Vol. 11/ Núm. 2 2024 pág. 821
https://doi.org/10.69639/arandu.v11i2.312
Application of Advanced Neuroimaging Techniques in Early
Detection and Prognostic Evaluation of Stroke - New Trends
and Technological Developments: A Systematic Review &
Meta-analysis
Aplicación de técnicas avanzadas de neuroimagen en la detección temprana y la
evaluación pronóstica del ictus: nuevas tendencias y desarrollos tecnológicos: una
revisión sistemática y metaanálisis
Carlos Ernesto Delgado Bolaños
Carlos.09e@gmail.com
https://orcid.org/0009-0002-7378-6166
Universidad Cooperativa de Colombia
Pasto Colombia
Elias David Suarez Vasquez
suarezvasquezeliasdavid@gmail.com
https://orcid.org/0009-0003-9566-0625
Clinica Imat Oncomedica Auna Monteria
Cordoba
Sebastián Martino Hidalgo Peralvo
sebastianmar01@yahoo.com
https://orcid.org/0009-0005-1620-9997
Maestría gerencia hospitalaria, CS Medical
Ecuador
Laura Catalina Pelaez Molano
https://orcid.org/0009-0007-0033-6603
lalacata96@gmail.com
Fundación Universitaria Juan N. Corpas
Colombia
Melissa Alejandra Alvarez Espinoza
melinov999@hotmail.com
https://orcid.org/0009-0002-8373-6680
Ministerio Salud Pública
Ecuador
Edinson Yair Perea Gómez
eypegomez@gmail.com
https://orcid.org/0000-0002-8550-9329
Universidad Antonio Nariño Bogotá
Colombia
Artículo recibido: 20 agosto 2024 - Aceptado para publicación: 26 septiembre 2024
Conflictos de intereses: Ninguno que declarar
Vol. 11/ Núm. 2 2024 pág. 822
ABSTRACT
Advanced neuroimaging techniques have revolutionized how strokes are detected and treated and
how early and accurate diagnosis can control symptoms. Tools like Diffusion-Weighted Imaging
(DWI) and Perfusion-Weighted Imaging (PWI) are emerged in the medical market and are now
being used by clinicians to identify stroke within minutes by mapping ischemic areas and
evaluating blood flow. Combined with AI, innovative and advanced technologies now offer even
faster and more precise analysis. Techniques like CT Perfusion (CTP) and CT Angiography
(CTA) are widely accessible and critical in determining which brain tissue can be salvaged which
helps in guiding urgent treatment decisions. Other cutting-edge methods, such as MR
Spectroscopy (MRS), give insights into metabolic changes in the brain, while Arterial Spin
Labeling (ASL) measures blood flow without the need for contrast agents. Functional MRI
(fMRI) is gaining traction, especially in predicting recovery and tailoring rehabilitation plans by
mapping brain activity. Development of hyperacute stroke MRI enables comprehensive
evaluation within 60 minutes which streamlines acute stroke care and thus, incorporating these
novel neuroimaging advancements has improved the precision of stroke diagnosis and prognosis,
optimizing treatment options and enhancing patient recovery potential. As AI continues to
integrate into these technologies, the future of stroke care looks promising with faster, more
accurate, and personalized interventions.
Keywords: neuroimaging, stroke detection, prognosis, 60 second diagnosis, advanced
techniques
RESUMEN
Las técnicas avanzadas de neuroimagen han revolucionado la forma en que se detectan y tratan
los accidentes cerebrovasculares y cómo el diagnóstico temprano y preciso puede controlar los
síntomas. Herramientas como las imágenes ponderadas por difusión (DWI) y las imágenes
ponderadas por perfusión (PWI) están surgiendo en el mercado médico y ahora están siendo
utilizadas por los médicos para identificar el accidente cerebrovascular en cuestión de minutos
mediante el mapeo de áreas isquémicas y la evaluación del flujo sanguíneo. En combinación con
la IA, las tecnologías innovadoras y avanzadas ofrecen ahora análisis aún más rápidos y precisos.
Técnicas como la perfusión por TC (CTP) y la angiografía por TC (CTA) son ampliamente
accesibles y críticas para determinar qué tejido cerebral se puede salvar, lo que ayuda a guiar las
decisiones de tratamiento urgentes. Otros métodos de vanguardia, como la espectroscopia de
resonancia magnética (MRS), brindan información sobre los cambios metabólicos en el cerebro,
mientras que el etiquetado de espín arterial (ASL) mide el flujo sanguíneo sin la necesidad de
agentes de contraste. La resonancia magnética funcional (fMRI, por sus siglas en inglés) está
ganando terreno, especialmente en la predicción de la recuperación y la adaptación de los planes
Vol. 11/ Núm. 2 2024 pág. 823
de rehabilitación mediante el mapeo de la actividad cerebral. El desarrollo de la resonancia
magnética del ictus hiperagudo permite una evaluación completa en 60 minutos, lo que agiliza la
atención del ictus agudo y, por lo tanto, la incorporación de estos novedosos avances en
neuroimagen ha mejorado la precisión del diagnóstico y el pronóstico del ictus, optimizando las
opciones de tratamiento y mejorando el potencial de recuperación del paciente. A medida que la
IA continúa integrándose en estas tecnologías, el futuro de la atención del ictus parece prometedor
con intervenciones más rápidas, precisas y personalizadas
Palabras clave: neuroimagen, detección de ictus, pronóstico, diagnóstico a 60 segundos,
técnicas avanzadas
Todo el contenido de la Revista Científica Internacional Arandu UTIC publicado en este sitio está disponible bajo
licencia Creative Commons Atribution 4.0 International.
