Advancing radiological AI: From robust foundations to clinical readiness

The sixth issue of the European Journal of Radiology Artificial Intelligence (EJR AI) reflects a field that is moving from technical ambition towards clinical accountability. This issue therefore tells a broader story. Innovation remains essential, but innovation alone is no longer sufficient. In radiological AI, credibility now depends on whether a method can be trusted, explained, implemented, and used responsibly. Four interconnected narratives emerge across the issue: First , clinically meaningful AI depends on robust methodological foundations, including representative data, reliable segmentation, consistent annotation, transparent metrics, and careful handling of uncertainty. Second , the value of AI is increasingly judged by its ability to support action in clinical workflows, not only by its ability to detect or classify findings. Reporting support, triage, decision guidance, workload reduction, and safety-oriented applications illustrate this shift from performance toward practical utility. Third , imaging AI is becoming more clinically relevant when it reflects the broader diagnostic context, including multimodal information, longitudinal data, and human expertise already embedded in radiological workflows. Finally , generative AI and large language models highlight both the promise and the risks of fluent, plausible outputs, reinforcing the need for human oversight, explainability, validation, and responsibility. Together, this issue presents a maturing discipline shaped by trust, transparency, implementation, and clinical dependability driven by an active radiological AI community.