Artificial Intelligence in Acute Ischemic Stroke Subtypes According to Toast Classification: A Comprehensive...

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Accurate ischemic stroke (IS) etiology recognition is crucial for timely treatment and secondary prevention.
• Identifying IS causes is challenging, relying on clinical data, imaging, and diagnostic tests.
• The TOAST classification system categorizes IS into five subtypes: large-artery atherosclerosis (LAAS), cardioembolism (CEI), small vessel disease (SVD), other determined etiology (ODE), and undetermined etiology (UDE).

Purpose of the Study:

• To review the most effective artificial intelligence (AI) models for differential diagnosis of ischemic stroke etiology based on the TOAST classification.
• To highlight AI's role in improving the accuracy of identifying specific IS causes.

Main Methods:

• Review of existing literature on AI applications in ischemic stroke diagnosis.
• Analysis of AI model performance in detecting key indicators for IS subtypes (e.g., carotid stenosis, atrial fibrillation, small vessel disease).
• Focus on AI's utility in classifying stroke etiology according to the TOAST system.

Main Results:

• AI models demonstrate increased sensitivity in diagnosing major IS causes, including tomographic detection of carotid stenosis, electrocardiographic recognition of atrial fibrillation, and MRI identification of small vessel disease.
• AI proves effective in identifying predictive factors for subtyping acute stroke patients within large, heterogeneous populations.
• AI particularly excels in clarifying the etiology of undetermined etiology ischemic stroke (UDE IS), especially in detecting cardioembolic sources.

Conclusions:

• AI is a valuable tool for the differential diagnosis of ischemic stroke etiology.
• AI enhances the ability to subtype acute stroke patients and identify underlying causes.
• AI shows significant promise in resolving undetermined stroke etiologies, improving patient management and outcomes.