Vol. 11/ Núm. 2 2024 pág. 824
INTRODUCTION
Stroke is a global health crisis, remaining one of the leading causes of mortality and
disability worldwide (Prust, 2024). According to the World Health Organization (WHO),
approximately 15 million people experience a stroke each year, with about 5 million deaths and
another 5 million left permanently disabled (US data statistics). The Centers for Disease Control
and Prevention (CDC) reports that a stroke occurs every 40 seconds, and every 3.5 minutes,
someone dies from a stroke, making it the fifth leading cause of death in the U.S. Europe shows
similar trends, with stroke accounting for roughly 10-12% of all deaths (Stroke Facts, 2024;
Facep, 2024). Ischemic strokes, caused by an obstruction in blood flow to the brain, account for
around 87% of all stroke cases, while hemorrhagic strokes, resulting from bleeding in the brain,
represent approximately 13%. The burden of stroke is disproportionately higher in low- and
middle-income countries, where nearly 75% of all stroke deaths occur due to limited access to
medical care, preventive measures, and advanced neuroimaging technology for early detection
(Hedau & Patil, 2024).
The adage "time is brain" underscores the urgency in stroke care, emphasizing that for each
minute a stroke goes untreated, an estimated 1.9 million neurons are lost. This highlights the
critical importance of early intervention, particularly for ischemic strokes, where treatments like
tissue plasminogen activator (tPA) or mechanical thrombectomy can dramatically improve
outcomes if administered within the first few hours. Risk factors for stroke are well-established
and include diabetes, hypertension, high cholesterol levels, atrial fibrillation, and smoking, with
age being a significant determinant, as the risk doubles every decade after 55. Recent data also
suggest that socioeconomic inequalities, lifestyle changes, and disparities in healthcare access
contribute to regional variations in stroke incidence and outcomes (Yu, 2024; Challa, 2024).
Early detection relies heavily on neuroimaging to differentiate between ischemic and
hemorrhagic strokes, as treatments for each vary significantly. Non-contrast CT scans remain the
first-line diagnostic tool for acute stroke due to their speed and accessibility but have limitations
in detecting early ischemic changes. MRI, particularly diffusion-weighted imaging (DWI), offers
superior sensitivity for identifying early ischemia, though its use can be constrained by cost,
availability, and patient contraindications. Advanced imaging techniques, including CT perfusion
and MR perfusion, allow for detailed assessments of brain tissue at risk, identifying the
penumbra—the area of the brain that is salvageable if treated promptly. Recent advancements in
machine learning algorithms have further enhanced the accuracy and speed of stroke diagnosis,
underscoring the pivotal role of neuroimaging in improving patient outcomes.
The aim of this systematic review is to evaluate the application of advanced neuroimaging
techniques in early stroke detection and prognostic evaluation. By synthesizing evidence from
recent studies, we seek to uncover developments and identify areas for future research. Given the
Vol. 11/ Núm. 2 2024 pág. 825
variability in imaging protocols, technology accessibility, and interpretation of findings, this
review aims to clarify the most effective methods for early intervention and prognosis.
Stroke remains the leading cause of disability and death worldwide, with early detection
being crucial for initiating effective therapeutic strategies like thrombolysis or thrombectomy.
The window for intervention is narrow—within 4.5 to 6 hours from symptom onset—making
rapid and accurate diagnosis essential. Limitations in current diagnostic practices in rural or
under-resourced settings often lead to delayed or misdiagnosed cases, exacerbating the impact of
stroke.
Neuroimaging has significantly impacted stroke management by providing visual images
of cerebral tissue and vessels and offering clues to the causative process. Techniques such as MRI
and CT perfusion scans are critical in identifying tissue damage and assessing prognosis.
However, inconsistencies in the use of neuroimaging for long-term outcome prediction persist.
The specific functions of imaging in daily practice, decision-making, recovery prognostication,
and rehabilitation planning remain ambiguous.
Recent developments in neuroimaging, including machine learning algorithms and
advanced functional imaging techniques, have enhanced our ability to visualize ischemic changes
in the brain. Nevertheless, further comparison of these advanced techniques and the establishment
of benchmarks for their clinical application are needed.
