RC 1713 - Clinical testing processes for artificial intelligence applications in radiology: from procurement to validation and quality control
March 7, 08:00 - 09:00 CET
5 min
Chairpersons' introduction
15 min
Optimising AI application procurement: key considerations for integration and compliance in radiology workflows
To identify essential technical, clinical, and regulatory criteria for evaluating AI solutions during the procurement process.
To understand the legal, ethical, and compliance considerations involved in AI procurement, including data privacy, cybersecurity, and adherence to local and international standards (e.g., Conformité Européenne (European CE mark), Food and Drug Administration (FDA), General Data Protection Regulation (GDPR)).
To assess strategies for effectively integrating AI applications into existing radiology workflows, ensuring compatibility with PACS/RIS systems, user experience optimisation, and clinician acceptance.
15 min
Clinical validation of AI applications in radiology
To understand the key phases of clinical validation for radiology AI applications.
To identify critical success factors and common pitfalls in designing and executing clinical validation processes.
To evaluate the role of stakeholders (medical physicists, radiologists, radiographers, IT support, clinicians, data scientists, and regulatory bodies) in shaping validation strategy, and how validation outcomes inform procurement decisions and quality assurance processes.
15 min
Ensuring quality and safety in AI-driven radiology systems
To describe the main principles and frameworks for assessing the quality and safety of AI systems in radiology, including performance monitoring, error analysis, and compliance with clinical standards.
To recognise potential risks and failure modes in the deployment of AI tools within radiological workflows, and explore strategies for mitigating harm through continuous validation and human-AI collaboration.
To apply best practices for post-deployment surveillance, auditing, and quality assurance, ensuring AI applications maintain clinical effectiveness and safety across diverse patient populations and imaging environments.
10 min
Panel discussion: How could we test AI applications before implementing them into clinical practice, taking into account the size and resources of the hospital/department?