Legal and Ethical Aspects of AI in Radiology

Developing, implementing, and deploying AI in radiology require more than only focusing on technological and medical aspects like functionality, effectiveness, and efficiency. To ensure basic levels of user-friendliness, user acceptance, and public support, it is important to also consider legal and ethical aspects. Legal requirements are a conditio sine qua non: non-compliance with legal norms would constitute an illegal practice. Ethical norms may be softer but may need consideration in cases in which there are no legal requirements, or when legal requirements are unclear or insufficiently detailed. In this chapter, the most relevant legal and ethical aspects of AI in radiology are discussed. Relevant legal aspects include safety, privacy and data protection, security, and bias and non-discrimination. Relevant ethical aspects include human dignity, autonomy, and accountability. These lists are not exhaustive; depending on the context, other legal and ethical aspects may be relevant as well. Therefore, some approaches (most notably impact assessments and value-sensitive design) are examined that can help identify legal and ethical aspects and take them into account for AI in radiology.