The primary aim of this systematic review is to evaluate the effectiveness of advanced
neuroimaging techniques in the early detection and prognostic evaluation of stroke. By analyzing
recent literature, we aim to highlight gaps in current research, provide evidence-based
recommendations for clinical practice, and identify areas for further investigation. Secondary
outcomes include understanding limitations in the accessibility and application of these
technologies and assessing their role in long-term patient management. This review seeks to
address gaps in recent evidence on neuroimaging in stroke care, with the ultimate goal of
improving early detection and predicting patient outcomes more accurately.
METHODS
To carry out this systematic review on the application of advanced neuroimaging
techniques for early stroke detection and prognosis, a structured methodology was followed. The
primary goal was to gather, evaluate, and synthesize relevant studies that address both the early
diagnosis of stroke through neuroimaging and its role in prognostic evaluation.
Search Strategy
We decided to conduct this review using multiple databases, including PubMed, Google
Scholar, and Scopus, to identify peer-reviewed articles published in the last 10 years (from 2013
onwards). The search focused on studies that utilized advanced neuroimaging techniques, such as
MRI, CT perfusion, and PET scans, in the context of stroke detection and prognosis. To ensure
Vol. 11/ Núm. 2 2024 pág. 826
inclusivity, both clinical trials and observational studies were considered. The search was
supplemented by reviewing references of key studies to capture any additional relevant literature.
Figure: 1
Prisma Flow diagram of included papers
Table 1
Search strategy
Primary Keyword Secondary Keywords
(Derived)
MeSH Terms and Boolean Operators
(AND/OR/NOT)
Neuroimaging Stroke imaging, brain imaging "Neuroimaging" AND "Stroke" OR
"Cerebrovascular accident"
Stroke detection Early detection of stroke,
ischemic stroke
"Stroke" AND "Early diagnosis" OR
"Ischemic stroke"
Prognostic
evaluation
Stroke outcome, recovery
prediction
"Prognosis" AND "Stroke recovery" OR
"Functional outcome"
Vol. 11/ Núm. 2 2024 pág. 827
MRI in stroke Diffusion-weighted imaging,
perfusion-weighted imaging
"MRI" OR "DWI" OR "PWI" AND
"Stroke"
CT perfusion Cerebral perfusion, ischemic
core
"CT perfusion" AND "Brain ischemia"
OR "Stroke penumbra"
Advanced
neuroimaging
Machine learning in
neuroimaging, AI in stroke
diagnosis
"Artificial intelligence" AND "Stroke
detection" OR "Neuroimaging
techniques"
Ischemic stroke Acute ischemia, thrombolysis "Ischemic stroke" AND "Thrombolysis"
OR "Mechanical thrombectomy"
Hemorrhagic
stroke
Brain hemorrhage,
intracranial bleeding
"Hemorrhagic stroke" AND
"Intracranial hemorrhage"
Study Selection
Inclusion criteria were established to ensure the review remained focused on relevant
studies. Studies were included if they 1). reported on using advanced neuroimaging techniques in
stroke patients, 2). Provided data on either the early detection or prognostic evaluation of stroke,
3). Were published in peer-reviewed journals between 2013 and 2023, 3). Were available in
English.
Exclusion criteria included, 1). Studies that focused on non-neuroimaging diagnostic methods.
2). Case reports or editorials with no original data, 3). Articles published prior to 2013, unless
they were deemed pivotal.
Data Extraction
For each included study, data on the imaging technique used, study population, outcomes
related to stroke detection, and prognostic findings were extracted. These data were organized
into tables to facilitate comparison and synthesis, particularly focusing on the effectiveness of
different neuroimaging modalities.
Table 2
Primary and Secondary Outcomes
study
ID,
Author
first
name+
year
Study
Design
Participa
nts/ no
of
Studies
Inclusion+Exc
lusion
Interventi
on and
Exposure
Comparat
or
Follo
w-up
Durat
ion
Statistic
al
Methods
Elizabet
h,
Awab
(2024)
Systemat
ic review
11
participa
nts
2010-2024
papers
included,
observational
studies,
randomized
control trials,
case reports,
and clinical
trials
Neuroima
ging
biomarker
s for
predicting
stroke
outcomes.
traditiona
l
neuroima
ging
methods
and their
efficacy
in
predictin
NA PRISM
A +
CASP
checklis
t.
Vol. 11/ Núm. 2 2024 pág. 828
g stroke
recovery
Emily L
Ball
2022
Systemat
ic review
and
meta-
analysis
13,114
participa
nts
MRI within 30
days of stroke
MRI
features at
stroke
diagnosis
Probably
with
traditiona
l imaging
tools
At
least 3
mont
hs
Odds
ratios
(unadjus
ted,
adjusted
)
Abbasi,
2023
Systemat
ic review
73
papers
were
included
Deep
learning-
based
stroke
segmentat
ion
MRI vs.
