CARS 2025 Computer Assisted Radiology and Surgery - 40th Anniversary and reflections on the role of modelling and AI
摘要
Based on CARS Congress events selected from its 40 year history, this editorial summarises the main challenges and solution concepts encountered, and what the future may hold for a model-centric world view in the specific domain of computer assisted radiology and surgery.
MethodsAltogether some 15,000 publications appeared in the CAR/CARS Congress Proceedings and Journal between 1985 and 2025, comprising approximately 3000 full papers and 12,000 long abstracts. Modelling occupied a central theme in many of these publications, particularly in the 2020s. Carefully selected statements from this period made by CARS Congress Presidents, members of the CARS Organizing Committee, Invited Speakers and Authors, serve to show how the CARS community has contributed to the evolution of model guided medicine.
ResultsIn terms of what can be derived from the modelling related themes and observations on their frequency in recent CARS Congress publications, there is an evolving timely focus on different aspects of a model guided medicine, such as methods and tools for situational models, process models and network models for distributed (AI) model services in health care. This indicates the need for multicentre AI models and associated clinical translational studies as well as model quality assurance, i.e. model verification, validation and evaluation.
ConclusionThe modelling-related themes presented at recent CARS Congresses are a first modest attempt to raise attention in the international community, particularly in the domain of radiology, surgery and informatics, that with modern modelling methods and tools, fundamental changes can be expected in health care. There is still a long way to go, as the awareness for model guided medicine is only gradually increasing. Future capabilities and acceptance of model guided medicine are strongly related to human-curated modelling methods and tools, as well as on effective human–computer interaction.