Beyond the algorithm: embedding ethics for trustworthy AI in radiology and oncology
-
By
-
April 20, 2026
Objective:
To explore specific ethical and societal aspects of AI in radiology and oncology, proposing a structured pathway for trustworthy AI development.
Key Findings:
- Ongoing interdisciplinary involvement is essential for addressing ethical issues in radiological AI.
- Four guiding dimensions of trustworthy AI were identified: explainability, trust, accountability, and fairness.
- Trustworthiness is relational and co-constructed through interactions among diverse stakeholders.
Interpretation:
Ethical issues in AI require contextual understanding and active engagement with stakeholders, not just technical solutions.
Limitations:
- High-level principles may become ineffective without specific contextual grounding, limiting their applicability.
- Trustworthiness as a concept is subject to philosophical criticism, which may challenge its operationalization.
Conclusion:
A shift from principle-driven ethics to embedded, interdisciplinary approaches is necessary for developing trustworthy AI in cancer care, emphasizing the importance of stakeholder engagement.