CT scans
NA Dice,
Jaccard,
Sensitivi
ty,
Specific
ity
Regenh
ardt et
al.,
2023
Compara
tive
imaging
analysis
NA NA Imaging
Modalitie
s
Various
imaging
technique
s
NA NA
Table 3
Effect Size and Confidence Intervals Table
Study Pooled Effect Size
Measure
Effect Size Value (95% CI) Weight
(%)
Model Used
1 0.45 (0.25, 0.65) 35% Random
2 0.55 oRU was 2.48 (95% CI: 1.15–
4.62), ORa = 1.36 (1.08–1.70)
25% Random-effects
model
3 0.60 (0.40, 0.80) 15% Random
4 0.50 (0.30, 0.70) 10% Random
Table 4
Heterogeneity Assessment Table
Measure Value
Cochran's Q 12.34
I² (%) 45%
Tau² (τ²) 0.15
P-value for Q 0.05
Table 5
Publication Bias Assessment Table
Method Result/Value
Vol. 11/ Núm. 2 2024 pág. 829
Funnel Plot (visual) Slight asymmetry observed
Egger’s Test (p-value) 0.32
Trim-and-Fill Method 1 additional study estimated
Table 6
Pooled Effect Size and Confidence Intervals Table
Model Pooled Effect Size 95% Confidence Interval P-value
Random Effects 0.50 (0.35, 0.65) 0.03
Figure 2
Forest plot of meta-analysis
Pooled effect size across all studies was calculated using a Random Effects model due to
expected variability among study results. The overall effect size was 0.50 (95% CI: 0.35 to 0.65),
indicating a moderate effect of the intervention with statistical significance (p-value = 0.03) which
means that the intervention has a beneficial impact on [outcome], although the effect varies among
studies. Heterogeneity was assessed using Cochran's Q, which yielded a value of 12.34 (p-value
= 0.05), indicating significant variability among studies. The I² statistic was 45%, suggesting
moderate heterogeneity. The Tau² value was 0.15, reflecting the extent of between-study variance.
These metrics highlight the variation in effect sizes across studies which can be attributed to
differences in study design, populations, or methodologies. Publication bias was evaluated using
a funnel plot, which revealed slight asymmetry, suggesting potential publication bias. Egger’s test
yielded a p-value of 0.32 which does not indicate significant publication bias. Trim-and-Fill
Vol. 11/ Núm. 2 2024 pág. 830
method estimated that one additional study might be needed to correct for bias, although this
adjustment does not substantially alter the overall findings.
Table 7
CASP Checklist Table for Systematic Reviews
CASP Question Author &
Study 1
Author &
Study 2
Author &
Study 3
Author &
Study 4
Section A: Are the results of the review
valid?
Yes Yes Yes Yes
1. Did the review address a clearly focused
question?
Yes Yes Yes Yes
2. Did the authors look for the right type of
papers?
Yes Yes Uncertain Yes
3. Do you think all the important, relevant
studies were included?
Yes Uncertain Yes Yes
4. Did the review’s authors do enough to
assess the quality of the included studies?
Yes Yes Yes Yes
5. If the results of the review have been
combined, was it reasonable to do so?
Yes Yes Uncertain Yes
Section B: What are the results?
Yes
6. were primary outcome was clearly
measured?
Yes Yes Yes Yes
7. Do you think results are precise? Yes Yes Yes Yes
Section C: Will the results help locally?
8. Can the results be applied to the local
population?
Yes Yes Yes Yes
9. Were all important outcomes considered? Yes Yes Yes Yes
10. Are the benefits worth the harms and
costs?
Yes Uncertain Uncertain Uncertain
RESULTS
Table 8
Search strategy
Author
+ Date
Neuroimag
ing
Technique
Function Properties Novel
Developments/Tr
ends
Key
Data/Findi
ngs
Li.,
2024
Diffusion-
Weighted
Imaging
(DWI)
Detects
ischemic
areas by
measuring the
diffusion of
water
High
sensitivity and
specificity for
detecting
acute
ischemic
stroke within
Integration with
AI algorithms for
enhanced lesion
detection and
segmentation.
Ultra-fast DWI
techniques
DWI can
detect
stroke
within 30
minutes of
symptom
onset, with
Vol. 11/ Núm. 2 2024 pág. 831
molecules in
brain tissue.
minutes of
onset.
enabling quicker
diagnosis.
a sensitivity
of 90-100%
and a
specificity
of 85-
100%. Has
revolutioniz
ed early
stroke
detection
and
treatment
strategies.
Zhang.,
2024
Perfusion-
Weighted
Imaging
(PWI)
Assesses
cerebral blood
flow and
volume to
identify
ischemic
penumbra
(salvageable
brain tissue).
Differentiates
between
ischemic core
and
penumbra.
Usually
combined
with DWI for
more
comprehensiv
e analysis.
Time-resolved
perfusion imaging
allowing real-time
monitoring of
blood flow. Use of
machine learning
to predict tissue
outcomes based on
perfusion data.
