ESTRO 2024 Congress report

At this year's ESTRO 2024 congress, I focused my attention on the technical advances in head-and-neck cancer radiotherapy, particularly in dose prediction and artificial intelligence (AI)-driven segmentation. These topics are pivotal to the enhancement of treatment precision and optimisation of patient outcomes.

Dose Prediction

One of the primary discussions was around dose prediction models, which are proving increasingly useful in the assessment of plan quality. It can be challenging to determine whether the doses to organs at risk (OARs) are minimised sufficiently or whether further optimisation is necessary. AI-based dose prediction models offer a promising solution to this problem as they can be used to estimate dose distributions more accurately and quickly than is possible for people. These models help oncologists and medical physicists to identify and adjust suboptimal plans accordingly, and therefore to ensure that patients receive the best possible radiotherapy plans. Integration of these models into clinical practice could streamline the planning process, reduce the burden on clinicians, and enhance the overall quality of radiotherapy plans.

AI for Segmentation

AI for segmentation was another major topic. Several centres have already implemented these technologies in their workflows. This was discussed extensively in the pre-conference clinical course titled "Embracing the AI Revolution: Practical Steps for Future-Proof Clinical Practice". The course featured numerous talks on the quality of AI-driven segmentation and its potential to revolutionise the roles of various professionals, including oncologists, radiation therapists and medical physicists. The use of AI models can significantly reduce the time required for manual contouring, increase consistency and improve the accuracy of OAR delineation. However, the implementation of these models must be carefully managed to ensure safety and effectiveness. The European Medical Device Regulation allows the use of home-developed AI models, provided they meet strict safety standards. This mirrors the stringent requirements for vendor-produced CE-marked models, in order to ensure a high level of patient safety regardless of the model's origin.

Automated Planning

The future of automated planning was also discussed at the ESTRO conference, although major developments in this area have been limited. Nonetheless, application of deep learning to dose prediction could play a role in the enhancement of automated planning systems. Treatment planners who use deep learning predictions to seed optimisations may be able to generate higher-quality radiotherapy plans. This approach leverages advanced algorithms to predict optimal dose distributions, which can be fine-tuned during the planning process. Use of this method should lead ultimately to better patient outcomes and potentially to more consistency in plan quality.

In conclusion, ESTRO 2024 highlighted significant advances in head-and-neck cancer radiotherapy. These advances are poised to improve the precision and efficiency of radiotherapy planning; however, collaboration among ESTRO members is essential to ensure that these technologies are used for all patients.

Christian Rønn Hansen

Department of Oncology, Odense University Hospital, Odense, Denmark

Danish Centre for Particle Therapy, Aarhus University Hospital, Aarhus, Denmark

Department of Clinical Research, University of Southern Denmark, Odense, Denmark