ESTRO 2024 - Abstract Book

S3527

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2024

An example of the three reporting outputs for our method is provided for brainstem and oral cavity for patient 1 in Figure 1. Figure 2 shows the number of plausible registrations evaluated for each OAR, summarised for each patient and registration pipeline. Out of a total of 2496 mDTA evaluations, 88% (2196) met the TG 132 threshold. No clear trends in performance were found per registration pipeline (rows in Figure 2a). For specific patients, however, fewer registrations were deemed plausible, for example, only 74% of all registrations for patient 3. This may indicate that anatomical changes seen in patient 3 are more challenging for the DIR algorithms compared to other patients. Clear trends for registration performance are observed for OARs (columns in Figure 2a and b). For example, all registrations satisfied the criteria of mDTA ≤ 3mm for brainstem, cochleae, and optic chiasm, but for the left/right submandibular glands, only 42.7%/36.5% registrations, respectively, met this criterion.

Conclusion:

We developed an approach for individualised quantification of geometric and dosimetric uncertainties for head and neck re-irradiation. Our results highlight the variation in performance of DIR between patients and OARs which will help identify patients and OARs where DIR is unreliable. This method will aid in optimising reRT planning, enabling more effective dose mapping and increased confidence in the delivery of re-irradiation. We are working on the integration of geometrical and dose mapping uncertainties, as well as radiobiological uncertainties. Additionally, we're exploring individualised structure-wise registrations to improve dose mapping for challenging OARs in ReRT. Our aim is to enable direct execution of our method within RayStation’s scripting interface, which will allow for its direct use in clinical practice.

Keywords: Dose mapping uncertainties, re-irradiation, H&N