Session Item

Radiation therapy is an extremely complex process that evolves and changes at a fast pace, and ensuring that our patients receive the intended treatment accurately remains a constant challenge to all disciplines involved. It is essential that the treatment is performed safely, but at the same time effective and efficient. Standardization and automation have shown to be powerful tools not only in increasing efficiency but also ensuring robust and high quality throughout these complex processes. In addition, it will be shown that standardization and individualization (or personalized treatment) are to be considered rather synergistic as opposed to contradicting concepts. As every new patient adds new information to the different steps in the radiotherapy workflow, generating a huge amount of structured data, the growing interest and impact of artificial intelligence (AI) has almost been inevitable. Machine learning (ML) applications are generally perceived as “black boxes”, which explains that their implementation is mainly accepted for automation of processes that can easily be inspected (eg delineation and treatment planning). Some examples will be provided in this presentation, illustrating the impact of automation in delineation and treatment planning on overall quality in the treatment process. Nevertheless, the increasing usage of AI models for automation creates awareness for the need of dedicated quality assurance (QA) procedures ensuring the safety of these AI processes (QA of AI). On the other hand, AI might become a powerful tool to enhance and optimize the QA procedures in the increasingly complex workflows, perhaps ultimately, generating a measureless framework to control and ensure the quality of the treatment delivery process (AI for QA). Whatever the application, it is obvious that the introduction of automation will require to rethink the role of the human operator from actually performing certain (routine) tasks, into training of models, monitoring the performance and safeguarding the quality of these processes.