ESTRO 37 Abstract book

S1129

ESTRO 37

2 Royal Marsden NHS Foundation Trust, Radiotherapy, Sutton, United Kingdom Purpose or Objective The soft tissue contrast provided by ultrasound (US) may be useful for assessing interfractional uterine motion. Here, we compare the uterus image quality on cone beam CT (CBCT) and US. Material and Methods The uterus of 11 cervical cancer patients was scanned with US using Clarity® (Elekta Ltd) and CBCT (Elekta Ltd) at 4 to 6 fractions during radiotherapy (RT) (59 image pairs acquired). 8 observers (blinded to US and CBCT pairings) were asked to perform the following tasks on both modalities: T1. Contour the uterus on a pre-selected central slice T2. Rate the certainty that the contour drawn on T1 reflected the true uterine anatomy (scale: 1 to 10) T3. Rate the percentage of the uterine boundary clearly visible (scale: 1 (0 – 20%) to 5 (80 – 100%)) All observers received training on one third of the data (randomly selected) for 1 to 2 hours for this study. 3 observers had previous experience in analysing US images (Group 1), 5 had previous training in analysing CBCT images, and 3 had no prior training. The remaining two thirds of the data were used for the final analysis. For each image, a ground-truth (GT) contour was generated using the Group 1 contours and Simultaneous Truth And Performance Level Estimation (STAPLE). The dice similarity coefficient (DSC) was measured between each of the 5 remaining observers and the GT contour. Pairwise DSC analysis was performed to determine the agreement between individual observers in Group 1. Wilcoxon signed rank tests were used to measure differences between US and CBCT DSCs (T1) and observer ratings (T2 and T3). Results T1: The median [interquartile range (IQR)] DSC between the GT contour and each of the 5 observer contours was 0.81 [0.14] and 0.84 [0.17] for CBCT and US respectively (p = 0.76). The median [IQR] pairwise DSC between observers in Group 1 was 0.82 [0.13] and 0.81 [0.14] for CBCT and US respectively (p = 0.97). (Fig 1a) T2: Observers were more confident in their contours drawn on US compared with CBCT (median ratings of 7 vs. 5, p < 0.0001, Fig 1b). T3: The observer ratings for the percentage of the uterine boundary clearly visible were greater in US than in CBCT (median of 40 to 60% vs. 20 to 40%, p < 0.0001, Fig 1c).

the IMPT plans compared to that planned with the VMAT technique (Figure 1). The mean relative intrafraction bladder volume difference was 50,2% (range 19,2% - 87,6%) and the mean relative interfraction bladder volume difference was 158,0% (range -4,9% - 726,3%).

Conclusion Robust optimized IMPT treatment plans for cervical cancer patients show equivalent target coverage robustness when compared to VMAT treatment plans, but offer better OAR sparing. For the bladder volume, the interfraction variability was larger than the intrafraction variability. Large bladder volume differences between simulation and delivery of treatment fractions result in degraded dose coverage of the ITV for both IMPT as well as VMAT treatment plans, showing that bladder volume should be maintained throughout the delivery of treatment fractions. EP-2057 Comparison of ultrasound and CBCT image quality for image guided radiotherapy for cervical cancer S. Mason 1 , I. White 2 , T. O'Shea 1 , H. McNair 2 , S. Alexander 2 , J. Bamber 1 , S. Lalondrelle 2 , E. Harris 1 1 Institute of Cancer Research, Physics, Surrey SM2 5NG, United Kingdom

Conclusion There was no difference in interobserver agreement on US and CBCT, which was high for both modalities (DSC > 0.8). Observers reported greater confidence in interpreting US images and that a greater portion of the uterine boundary was clearly visible on US compared to