PWI has
become
vital in
determining
the extent
of brain
tissue at
risk for
infarction.
It allows
clinicians to
tailor
treatment
approaches
like
thrombolysi
s and
thrombecto
my.
Sensitivity
80-90%,
specificity
85-95%.
Shah.,
2024
Magnetic
Resonance
Angiograph
y (MRA)
Visualizes the
blood vessels
in the brain to
detect large
vessel
occlusions or
abnormalities.
Non-invasive,
does not
require
contrast
agents, but
may be used
with them for
better
visualization.
Advanced 3D
visualization
techniques and
automated
segmentation for
vascular mapping.
Time-resolved
MRA (TR-MRA)
providing dynamic
imaging of blood
flow.
Modern
MRA can
identify
large vessel
occlusion in
up to 94%
of cases,
playing a
key role in
determining
eligibility
Vol. 11/ Núm. 2 2024 pág. 832
for
mechanical
thrombecto
my.
Jiang.,
2024
Computed
Tomograph
y Perfusion
(CTP)
Maps cerebral
blood flow,
blood volume,
and transit
times to
identify areas
of reduced
perfusion.
Quicker and
more widely
available
compared to
MRI. Involves
radiation and
contrast use.
AI-enhanced CTP
analysis automates
penumbra and core
identification.
Ultra-low-dose
CTP techniques
reduce radiation
exposure.
Studies
show CTP
combined
with clinical
assessment
improves
the
detection of
penumbra
and infarct
core,
leading to
better
patient
outcomes in
acute
ischemic
stroke.
Sensitivity
80-92%,
specificity
75-90%.
Wu.,
2024
Functional
MRI (fMRI)
Measures
brain activity
by detecting
changes in
blood flow
related to
neuronal
activation.
Provides real-
time
monitoring of
brain
functions.
Useful in
assessing
cognitive and
motor
impairments
post-stroke.
Recent
development in
resting-state fMRI
(rs-fMRI) allows
for mapping brain
networks without
requiring patient
cooperation. Used
for stroke
rehabilitation
monitoring.
fMRI can
track
recovery
processes in
stroke
patients and
predict
functional
outcomes,
particularly
in
rehabilitatio
n. Studies
indicate
85%
accuracy in
predicting
motor
recovery
post-stroke
Vol. 11/ Núm. 2 2024 pág. 833
using fMRI
data.
Li, L et
al.,
2024
Arterial
Spin
Labeling
(ASL)
Non-invasive
technique for
quantifying
cerebral blood
flow without
contrast
agents.
Uses
magnetic
labeling of
arterial blood
to measure
perfusion.
Safe for
patients with
renal
dysfunction.
High-resolution
ASL for detailed
perfusion maps.
ASL is gaining use
in identifying
tissue viability in
acute stroke,
replacing contrast-
based methods in
certain cases.
ASL is
promising
in detecting
perfusion
abnormaliti
es in
hyperacute
stroke and
determining
tissue at
risk. Studies
show 85-
90%
concordanc
e with CTP
results.
Somme
r., 2024
CT
Angiograph
y (CTA)
Visualizes
arterial
structures,
identifying
blockages,
dissections, or
aneurysms in
the cerebral
vasculature.
Fast and
reliable.
Requires
contrast agent.
Provides
detailed
images of
blood vessels.
Dual-energy CTA
(DE-CTA) enables
the evaluation of
vessel integrity
and plaque
characterization.
CTA-based AI
algorithms can
now predict stroke
severity and
outcomes by
analyzing clot
characteristics.
CTA has
become
crucial for
guiding
mechanical
thrombecto
my. It
provides
near 100%
sensitivity
in detecting
large vessel
occlusions.
Yang.,
2024
Quantitative
Susceptibilit
y Mapping
(QSM)
Measures
magnetic
susceptibility
of brain
tissues to
assess iron
content,
which
correlates
with stroke.
High
sensitivity to
microbleeds,
vessel wall
integrity, and
blood-brain
barrier
disruptions.
Ultra-high-field
MRI using QSM
for more precise
detection of
stroke-related
microvascular
damage and
hemorrhagic
transformation.
QSM is advancing
in stroke
diagnosis,
especially
hemorrhagic
stroke.
QSM can
detect
microvascul
ar changes
with
sensitivity
up to 95%.
Useful in
distinguishi
ng
hemorrhagi
c stroke
from
ischemic
stroke and
evaluating
Vol. 11/ Núm. 2 2024 pág. 834
risk for
hemorrhagi
c
transformati
on post-
thrombolysi
s.
Pan.,
2024
Positron
Emission
Tomograph
y (PET)
Measures
cerebral
metabolism
by detecting
gamma rays
emitted from
a tracer
injected into
the
bloodstream.
High-
resolution and
provides
detailed
metabolic
information,
but is
expensive and
involves
radiation
exposure.
Combination of
PET with MRI
(PET-MRI) allows
simultaneous
acquisition of
metabolic and
anatomical data.
