Human-AI Interaction in Interventional Radiology: A Narrative Review of Current Applications, Challenges, and Future...

Area of Science:

• Interventional Radiology
• Artificial Intelligence
• Human-AI Interaction

Background:

• Traditional AI evaluations in interventional radiology (IR) focus on algorithmic performance, neglecting real-world clinical impact.
• The dynamic, operator-dependent, and time-sensitive nature of IR requires a nuanced approach to AI integration.

Purpose of the Study:

• To review the state of the art of AI integration in IR through human-AI interaction (HAI).
• To critically assess AI's clinical integration, limitations, and future directions in IR.
• To explore AI applications across different procedural phases in IR.

Main Methods:

• A comprehensive literature survey of AI applications in IR.
• Focus on AI systems for decision support, real-time procedural verification, and immersive interfaces.
• Critical evaluation of determinants for effective clinical adoption.

Main Results:

• AI demonstrates preliminary potential to support operator performance in select IR tasks.
• Most AI applications are experimental, retrospective, or preclinical.
• Benefits include aiding patient selection, procedural planning, and outcome assessment via immersive technologies.

Conclusions:

• Effective AI clinical utility in IR depends on human-AI interaction (HAI), not just algorithmic accuracy.
• Interpretability, workflow integration, and trust calibration are key determinants for AI adoption.
• A human-centered, interaction-based model is essential for developing adaptive AI systems for IR.