Interventional Radiology Reporting Standards and Checklist for Artificial Intelligence Research Evaluation (iCARE).

Area of Science:

• Medical Imaging and Artificial Intelligence
• Interventional Radiology Research

Background:

• Artificial intelligence (AI) is increasingly used in interventional radiology (IR).
• Ensuring the robustness of AI systems in IR research and practice is crucial.
• Novel AI technologies require standardized evaluation before clinical integration.

Purpose of the Study:

• To introduce comprehensive standards and an evaluation checklist (iCARE) for AI in IR.
• To ensure the robustness and reliability of AI applications in interventional radiology.
• To support the development of safe and generalizable AI technologies for IR.

Main Methods:

• Development of the iCARE (interventional radiology AI Robustness Evaluation) checklist.
• The checklist covers the AI development pipeline from "code-to-clinic".
• Key areas include dataset curation, training, explainability, privacy, bias, reproducibility, and deployment.

Main Results:

• The iCARE checklist provides a structured framework for evaluating AI in IR.
• It addresses critical aspects of AI development and deployment.
• Facilitates the assessment of AI systems for clinical use in interventional radiology.

Conclusions:

• The iCARE checklist aims to enhance the safety and generalizability of AI in IR.
• It supports the integration of AI into IR workflows for improved patient care and outcomes.
• Standardized evaluation is essential for trustworthy AI in medical practice.