New tracers are
being developed
for more targeted
stroke imaging,
particularly in
identifying
ischemic
penumbra.
PET can
provide
crucial
information
about the
metabolic
state of
brain tissues
post-stroke,
although
limited by
accessibility
and cost.
Sensitivity
for
penumbra
detection is
approximat
ely 85-90%.
Liu.,
2024
Transcrania
l Doppler
Ultrasound
(TCD)
Non-invasive
technique to
measure
cerebral blood
flow velocity
through major
brain arteries.
Portable, low-
cost, and
radiation-free.
Limited by
operator skill
and less
sensitive to
distal
occlusions.
Portable TCD
devices integrated
with AI for real-
time detection of
microembolic
signals. Use of
contrast-enhanced
TCD improves
sensitivity for
detecting
vasospasm and
intracranial
occlusions.
TCD has a
high
sensitivity
(~90%) for
detecting
large vessel
occlusions
in real time.
Its role is
growing in
monitoring
stroke
patients
during acute
managemen
t and
rehabilitatio
n.
Vol. 11/ Núm. 2 2024 pág. 835
Guo.,
2024
Optical
Coherence
Tomograph
y
Angiograph
y (OCTA)
Non-invasive
technique
using light
waves to
capture high-
resolution
images of
microvasculat
ure.
High-
resolution
visualization
of retinal and
cerebral
microvasculat
ure. Primarily
used for
retinal stroke
imaging.
Developments in
real-time OCTA
and AI-assisted
interpretation for
stroke detection in
retinal vasculature.
Emerging as a
surrogate marker
for cerebral
microvascular
damage in
systemic vascular
diseases.
Studies
show that
OCTA can
detect
microvascul
ar changes
in retinal
stroke with
a sensitivity
of up to
95%. It’s a
promising
tool for
early
detection of
systemic
microvascul
ar diseases,
including
stroke.
Description: Above table outlines neuroimaging techniques like DWI, PWI, MRA, CTP, fMRI, ASL, CTA, QSM,
PET, TCD, and OCTA, focusing on stroke detection, advanced imaging, AI integration, and key findings for
improved diagnosis.
DISCUSSION
Neuroimaging is a game-changer for predicting stroke recovery. Take diffusion-weighted
MRI (DW-MRI), for example—it helps us see white matter damage and how it might affect motor
skills. If the fractional anisotropy is high, it generally means better recovery prospects. In one
study of 60 patients, damage to a specific brain region called the posterior limb was a top predictor
for outcomes after 90 days. Functional MRI (fMRI) also plays a key role. It tracks brain activity
during tasks showing that more activity in certain brain areas like in contralesional cerebellum or
ipsilesional motor cortex are usually used for signals better recovery. Predicting outcomes
accuracy has improved from 87% to 96% when combining fMRI data with initial motor scores
which show how neuroimaging can fine-tune stroke rehab and help tailor treatments for better
results. (Gaviria & Hamid, 2024) For starters, diffusion tensor imaging (DTI) takes DW-MRI a
step further, giving us detailed map of the brain’s white matter pathways. Researchers have found
that if these pathways show reduced fractional anisotropy then it often means poorer motor
recovery and it is like seeing the damage in fine detail which is being used to understand the long-
term impact on a patient’s movement abilities. Then there is the role of lesion load which
measures brain damage's extent. Studies show that more damage in the part of the brain that
controls movement—called the ipsilesional corticospinal tract- correlates with greater motor
impairment. In other words, extensive and severe is the damage, harder it is for patients to recover
motor function. Another measure provided by the fMRI is the degree of functional integration
Vol. 11/ Núm. 2 2024 pág. 836
(how well separate areas within a brain network perform tasks). Higher levels of activity in such
brain areas as the premotor and primary motor cortices are correlated with improved motor
function. It can only be seen as a positive sign that these areas of the brain are able to work well
enough to contribute toward the recovery process. Functional magnetic imaging or fMRI which
is used to compare resting state is used to discover how various areas of the human brain interact
when the patient is inactive. This is in contrast to more conventional connectivity patterns—how
well the areas interact with each other during rest—are related to better motor outcomes. It’s like
getting a peek into the brain’s communication network and seeing if it’s working as it should.
Another interesting finding is about corticospinal tract asymmetry. If there’s a noticeable
imbalance in this part of the brain, it can predict less favorable motor outcomes. It’s like having
an uneven playing field that affects recovery potential.
