ESTRO 2025 - Abstract Book

S4392

RTT - Treatment planning, OAR and target definitions

ESTRO 2025

3887

Digital Poster Inter-fraction motion evaluation of the internal pudendal arteries during a stereotactic irradiation of prostate cancer using the 1.5T MR-Linac Stefania Stepanov, Corina Udrescu, Antoine Ailloud, Anda Ionita, Ciprian Enachescu, Vlad Galatan, Olivier Chapet Radioterapy-Oncology, Centre Hospitalier Lyon Sud, Lyon, France Purpose/Objective: Erectile dysfunction after radiotherapy (RT) remains an important problem in a majority of patients treated for prostate cancer (PCa). The MR-Linac allows real time visualization of the anatomical changes during the treatment. The internal pudendal arteries (IPAs) could play a key role in the erectile function, but their motion and the margins needed during adaptive planning are still unclear. The aim of this study was to assess the inter-fraction motion of the IPAs, between each fraction of RT. Material/Methods: A radiation oncologist, expert in pelvic cancer, delineated the IPAs on the pre-treatment (MRISim) and the five daily T2-MRI sequences from 5 patients previously treated with SBRT for PCa. The IPAs from each fraction were then copied onto the MRISim using translations only type of fusion on bone structures, to create a common volume (CV). This CV was compared with the volumes of 3 and 5 mm margins around the IPAs from MRISim to define the most adapted planning organ’s at risk volume. Three similarity coefficients were used to evaluate the IPAs position variations: the Jaccard, the Dice and the Szymkiewicz-Simpson coefficients. The Jaccard and the Dice coefficients are particularly useful in quantifying the similarity or overlap between two sets or binary data. Both coefficients yield values between 0 and 1, with 0 indicating no similarity and 1 indicating complete overlap (identical sets). The Szymkiewicz-Simpson coefficient specifically measures the overlap or commonality between two datasets, with 0 indicating no overlap and 1 indicating that one set is a subset of the other set. Results: The CV was systematically covered by the 5 mm margin around the IPAs. In Table 1 are presented the results for the similarity coefficients calculated between the CV and the 3 mm margin volume. The Dice coefficient was slightly higher than the Jaccard coefficient, due to it being less sensitive to set size differences. The Szymkiewicz–Simpson coefficient has almost a consistent value of 1 signifying a high level of commonality between volumes. The highest discrepancy was seen in the right IPA of patient 1, with the CV representing 81% for the 3 mm margins.

Conclusion: The values obtained from the similarity assessment suggest that the IPAs have minimal inter-fraction motion, and any discrepancies may be attributed to uncertainties in delineation and/or translation fusion uncertainties. During the MR-Linac RT, the inter-fraction motion of the IPAs was insignificant and considering a 3 mm margin should be sufficient.