ESTRO 38 Abstract book

S543 ESTRO 38

Conclusion This study has demonstrated feasibility of CTV delineation and volume selection based on CBCT for certain anatomical regions of the stomach. The large variations in these regions showed the urgent need for improvement of the current clinical practice and these findings justify further research towards the development and implementation of gastric ART. However, for other regions of the stomach, CTV delineation and volume selection were unreliable due to poor visibility on CBCT. Consequently, the potential benefit of gastric fiducial markers will be explored in a follow-up study. PO-0989 Deep learning improves robustness of contour propagation for online adaptive IMPT of prostate cancer M. Elmahdy 1 , T. Jagt 2 , R. Zinkstok 3 , C. Marijnen 3 , M. Hoogeman 2 , M. Staring 1 1 Leiden University Medical Center, Division of Image Processing- Department of Radiology, Leiden, The Netherlands ; 2 Erasmus MC Cancer Institute, Department of Radiation Oncology, Rotterdam, The Netherlands ; 3 Leiden University Medical Center, Department of Radiation Oncology, Leiden, The Netherlands Purpose or Objective Online-adaptive radiotherapy holds the promise to mitigate daily anatomical uncertainties thereby increasing treatment precision. One of the major challenges here is the development and validation of sufficiently fast, accurate, and robust segmentation algorithms for target volumes and organs at risk. The purpose of this study is to improve contour propagation in the pelvic region by combining deep-learning based auto-segmentation with image registration and to validate it geometrically and dosimetrically for online-adaptive Proton Therapy (PT) for prostate cancer. Material and Methods The proposed registration pipeline registers the daily CT scan to the planning, based on a combination of image intensities and, moreover, by an automatic segmentation of the bladder from the daily CT scan using a novel 3D convolutional neural network. The bladder was chosen for its influence on prostate motion. Registration performance is further enhanced by a digital inpainting of gas pockets with realistic bowel content, using a state-of- the-art generative adversarial network. We further

Results For all patients, the CTV could be delineated on the first 5 CBCTs. However, often no adjustments with respect to CTV_pCT were made in caudal, ventral and left-lateral direction because of poor visibility due to low image quality and changing volumes of gas in the stomach and colon (Fig 1A). All 15 CTV_CBCTs were partially outside the pCT_PTV, with on average 3.4% of the volume (range 0.0%–9.3%). PSV selection showed no apparent pattern over time (Fig. 2A). Overall, PSV_large was selected in 28 fractions (46.7%), PSV_medium in 20 fractions (33.3%) and PSV_small in 11 fractions (18.3%; Fig. 2B). For two patients, only PSV_medium and PSV_large were selected; for these patients, PSV_small equaled CTV_pCT. For only one fraction (1.7%), it was not possible to select a PSV.