ESTRO 2020 Abstract book

S1081 ESTRO 2020

followed the same bladder and rectum preparation: to urinate and drink 330ml of water and to empty the rectum 1 hour before planning CT and treatment. They were also instructed to use probiotics during treatment. Philips Brilliance Big Bore CT was used. VMAT / IMRT planning was performed with ECLIPSE (ACCUROS algorithm) V13.5. Treatment was delivered using 2100CD Clinac and TrueBeam linear accelerators (Varian) equipped with OBI. All patients were treated in supine position and with Combifix TM (CIVCO) immobilizers. A non‐action level offline setup protocol with five fractions was used. Orthogonal KV‐MV images were taken the first 5 days and then weekly. At least a CBCT was taken weekly. For this study, one dosimetrist supervised by a Radiation Oncologist delineated the OARs (bladder and rectum) in the CT and CBCTs for all patients. Rectal and bladder volumes and their center of mass with respect to the center of mass on planning CT, and for each CBCT, were recorded. Variations in volume and center of mass position were analyzed. Results The distribution of rectal and bladder volumes for all patients, during CT‐plan and treatment fractions is shown in figure 1. Although the inter‐fraction variation of rectal volume was low we observed that bladder volume significantly decreased during the treatment. The patients were unable to keep bladder filling as in the planning CT for the last half of the treatment.

Auto‐segmentation tools have been shown to reduce inter‐ and intra‐ observer variability in target delineation. Repeated target delineation over the course of RT is more frequently required due to a move towards adaptive strategies and it has recently been shown that target delineation inter‐observer variability increases during mid treatment adaptive RT. This research evaluates the geometric accuracy of an auto‐segmentation tool over the course of conventionally fractionated RT in LA‐NSCLC to assess if geometric agreement varies over time. Material and Methods 8 cases, with longitudinal 4DCT scans (n=44) acquired over the course of conventionally fractionated RT were assessed. Each 4DCT had a physician defined GTV created on a single breathing phase of the dataset. A target volume auto‐contour was created using Varian Smart Segmentation® and manually adjusted, if required, by a single user to create a corresponding user‐adjusted auto‐ contour (UA‐AC) for each scan. Geometric agreement of UA‐AC to the GTV was assessed using Dice Similarity Coefficient (DSC), centre of mass (COM) shift from GTV in cm and structure volume difference from GTV in cm 3 . Differences in the geometric accuracy measures from the first scan to the last of each patient were evaluated using a paired T‐Test or Wilcoxon Signed Ranks Test (depending on data normality tested using Shapiro Wilk) to assess the reliability of the UA‐AC over the duration of RT. Results Between 5 and 8 CT scans were available for each case. Scan dates ranged from 26 days pre‐RT to day 47 of RT. Median time difference between first and last scan was 47.5 days (range 27‐65 days). Median DSC at the first scan was 0.82 (range 0.76‐0.89) and 0.76 (range 0.49‐0.86) at last scan. Mean DSC decreased from 0.82 to 0.72 ( p = 0.05 paired T‐Test) between first and last scan. Median COM shift from GTV at first scan was 0.22 cm (range 0.09‐0.71 cm), at last scan was 0.23 cm (range 0.09‐ 1.99 cm). Mean COM shift increased from 0.31 cm to 0.52 cm ( p =0.058, Wilcoxon Signed Ranks Test). Median Volume difference between GTV and UA‐AC at first scan was 3.15 cm 3 (range 0.8‐32.4 cm 3 ), at the last scan was 1.0 cm 3 (range 0.7‐60.7 cm 3 ). Mean Volume difference increased from 6.6 cm 3 to 9.9 cm 3 ( p =0.866 Wilcoxon Signed Ranks Test). Conclusion No statistically significant changes in geometric agreement between the physician defined GTV and the UA‐AC were observed over the course of RT. However, all measures of agreement, except median volume difference, worsened at the later time point. Further evaluation is warranted on a larger dataset to increase the statistical power of the analysis. PO‐1846 Inter‐faction reproducibility of organs at risk in prostate and rectum RT treatments. N. Espinosa 1 , A. Coral Gallart 1 , S. Bermejo Martinez 2 , X. Nolla Nieto 1 , G. Gomez De Segura Melcon 1 , M. Mancera Soto 1 , M. Lizondo Gisbert 3 , N. Jornet Sala 3 1 Hospital de la Santa Creu i Sant Pau, Department of Medical Physics- Dosimetrist, Barcelona, Spain ; 2 Hospital de la Santa Creu i Sant Pau, Department of Radiation Oncology- Technician, Barcelona, Spain ; 3 Hospital de la Santa Creu i Sant Pau, Department of Medical Physics- Medical Physicist, Barcelona, Spain Purpose or Objective To check the reproducibility of bladder and rectum preparation and position in patients treated with radiotherapy in the pelvic area and analyze their inter‐ fraction motion. Material and Methods We included 63 patients with prostate and rectal cancer treated with external body radiotherapy. These patients

The distributions of the center of mass shifts in the three directions are shown in figure 2. The rectal center of mass had a larger motion antero‐posterior than for the other directions; these distensions maybe produced by the gas, so we should insist more on a correct diet. For the bladder, the axis antero‐posterior and craneo‐caudal axis present similar differences.