On the Challenges and Perspectives of Stroke Segmentation Using AI

Stroke is one of the leading causes of disability and death worldwide, underscoring the need for early and precise diagnosis to enhance patient outcomes. Artificial intelligence (AI) has emerged as a promising tool for stroke diagnosis, offering the potential to significantly improve diagnostic accuracy and efficiency. This paper reviews recent AI-based methods for stroke diagnosis, focusing on available datasets with various modalities and the AI techniques employed. We analyze the limitations of existing approaches, including data and model biases, lack of generalizability, and transparency issues. We then propose future directions for AI in stroke diagnosis, aiming to address these identified gaps, including the development of more robust and adaptable models, and improved data integration strategies. Finally, we emphasize the importance of ethical considerations and regulatory frameworks to ensure the responsible implementation of AI in this critical field. A GitHub repository is created to include recent codes and platforms that are available for stroke diagnosis.