ESTRO 2024 Congress Report
The ESTRO24 conference highlighted the transformative role of Artificial Intelligence (AI) in Radiotherapy through a series of insightful sessions, showcasing its rapid integration into clinical practice and future potential. Below is a summary of the key points discussed in some of those sessions, emphasising the perspectives of Physicists, Oncologists, Radiation Therapists (RTTs), and the broader implications of AI in Radiotherapy.
Symposium: Artificial Intelligence in Daily Radiotherapy Practice
From a Physicist's Perspective Speaker: C. Van den Berg
AI is profoundly reshaping the Radiotherapy workflow, enhancing productivity and quality gains. Van den Berg emphasised that, while AI increases efficiency, its successful integration hinges on user confidence and usability, and this applies especially for online adaptative Radiotherapy where AI must indicate confidence levels to users. Interactive tools for AI, especially for clinical target volume (CTV) contouring, are still lacking. Hybrid large language models (LLMs) combined with vision models are seen as promising, allowing for text-prompted modifications of contours. Van den Berg projected that within the next 5 to 10 years, AI assistants will become an integral part of daily Radiotherapy, aiding in consulting and data analysis. A significant challenge lies in the fusion of imaging with AI, which promises substantial gains in terms of productivity, consistency and quality. Ultimately, autonomous workflows under human supervision are envisaged. However, this supervision will also cost time and require a new skill set.
From an Oncologist's Perspective Speaker: JG. Eriksen
Eriksen highlighted AI's potential to manage increasing workloads and improve delineation quality, reducing inter-observer variation and contouring time. AI-assisted contouring is poised to become standard in clinics, necessitating continuous evaluation and regular updates of AI models. The next generation must be educated to critically evaluate AI without extensive delineation experience. Eriksen pointed out vast opportunities like multi-institutional trials but stressed the importance of foundational steps: "We need to crawl before we can walk."
From a Radiation Therapist's Perspective Speaker: S. Perryck
Perryck presented a comprehensive overview of AI's role in Radiotherapy, referencing a review paper on AI in clinical practice (https://doi.org/10.1259/bjro.20230030). She discussed AI's applications across the treatment workflow, from imaging to treatment planning and delivery. As AI technologies continue to evolve, they promise to improve treatment efficiency by reducing variability and saving time. The importance of training RTTs with digital skills (https://doi.org/10.1016/j.radi.2022.06.017) to effectively integrate AI into the workflow was also underscored, ensuring that AI serves as a collaborative tool rather than a replacement. Looking ahead, the synergy between humans and AI could address critical issues like staff shortages and future workloads, ultimately enhancing patient care. Perryck also addressed ethical and generalizability challenges, reinforcing that AI requires high-quality input data to produce reliable outputs.
Symposium: AI Fairness in Radiotherapy: From Training to Impact
Fairness Towards All Users Speaker: Stine S. Korreman
Korreman discussed AI's potential to democratise Radiotherapy access globally, especially in low- and middle-income countries (LMICs). AI can alleviate staff shortages and streamline time-consuming tasks, enhancing standard-of-care treatment accessibility. However, the cost of AI tools could exacerbate inequalities if only high-resource clinics can afford them. Korreman highlighted a study with the International Atomic Energy Agency, demonstrating AI's effectiveness in reducing inter-observer variability and delineation time in LMICs. The session recommended affordable, open-source AI tools to promote global fairness in Radiotherapy.
Symposium: What's New in AI Developments and Automation for Brachytherapy
Can ChatGPT Help Our Professional Efforts in Brachytherapy? Speaker: Stefanie Corradini
Corradini explored ChatGPT's role in improving patient-clinician communication and education in brachytherapy. The AI model can translate complex medical terms into patient-friendly language, enhancing understanding and treatment adherence. ChatGPT's ability to answer patient queries and provide personalised education materials was highlighted. The findings also emphasised the necessity of human oversight to ensure the accuracy and ethical use of AI-driven communications, paving the way for a synergistic integration of AI in Radiotherapy.
Closing Debate: The Great AI Debate - This House Believes That the Radiation Therapy Care Pathway Will Be Delivered Entirely by Bots by 2040
Clan for the Motion: Speakers: A. Hope and S. Korreman
Proponents argued that automation (bots) is necessary to handle the cancer demands and alleviate staff shortages. They emphasised that AI will not replace jobs but will transform roles, requiring humans to monitor and improve bots.
Clan Against the Motion: Speakers: A. Dekker and EM. Vasquez Osorio
Opponents contended that full automation by 2040 is unrealistic due to the complexity of the Radiotherapy care pathway. They raised concerns about AI biases, accountability, and the importance of human empathy in patient care.
The opposition won the voting debate, suggesting scepticism about the complete automation of the Radiotherapy care pathway in the near future.
![Picture25nb.jpg Picture25nb.jpg](/getattachment/About/Newsroom/Newsletter/RTT-Corner/Artificial-Intelligence-in-Radiotherapy-Practice-H/Picture25nb.jpg.aspx?width=200&height=180)
Bárbara Barbosa
Specialist Radiation Therapist
Instituto Português de Oncologia do Porto (IPO Porto), Portugal
barbara.barbosa@ipoporto.min-saude.pt