Hamza Memon

Background:  Stroke is the 5th most common cause of death and a major cause of disability in the US, with a prevalence of 2.5% of adults.¹ In addition, its economic burden in the US is estimated at $34 billion per year.² Strokes are categorized as hemorrhagic or ischemic, either of which lead to a lack of oxygen and nutrients supplied to the brain. Hemorrhagic stroke is often a result of vascular rupture, while ischemic stroke is from vascular occlusion.³ Lack of nutrients and oxygen to the brain eventually results in necrosis or tissue death.⁴ The diagnosis of an ischemic stroke is often confirmed by imaging modalities such as computed tomography (CT) and/or magnetic resonance imaging (MRI).¹ CT is usually the first imaging modality utilized because it can detect hemorrhage and is readily available, time-efficient, and cost-effective. Unfortunately, CT has a high rate of false positives in the detection of ischemic stroke.⁵ Thrombolysis is the most efficient treatment for ischemic stroke but carries a 6% risk of intracranial hemorrhage.⁷ If an ischemic stroke is accurately identified within 4.5 hours from onset, the patient can receive thrombolytics.⁶

Research Objective:  Since stroke management is time sensitive, multiple models incorporating artificial intelligence (AI) have been developed to improve accuracy, time to diagnosis, and treatment outcomes. In this review, three different roles of AI are analyzed: stroke detection, onset approximation, and prediction of adverse reactions post-treatment to evaluate the potential of AI in the management of stroke from detection to treatment.

Methods:  A search within PubMed and Google Scholar was performed to identify studies within the past 5 years that developed AI models for ischemic stroke management. Multiple studies were analyzed, each with a different role in stroke management. Each AI model utilized different MRI or CT datasets to train and test their models.

Results:  The AI model for stroke detection showed favorable outcomes for detecting infarct area.⁵ Similarly, studies evaluating onset detection and prediction of treatment outcome revealed higher accuracy using AI than with traditional methods.^7,8

Conclusion:  The utilization of AI in medicine has unparalleled potential and is rapidly progressing.  Multiple AI models are already being investigated for stroke management, yielding positive results. However, there is a need for these models to be further developed and validated before clinical translation.

Work Cited:

1. Feske SK. Ischemic Stroke. Am J Med. 2021;134(12):1457-1464. doi:10.1016/j.amjmed.2021.07.027
2. Soun JE, Chow DS, Nagamine M, et al. Artificial Intelligence and Acute Stroke Imaging. AJNR Am J Neuroradiol. 2021;42(1):2-11. doi:10.3174/ajnr.A6883
3. Nishio M, Koyasu S, Noguchi S, et al. Automatic detection of acute ischemic stroke using non-contrast computed tomography and two-stage deep learning model. Comput Methods Programs Biomed. 2020;196:105711. doi:10.1016/j.cmpb.2020.105711
4. Maida CD, Norrito RL, Daidone M, Tuttolomondo A, Pinto A. Neuroinflammatory Mechanisms in Ischemic Stroke: Focus on Cardioembolic Stroke, Background, and Therapeutic Approaches. Int J Mol Sci. 2020;21(18):6454. Published 2020 Sep 4. doi:10.3390/ijms21186454
5. Kaothanthong N, Atsavasirilert K, Sarampakhul S, Chantangphol P, Songsaeng D, Makhanov S. Artificial intelligence for localization of the acute ischemic stroke by non-contrast computed tomography. PLoS One. 2022;17(12):e0277573. Published 2022 Dec 1. doi:10.1371/journal.pone.0277573
6. Lee H, Lee EJ, Ham S, et al. Machine Learning Approach to Identify Stroke Within 4.5 Hours. Stroke. 2020;51(3):860-866. doi:10.1161/STROKEAHA.119.027611
7. Bentley P, Ganesalingam J, Carlton Jones AL, et al. Prediction of stroke thrombolysis outcome using CT brain machine learning. Neuroimage Clin. 2014;4:635-640. Published 2014 Mar 30. doi:10.1016/j.nicl.2014.02.003
8. Ho KC, Speier W, Zhang H, Scalzo F, El-Saden S, Arnold CW. A Machine Learning Approach for Classifying Ischemic Stroke Onset Time From Imaging. IEEE Trans Med Imaging. 2019;38(7):1666-1676. doi:10.1109/TMI.2019.2901445