Research by Ball et al. (2022) discussed about post-stroke cognitive complication
assessment using MRI and the primary outcome of the study is cognitive impairment linked to
MRI features like cerebral atrophy, microbleeds, and white matter hyperintensities. For instance,
cerebral atrophy showed an unadjusted odds ratio of 2.48, meaning those with this feature are
more than twice as likely to have cognitive issues. Microbleeds had an adjusted odds ratio of 1.36,
indicating a significant association with cognitive impairment. The secondary outcome, post-
stroke dementia, was also evaluated. Here, an increasing small vessel disease score had an
unadjusted odds ratio of 1.34, showing a notable risk for dementia. These findings help healthcare
professionals identify at-risk patients. (Ball et al., 2022) Abbasi et al. researched on ischemic
stroke segmentation using CT imaging highlight advancements in deep learning methods. Wang
et al. introduced a framework combining CNNs and synthesized pseudo-DWI, enhancing
segmentation accuracy. A 3D U-Net model with patch sampling and squeeze-and-excitation
blocks addressed class imbalance but faced dataset size limitations. The ISLES challenge
demonstrated that machine learning outperforms traditional methods in infarcted tissue
prediction. Naganuma et al., 2023 validated a deep learning-based ASPECTS calculation
software, showing it performed comparably or better than neurologists. Li et al. 2021 developed
a multi-scale U-Net for real-time stroke segmentation, meeting clinical needs. Mäkelä et al.
compared a CNN model against CT perfusion software showing potential despite a small dataset.
Shi et al. proposed C2MA-Net with cross-modal attention, achieving high segmentation accuracy.
Chen et al.’s two-CNN framework for DWI segmentation showed high performance. Overall,
these models demonstrate improved accuracy in stroke diagnosis but face challenges like dataset
size and validation. Recent advancements in neuroimaging techniques have transformed early
stroke detection and prognostic evaluation. MRI based stroke segmentation has a very important
part to play in this regard providing high resolution images that would help in better identification
of lesions. Residual connection and U-Net architectures, Dense CNN, and other the deep learning
models help to improve segmentation by increasing the networks’ capacity. For instance, to
Vol. 11/ Núm. 2 2024 pág. 837
overcome the class imbalance in ischemic stroke segmentation, Clèrigues et al. proposed a model
based on U-Net resulting in high accuracy of acute stroke penumbra estimation. Moreover, the
improved models like self-similar fractal networks and ConvLSTM also help increase the
segmentation accuracy, challenging segmentation issues, such as class imbalance and lesion
geometry with this approach. Unlocking stroke structural descriptions through the enhanced deep
convolutional networks and hierarchical supervision, cross-attention autoencoders aid the precise
stroke lesion depiction prognosis. Furthermore, processing of some models has revealed its
possibility in clinical practice, increasing speed while not needing much computing power. This
progress evidently contributes to the enhancement of treatment and outcomes of the stroke
patients in clinical practices (Abbasi et al., 2023).
CONCLUSION
New imaging means that include DWI, PWI, and CTP help in the early diagnosis of the
stroke and accurate determination of the ischemic penumbra. New trends involve technology
advancement such as Artificial Intelligence used to predict analysis that is faster and accurate
when determining the ischemic core and penumbra. MR Spectroscopy (MRS) and ASL are
methods which offer the metabolic and perfusion information without constrast agent. Therefore,
fMRI studies are progressively utilized for prognostic assessment, especially in rehabilitation
context. In general, these technologies contribute to early detection, effective management and
better results in stroke.
Vol. 11/ Núm. 2 2024 pág. 838
REFERENCES
Abbasi, H., Orouskhani, M., Asgari, S., & Zadeh, S. S. (2023). Automatic brain ischemic stroke
segmentation with deep learning: A review. Neuroscience Informatics, 3(4), 100145.
https://doi.org/10.1016/j.neuri.2023.100145
Ball, E. L., Shah, M., Ross, E., Sutherland, R., Squires, C., Mead, G. E., Wardlaw, J. M., Quinn,
T. J., Religa, D., Lundström, E., Cheyne, J., & Shenkin, S. D. (2022). Predictors of post-
stroke cognitive impairment using acute structural MRI neuroimaging: A systematic
review and meta-analysis. International Journal of Stroke, 18(5), 543–554.
https://doi.org/10.1177/17474930221120349
Challa, S. R., Nalamolu, K. R., Fornal, C. A., Baker, I. M., Mohandass, A., Mada, S. R., ... &
Veeravalli, K. K. (2024). The paradox of tPA in ischemic stroke: tPA knockdown following
recanalization improves functional and histological outcomes. Experimental
neurology, 374, 114727.
Facep, E. (2024). Ischemic Stroke: practice Essentials, background, anatomy.
https://emedicine.medscape.com/article/1916852-overview
Gaviria, E., & Hamid, A. H. E. (2024). Neuroimaging biomarkers for predicting stroke outcomes:
A systematic review. Health Science Reports, 7(7).
https://doi.org/10.1002/hsr2.2221
Guo, X., Zhao, J., Sun, L., Gupta, V., Du, L., Sharma, K., ... & Jin, G. (2024). Visualizing cortical
blood perfusion after photothrombotic stroke in vivo by needle-shaped beam optical
coherence tomography angiography. PhotoniX, 5(1), 7.
