal
cancer
datasets
were
validate
model,
achieving
average
0.90
(F1
score)
surpassing
state-of-the-art
by
(23%).
Journal of Imaging,
Год журнала:
2023,
Номер
9(9), С. 173 - 173
Опубликована: Авг. 27, 2023
Computer-assisted
diagnostic
systems
have
been
developed
to
aid
doctors
in
diagnosing
thyroid-related
abnormalities.
The
aim
of
this
research
is
improve
the
diagnosis
accuracy
thyroid
abnormality
detection
models
that
can
be
utilized
alleviate
undue
pressure
on
healthcare
professionals.
In
research,
we
proposed
deep
learning,
metaheuristics,
and
a
MCDM
algorithms-based
framework
detect
abnormalities
from
ultrasound
histopathological
images.
method
uses
three
recently
learning
techniques
(DeiT,
Swin
Transformer,
Mixer-MLP)
extract
features
image
datasets.
feature
extraction
are
based
Image
Transformer
MLP
models.
There
large
number
redundant
overfit
classifiers
reduce
generalization
capabilities
classifiers.
order
avoid
overfitting
problem,
six
transformation
(PCA,
TSVD,
FastICA,
ISOMAP,
LLE,
UMP)
analyzed
dimensionality
data.
five
different
(LR,
NB,
SVC,
KNN,
RF)
evaluated
using
5-fold
stratified
cross-validation
technique
transformed
dataset.
Both
datasets
exhibit
class
imbalances
hence,
used
evaluate
performance.
MEREC-TOPSIS
for
ranking
at
analysis
stages.
first
stage,
best
classification
chosen,
whereas,
second
reduction
wrapper
selection
mode.
Two
best-ranked
further
selected
weighted
average
ensemble
meta-heuristics
FOX-optimization
algorithm.
PCA+FOX
optimization-based
+
random
forest
model
achieved
highest
TOPSIS
score
performed
exceptionally
well
with
an
99.13%,
F2-score
98.82%,
AUC-ROC
99.13%
Similarly,
90.65%,
92.01%,
95.48%
This
study
exploits
combination
novelty
algorithms
cancer
capabilities.
outperforms
current
state-of-the-art
methods
significantly
medical
professionals
by
reducing
excessive
burden
fraternity.
Diagnostics,
Год журнала:
2024,
Номер
14(4), С. 422 - 422
Опубликована: Фев. 14, 2024
Breast
cancer
remains
a
significant
global
public
health
concern,
emphasizing
the
critical
role
of
accurate
histopathological
analysis
in
diagnosis
and
treatment
planning.
In
recent
years,
advent
deep
learning
techniques
has
showcased
notable
potential
elevating
precision
efficiency
data
analysis.
The
proposed
work
introduces
novel
approach
that
harnesses
power
Transfer
Learning
to
capitalize
on
knowledge
gleaned
from
pre-trained
models,
adapting
it
nuanced
landscape
breast
histopathology.
Our
model,
Learning-based
concatenated
exhibits
substantial
performance
enhancements
compared
traditional
methodologies.
Leveraging
well-established
pretrained
models
such
as
VGG-16,
MobileNetV2,
ResNet50,
DenseNet121—each
Convolutional
Neural
Network
architecture
designed
for
classification
tasks—this
study
meticulously
tunes
hyperparameters
optimize
model
performance.
implementation
is
systematically
benchmarked
against
individual
classifiers
data.
Remarkably,
our
achieves
an
impressive
training
accuracy
98%.
outcomes
experiments
underscore
efficacy
this
four-level
advancing
By
synergizing
strengths
transfer
learning,
holds
augment
diagnostic
capabilities
pathologists,
thereby
contributing
more
informed
personalized
planning
individuals
diagnosed
with
cancer.
This
research
heralds
promising
stride
toward
leveraging
cutting-edge
technology
refine
understanding
management
cancer,
marking
advancement
intersection
artificial
intelligence
healthcare.
Frontiers in Oncology,
Год журнала:
2024,
Номер
13
Опубликована: Янв. 17, 2024
The
field
of
histopathological
image
analysis
has
evolved
significantly
with
the
advent
digital
pathology,
leading
to
development
automated
models
capable
classifying
tissues
and
structures
within
diverse
pathological
images.
