ESTRO 2024 - Abstract Book

S336

Beachytherapy - Physics

ESTRO 2024

Conclusion:

Model performance was comparable before and after the clinical introduction of auto-segmentation. An exception was found for the 95 HD of the bladder. Due to the high overall similarity between manually segmented and auto segmented bladder contours, combined with the sensitivity of 95 HD to outliers, it may not be a suitable choice for assessing the consistency of bladder segmentations, and we therefore conclude that model performance was as expected. A bias towards the auto-segmentation model output was introduced with the introduction of the auto-segmentation workflow. However, as the RTTs verify and if necessary edit the segmentations, the resulting edited structures lie within the variability of clinically acceptable segmentations. The median time taken by the RTTs to segment the four OARs decreased by half to about 15 minutes, demonstrating the value of the clinical introduction of auto-segmentation in the brachytherapy treatment planning of cervical cancer patients.

Keywords: Autosegmentation, Deep learning, evaluation

References:

[1] González, P. et al. (2021) PD-0927 MRI-based deep learning auto-contouring for organs-at-risk in gynecological brachytherapy. Radiotherapy and Oncology (vol 161-1). DOI: http://dx.doi.org/10.1016/S0167-8140(21)07206-6

978

Digital Poster

Dose model-based dose calculation have impact to dose distribution in brachytherapy?

Georgina Fröhlich 1,2 , Zsófia Jólesz 2 , Júlia Vízkeleti 1 , András Bajcsay 1 , Tibor Major 1,3 , Zoltán Nagy Takácsi 1,3

1 National Institute of Oncology, Centre of Radiotherapy, Budapest, Hungary. 2 Eötvös Loránd University, Department of Biophysics, Budapest, Hungary. 3 Semmelweis University, Department of Oncology, Budapest, Hungary

Purpose/Objective:

To investigate the impact of model-based dose calculation algorithm (MBDCA) around a high-dose-rate Ir-192 source in cervix and bronchus brachytherapy (BT).

Material/Methods:

Treatment plans of 30 intraluminal bronchus and 28 intracavitary cervix high-dose-rate BT were selected. The medium around the source is air in the bronchial and steel (metal applicator) in cervical cases. The dose calculation using the traditional water-based AAPM TG-43 formalism was recalculated with a model-based calculation algorithm (ACE, Advanced Collapsed cone Engine, TG-186 formalism) implemented in the same Oncentra Brachy v4.5 planning system (Elekta Brachytherapy, Veenendaal, The Netherlands). With MBBCA, the tissue heterogeneities and scatter conditions were taken into consideration with high accuracy. The dose-volume