Page 8 - AIH-2-4

Artificial Intelligence in Health AI in acute stroke imaging

increased life expectancy. This figure is projected to rise 2026. AI-powered tools further enhance radiologic image
15
to 186.88 million by 2030 and 224.86 million by 2050, as analysis, enabling fast and precise identification of ischemic
reported by Cheng et al. in the Global Burden of Disease and hemorrhagic strokes, as well as vascular abnormalities,
3
2021 study. In India, stroke is the fourth-leading cause of thereby supporting swift and effective stroke management.
death and the fifth-leading cause of disability. The stroke- Machine learning algorithms assist in analyzing CT
related death rate in India has increased from 44 to 55 angiograms and identifying large vessel occlusions (LVOs)
people per 100,000 population between 1990 and 2021. in real time. Studies have shown that these AI tools reduce
2
According to a study conducted by Pandian et al., India door-to-treatment times by promptly alerting clinicians.
4
16
has the highest DALYs due to stroke among countries in AI can assess imaging data to determine whether a patient
the Southeast Asia Region. is eligible for procedures like mechanical thrombectomy
A striking feature in India is that a large proportion or tissue plasminogen activator administration. AI
of the stroke patients are from the younger population, models also analyze electronic health records, imaging
unlike in developed countries. Nearly 20% of patients data, and outputs from wearable devices to assess stroke
hospitalized with a first-time stroke are under 40 years risk. For example, predictive algorithms can detect atrial
of age. Younger individuals are increasingly at risk due fibrillation, a major stroke risk factor, from smartwatch
5
to sedentary lifestyles, substance use (including tobacco, electrocardiogram data with high sensitivity. 17
nicotine, alcohol, and illicit drugs), and stress. Other Several studies have evaluated the role of AI in stroke
risk factors involve elevated blood pressure, blood sugar, management and patient care. 18-23 A review article by Liu
cholesterol, and body weight. 6,7 et al. highlights the role of AI in areas such as automated
18
The paradigm “time is brain” is pivotal in stroke care, segmentation of infarct areas, identification of LVOs,
as millions of neurons die with each minute that a stroke stroke outcome prediction, analysis of hemorrhagic
goes untreated. Therefore, treating stroke patients within the transformation risk, prediction of recurrent ischemic
critical window period or the golden hour (within 60 min stroke, and automated grading of collateral circulation.
of symptom onset) is essential. During this time, physicians Al-Janabi et al. provided an overview of the AI tools used
19
should administer medication and initiate treatment as to identify strokes and guide acute ischemic stroke care.
quickly as possible. Beyond this golden hour, irreversible This review paper explores the transformative
8
brain damage occurs, making treatment less effective. potential of AI in stroke care. It provides an overview of
Treatment strategies include intravenous tissue plasminogen AI applications in acute stroke care imaging, focusing on
activator for thrombolysis and endovascular treatment the advancements in detection and screening, triaging and
(EVT). Timely intervention is critical to minimize damage prioritization, quantification and prognosis, automated
9
and improve outcomes. However, disparities in stroke care image interpretation, and workflow optimization, supported
persist due to delays in diagnosis, limited access to treatment, by published review articles on the subject.
and a shortage of radiologists and stroke care experts.
Teleradiology has transformed stroke care by enabling 2. Methodology
rapid, high-quality, and accurate radiologic interpretation,
even from remote locations. 8,10-12 Still, the sharp rise in the A comprehensive literature search was conducted using
volume of radiologic imaging, without a corresponding the PubMed and Google Scholar databases, focusing
increase in the number of trained radiologists, necessitates on papers evaluating the use of AI in stroke imaging,
more scalable and efficient solutions. 13 published between 2014 and 2024. Keywords used for the
search included “artificial intelligence in stroke,” “AI in
Neuroimaging is essential for identifying acute strokes acute stroke,” “AI in hemorrhage,” “AI in ASPECTS score,”
and distinguishing between ischemic and hemorrhagic “AI in large vessel occlusion,” and “AI in midline shift.”
types. Tools such as computed tomography (CT)
14
and magnetic resonance imaging (MRI) are pivotal Studies were included if they focused on the application
in detecting, characterizing, and diagnosing strokes. of AI in stroke imaging, specifically involving acute
Artificial intelligence (AI) is a rapidly advancing field that stroke, hemorrhage detection, Alberta Stroke Program
offers powerful tools for fast and efficient imaging analysis. Early Computed Tomography Scoring (ASPECTS), LVO
Its emergence has enabled the analysis of large datasets, detection, or midline shift assessment. Only original
pattern recognition, and prediction with unprecedented research articles, reviews, and systematic reviews written
speed and accuracy. In healthcare, AI is growing at a in English and with full-text availability were considered.
rate of 40% per annum and is projected to help decrease Studies were excluded if they were unrelated to medical
healthcare costs by United States Dollar 150 billion by imaging or the application of AI in stroke, or if they

Volume 2 Issue 4 (2025) 2 doi: 10.36922/AIH025140025