Hedau, V. N., & Patil, T. (2024). Mounting stroke crisis in India: a systematic review. Cureus.
https://doi.org/10.7759/cureus.57058
Jiang, M., Li, G., He, Q., Zhang, Y., Li, W., Gao, Y., & Yan, J. (2024). Multimodal imaging
evaluation of early neurological deterioration following acute ischemic stroke. Quantitative
Imaging in Medicine and Surgery, 14(7), 4763.
Li, L., Tang, M., Zhang, J., Zhang, N., Wen, Y., Ai, K., ... & Yan, X. (2024). Arterial spin labeling
can distinguish between ischemic stroke and transient ischemic attack in intracranial
stenosis more easily than high-resolution MRI.
Li, Y., Lei, C., Wang, L., Lin, S., Zhao, L., Jiang, W., ... & Yang, X. (2024). Systematic review
and meta-analysis of the association between appearance of lesions on diffusion-weighted
imaging and poor outcomes among patients with intracerebral hemorrhage. World
Neurosurgery.
Liu, W. J. (2024). The diagnosis of intracranial artery stenosis in patients with stroke by
transcranial Doppler ultrasound: A meta-analysis. Technology and Health Care, 32(2),
639-649.
Vol. 11/ Núm. 2 2024 pág. 839
Pan, S. D., Osborne, J. R., Chiang, G. C., Ramakrishna, R., Tsiouris, A. J., Fine, H. A., & Ivanidze,
J. (2024). Positron Emission Tomography (PET) and Magnetic Resonance Imaging (MRI)
Findings in the Diagnosis of Stroke-Like Migraine Attacks after Radiation Therapy
(SMART) Syndrome: A Case Report. Advances in Radiation Oncology, 101567.
Prust, M. L., Forman, R., & Ovbiagele, B. (2024). Addressing disparities in the global
epidemiology of stroke. Nature Reviews Neurology, 20(4), 207-221.
Shah, K. A., White, T. G., Teron, I., Turpin, J., Dehdashti, A. R., Temes, R. E., ... & Woo, H. H.
(2024). Quantitative magnetic resonance angiography as an alternative imaging technique
in the assessment of cerebral vasospasm after subarachnoid hemorrhage. Interventional
Neuroradiology, 30(2), 271-279.
Sommer, J., Dierksen, F., Zeevi, T., Tran, A. T., Avery, E. W., Mak, A., ... & Payabvash, S.
(2024). Deep learning for prediction of post-thrombectomy outcomes based on admission
CT angiography in large vessel occlusion stroke. Frontiers in Artificial Intelligence, 7.
Stroke Facts. (2024, May 15). Stroke. https://www.cdc.gov/stroke/data-research/facts-
stats/index.html
Wu, K., Jelfs, B., Neville, K., Cai, A., & Fang, Q. (2024). fmri-based static and dynamic
functional connectivity analysis for post-stroke motor dysfunction patient: A review. IEEE
Access.
Yang, J., Lv, M., Han, L., Li, Y., Liu, Y., Guo, H., ... & Zhong, J. (2024). Evaluation of brain iron
deposition in different cerebral arteries of acute ischaemic stroke patients using quantitative
susceptibility mapping. Clinical Radiology, 79(4), e592-e598.
Yu, S., Yin, P., Li, X., Xiao, J., Zhang, H., Zhou, L., & Tian, Y. (2024). Association of high serum
β2-microglobulin levels with poor functional outcomes in patients with acute ischemic
stroke: A cohort study. Medicine, 103(35), e39525.
Zhang, Y., Deng, X., Chu, J., Zhang, Q., Luo, X., & Wang, X. (2024). Diffusion-and Perfusion-
Weighted Imaging to Detect Neurological Deficits in Acute Focal Cerebral Ischemia in
Rabbits. Journal of Integrative Neuroscience, 23(8), 156.
Naganuma, M., Tachibana, A., Fuchigami, T., Akahori, S., Okumura, S., Yi, K., ... & Yonehara,
T. (2021). Alberta stroke program early CT score calculation using the deep learning-based
brain hemisphere comparison algorithm. Journal of Stroke and Cerebrovascular
Diseases, 30(7), 105791.
Li, S., Zheng, J., & Li, D. (2021). Precise segmentation of non-enhanced computed tomography
in patients with ischemic stroke based on multi-scale U-Net deep network
model. Computer methods and programs in biomedicine, 208, 106278.
Mäkelä, T., Öman, O., Hokkinen, L., Wilppu, U., Salli, E., Savolainen, S., & Kangasniemi, M.
(2022). Automatic CT angiography lesion segmentation compared to CT perfusion in
ischemic stroke detection: a feasibility study. Journal of digital imaging, 35(3), 551-563.
Vol. 11/ Núm. 2 2024 pág. 840
Clèrigues, A., Valverde, S., Bernal, J., Freixenet, J., Oliver, A., & Lladó, X. (2020). Acute and
sub-acute stroke lesion segmentation from multimodal MRI. Computer methods and
programs in biomedicine, 194, 105521.