Artificial
intelligence
algorithms,
such
as
convolutional
neural
networks,
have
shown
remarkable
capabilities
in
pathology
tasks,
including
tumor
identification,
metastasis
detection,
patient
prognosis
assessment.
However,
traditional
manual
methods
generally
low
accuracy
diagnosing
colorectal
cancer
using
This
study
investigates
use
AI
classification
analytics
images
histogram
oriented
gradients
method.
develops
an
AI-based
architecture
for
images,
aiming
achieve
high
performance
less
complexity
through
specific
parameters
layers.
In
this
study,
we
investigate
complicated
state
classification,
explicitly
focusing
on
categorizing
nine
distinct
tissue
types.
Our
research
used
open-source
multi-centered
datasets
that
included
records
100.000
non-overlapping
from
86
patients
training
7180
50
testing.
compares
two
approaches,
artificial
intelligence-based
algorithms
machine
learning
models,
automate
classification.
comprises
primary
tasks:
binary
distinguishing
between
normal
tissues,
multi-classification,
encompassing
types,
adipose,
background,
debris,
stroma,
lymphocytes,
mucus,
smooth
muscle,
colon
mucosa,
tumor.
findings
show
systems
can
0.91
0.97
multi-class
classifications.
comparison,
directed
gradient
features
Random
Forest
classifier
achieved
rates
0.75
0.44
classifications,
respectively.
are
generalizable,
allowing
them
be
integrated
into
histopathology
diagnostics
procedures
improve
diagnostic
efficiency.
CNN
model
outperforms
existing
techniques,
demonstrating
its
potential
precision
effectiveness
analysis.
emphasizes
importance
maintaining
data
consistency
applying
normalization
during
preparation
stage
It
particularly
highlights
assess
Diagnostics,
Год журнала:
2024,
Номер
14(4), С. 388 - 388
Опубликована: Фев. 10, 2024
Digital
pathology
(DP)
has
begun
to
play
a
key
role
in
the
evaluation
of
liver
specimens.
Recent
studies
have
shown
that
workflow
combines
DP
and
artificial
intelligence
(AI)
applied
histopathology
potential
value
supporting
diagnosis,
treatment
evaluation,
prognosis
prediction
diseases.
Here,
we
provide
systematic
review
use
this
field
hepatology.
Based
on
PRISMA
2020
criteria,
search
PubMed,
SCOPUS,
Embase
electronic
databases
was
conducted,
applying
inclusion/exclusion
filters.
The
articles
were
evaluated
by
two
independent
reviewers,
who
extracted
specifications
objectives
each
study,
AI
tools
used,
results
obtained.
From
266
initial
records
identified,
25
eligible
selected,
mainly
conducted
human
tissues.
Most
performed
using
whole-slide
imaging
systems
for
acquisition
different
machine
learning
deep
methods
image
pre-processing,
segmentation,
feature
extractions,
classification.
Of
note,
most
selected
demonstrated
good
performance
as
classifiers
histological
images
compared
pathologist
annotations.
Promising
date
bode
well
not-too-distant
inclusion
these
techniques
clinical
practice.
Colorectal
carcinoma,
a
prevalent
and
deadly
malignancy,
necessitates
precise
histopathological
assessment
for
effective
diagnosis
prognosis.
Artificial
intelligence
(AI)
emerges
as
transformative
force
in
this
realm,
offering
innovative
solutions
to
enhance
traditional
methods.
This
narrative
review
explores
AI's
pioneering
role
colorectal
carcinoma
histopathology,
encompassing
its
evolution,
techniques,
advancements.
AI
algorithms,
notably
machine
learning
deep
learning,
have
revolutionized
image
analysis,
facilitating
accurate
prognosis
prediction.
Furthermore,
AI-driven
analysis
unveils
potential
biomarkers
therapeutic
targets,
heralding
personalized
treatment
approaches.
Despite
promise,
challenges
persist,
including
data
quality,
interpretability,
integration.
Collaborative
efforts
among
researchers,
clinicians,
developers
are
imperative
surmount
these
hurdles
realize
full
care.
underscores
impact
implications
future
oncology
research,
clinical
practice,
interdisciplinary
collaboration.