Using artificial intelligence system for assisting the classification of breast ultrasound glandular tissue components in dense breast tissue
Hongju Yan,
No information about this author
Chaochao Dai,
No information about this author
Xiaojing Xu
No information about this author
et al.
Scientific Reports,
Journal Year:
2025,
Volume and Issue:
15(1)
Published: April 6, 2025
To
investigate
the
potential
of
employing
artificial
intelligence
(AI)
-driven
breast
ultrasound
analysis
models
for
classification
glandular
tissue
components
(GTC)
in
dense
tissue.
A
total
1,848
healthy
women
with
mammograms
classified
as
were
enrolled
this
prospective
study.
Residual
Network
(ResNet)
101
model
and
ResNet
fully
Convolutional
Networks
(ResNet
+
FCN)
segmentation
trained.
The
better
effective
was
selected
to
appraise
performance
3
radiologists
non-breast
radiologists.
evaluation
metrics
included
sensitivity,
specificity,
positive
predictive
value
(PPV).
ResNet101
demonstrated
superior
compared
FCN
model.
It
significantly
enhanced
sensitivity
all
by
0.060,
0.021,
0.170,
0.009,
0.052,
0.047,
respectively.
For
P1
P4
glandular,
PPVs
increased
0.154,
0.178,
0.027,
0.109
Ai-assisted.
Notably,
experienced
a
particularly
substantial
rise
PPV
(p
<
0.01).
This
study
trained
deep
learning
is
reliable
accurate
system
assisting
different
differentiate
images.
Language: Английский
Beyond dental radiographs, a radiomics-based study for the classification of caries extension and depth
Journal of Dental Sciences,
Journal Year:
2025,
Volume and Issue:
unknown
Published: April 1, 2025
Language: Английский
Classification of Parotid Tumors with Robust Radiomic Features from DCE- and DW-MRI
Journal of Imaging,
Journal Year:
2025,
Volume and Issue:
11(4), P. 122 - 122
Published: April 17, 2025
This
study
aims
to
evaluate
the
role
of
MRI-based
radiomic
analysis
and
machine
learning
using
both
DWI
with
multiple
B-values
dynamic
contrast-enhanced
T1-weighted
sequences
differentiate
benign
(B)
malignant
(M)
parotid
tumors.
Patients
underwent
DCE-
DW-MRI.
An
expert
radiologist
performed
manual
selection
3D
ROIs.
Classification
vs.
tumors
was
based
on
features
extracted
from
DCE-based
DW-based
parametric
maps.
Care
taken
in
robustness
evaluation
no-bias
features.
Several
classifiers
were
employed.
Sensitivity
specificity
ranged
0.6
0.8.
The
combination
LASSO
+
neural
networks
achieved
highest
performance
(0.76
sensitivity
0.75
specificity).
Our
identified
a
few
robust
respect
ROI
that
can
effectively
be
adopted
classifying
Language: Английский