ESTRO 2022 - Abstract Book

S291

Abstract book

ESTRO 2022

Figure 1: A simulated tumor invasion shown in terms of cell concentration. V 0 and V correspond to the segmented and simulated GTVs, respectively. Conclusion Both analyses of c enc and d H (c) indicate poor agreement between the CTV delineation and the simulated tumor invasion. Tumor invasion modelling could therefore be a valuable tool for assisting the CTV delineation of HGGs.

PD-0328 Impact of image registration uncertainties on dose accumulations in online adaptive radiotherapy

A.B. Kristensen 1 , L.M. Åström 1,2 , C.E. Andersen 2 , P. Sibolt 1

1 Copenhagen University Hospital – Herlev and Gentofte, Dept of Oncology, Copenhagen, Denmark; 2 Technical University of Denmark, Dept of Health Technology, Roskilde, Denmark Purpose or Objective Modern re-optimization based online adaptive radiotherapy (oART) systems have for certain disease sites been demonstrated to improve target coverage and lower the dose to organs at risk (OARs) by adapting to the anatomy of the day. In order to evaluate the efficacy of the overall treatment, dose accumulation can be applied by propagating the dose from each adaptive fraction back to the planning simulation CT and summing the dose matrices. However, inherent registration uncertainties when relating the daily anatomy to the planning image result in dosimetric uncertainties, potentially affecting the clinical evaluation of the delivered treatment. Materials and Methods A total of 30 oART sessions for five patients with bladder cancer treated with 6 Gy per fraction, 6 fractions, to a total of 36 Gy on the Ethos oART system (Varian Medical Systems) were retrospectively collected . Data entailed DICOM files holding the structures delineated on patient planning CT (pCT), structures delineated on individual session images and the reconstructed dose from each session, before and after being propagated back to the pCT. Dose-Volume Histograms (DVH) were computed using the dicompyler-core package in python. Clinically relevant DVH parameters were extracted for each session using pCT-structures + propagated dose (DVH1), and session-structures + non-propagated dose (DVH2) . Differences in DVH parameters derived from DVH1 and DVH2, purely reflecting trends or uncertainties due to registration uncertainties, were evaluated relative changes towards DVH2. Results Also shown in Figure 1, the median (interquartile range, IQR) relative reduction in dose delivered to 99% of the clinical target volume (CTV99%) when comparing the propagated dose on the reference structure to the online adaptive dose was 0.01% (-0.02%, 0.16%). Similarly, for the planning target volume (PTV), the relative reduction in the dose delivered to 99% of the volume (PTV99%) and the maximum dose (PTV max ) are -3.17% (-6.23%, -2.02%) and -0.95% (-1.10%, -0,48%). Correspondingly, for the OARs, the critical DVH parameter V2.8Gy Rectum also showed a large relative decrease in median (IQR) of -11,7% (-35.3%, 3,8%) and V4.7Gy Bowel Bag showed a relative increase in the median (IQR) with 5.0% (-4.5%, 18.6%).