ESTRO 2024 Congress Report

1/ Dose accumulation for patient treatment

Dose mapping and accumulation are important issues in the vision of optimal health for all, together. It will improve the quality of treatment by estimating the fractional doses delivered, in replanning situations, or in determining the total dose when combined modalities (external beam radiotherapy and brachytherapy) are delivered to the patient in the same course. This process can be divided into three main tasks: i) image registration and deformation, ii) dose mapping and iii) dose accumulation. It was clear from this session of ESTRO 2024 that the first point is very challenging, not only due to patient anatomical variations in intra- and inter-fraction, but also due to the lack of reliable metrics to assess the geometric registration performed by the algorithms. Advanced registration solutions should be implemented in the available commercial systems. Dose accumulation can be applied by simply summing the physical dose or by considering the biological effect of dose deposition through the use of a forward or backward methodology. The associated uncertainty, due to geometric, biological or case complexity uncertainties, depends on the time difference between patient images, but its relevance is application-dependent.

To avoid large variations in accumulated doses, this process requires a high degree of consensus and the establishment of a standardised procedure.

2/ Debate: This house believes that AI will do dosimetry for us and there is no longer any need to train in it

Despite this being the artificial intelligence (AI) era, 74% of the audience was against the motion before the debate began.

During the debate, the AI corner argued that this technology could be important in linear accelerator commissioning, given that the beam models of these machines tend to be very similar and some machines come already pre-commissioned. AI could also help to prevent accidental exposures. In fact, according to reports from the International Atomic Energy Agency and the World Health Organization, dosimetry was a cause of such incidents. The lack of standardisation, automation and the expectation of standard results and benchmarking in dosimetry tasks were also cited as arguments in favour of AI. It was also noted that dosimetry was currently of little importance and that not every medical physicist should be a dosimetry expert. The training programme for medical physicists should be adapted to drive efficiently and support the introduction of new technologies in this field.

Against the motion, it was argued that the application of AI in dosimetry was constrained by the fact that AI worked well on relatively simple, well-established and routine processes and was based on only a limited set of data. The studies with AI applications in dosimetry are relatively small. For this reason, the implementation of AI in dosimetry should be undertaken with the objective of improving efficiency and quality and reducing the number of required repetitive tasks as much as possible. However, it should not be employed in place of human expertise. Medical physicists are trained to interpret and validate dosimetry results. The dosimetry task is an exclusive role of the medical physicist. In fact, incidents in radiotherapy in the past were due to a lack of education and proper training.

At the end of the debate, the voting behaviour remained unchanged. Following the presentation of all arguments, the vote against the motion was reinforced, with 83% of the audience in favour of this position.

Tiago Ventura

Medical physicist, Medical Physics Department, IPO Coimbra, Portugal

Member of ESTRO Dosimetry and quality assurance focus group