ESTRO 38 Abstract book

S1088 ESTRO 38

Purpose or Objective Craniospinal axis irradiation (CSI) is important in specific brain tumors like medulloblastoma. This is a challenging tumor site due to the large size of the fields and complicated patient positioning. Little is known about these patients’ inter and intrafraction motion, while these are important components of the PTV margin. The purpose of this study is to determine the magnitude of the inter and intrafraction motion for the thoracic and lumbar spine, and to compare these to the currently used PTV margins. Additionally, we investigate whether 2D kV/DRR position verification yields the same results as 3D CBCT/CT, as the 2D method would lead to lower imaging We retrospectively included 14 patients [median age 8.5y; range 5-14y] treated with CSI. The thoracic and lumbar CTVs were expanded with anisotropic margins of 0.8cm in LR direction, 1.0cm in caudal direction in the lumbar field and 1.0cm in AP direction for the thoracic and lumbar CTV. All patients were treated in supine position on a thin matrass and knee support, immobilized with a 5 points mask. For clinical position verification, pre and post treatment CBCTs were acquired of the thoracic and lumbar CTV, using a hybrid offline/online protocol. Average time between the pre and post fractional CBCTs was 22min [range 17-33min]. In total 180 pre and 154 post fraction CBCTs were analyzed. Inter and intrafraction motion in LR, CC and AP directions were determined based on clinically used CBCT/CT automated matches of the bony anatomy. Distributions of systematic and random errors (standard deviations Σ and σ, respectively) were calculated. PTV margins were calculated using inter and intrafraction motion, without considering other sources of error. We simulated 2D kV/DRR position verification by extracting two orthogonal radiographs from the CBCT series under angles with optimal visibility of the target. These were matched manually to the bony anatomy on DRRs of corresponding angles from the planning CT. The corrections resulting from these matches were compared to the corrections from the 3D CBCT/CT matches and Pearson correlation coefficients were calculated. Results The systematic and random errors and resulting PTV margins for inter and intrafraction motion are shown in table 1 for the 3D CBCT/CT method. The clinically used PTV margins are sufficient for these uncertainties. In figure 1 the corrections from the 2D matches are plotted versus those from the 3D matches for all directions and linear regression lines are shown. A moderate to strong correlation was found as indicated by the Pearson correlation coefficients. dose for the patients. Material and Methods

Conclusion The current PTV margins are sufficient to incorporate the inter and intrafraction motion measured on the CBCT for both the thoracic and lumbar CTV. The moderate to strong correlation between the 2D and 3D matches implies that larger margins may be needed when using 2D position verification to account for the additional uncertainty. EP-1992 Dose warping protocol for interfraction variation in Bladder filling in ano-rectal cancer patients C. Fiandra 1 , E. Trino 1 , F. Arcadipane 2 , F. Olivero 1 , U. Ricardi 1 , P. Franco 1 1 University of Turin, Oncology, Turin, Italy ; 2 Città della Salute e della Scienza, Radiation Oncology, Turin, Italy Purpose or Objective During pelvic radiotherapy (RT), bladder filling may randomly change during the course of treatment. The more reliable approach to account for this variation is represented by Plan of the Day approach; This relies on the robustness of deformable image registration (DIR) algorithms and software to match dissimilar volumes in different scansets. The main purpose of this work was to investigate a framework with 2 different algorithms to validate DIR performed between planning CT and each CBCT to explore the relationship between the “real” dose received by the Bladder and its volume change. Material and Methods Ten patients were enrolled in the study; a total of 233 CBCTs were analyzed and contoured. Bladder, femoral heads and pubic symphysis were mapped from planning CT (pCT) to each CBCT and then verified by physicians. Each contour was then mapped back on the planning CT by means of specific deformable vector field (DVF) using both Hybrid and Biomechanical DIR algorithms. The goodness of the registration was double-checked first by visual inspection of each registration by physician and then by structures correspondence using Dice similarity coefficient (DSC) and Hausdorff distance (HD). The plan was then recalculated on each CBCT and dose warping was applied to accumulate dose on the planning CT with both algorithms, only in case of DSC ≥ 0,8 and HD ≤ 3mm. Each re-calculated and deformed dose was used to evaluate the average dose (AD) and dose received by 2% volume (D 2 ) of the Bladder for each treatment session and their correlation with percentage variation of volume of the Bladder (PVB) was investigated; Spearman's coefficient of rank correlation (r) was calculated for both algorithms for AD and D 2 in function of PVB Results Percentage variation of Bladder volume compared to the pCT was -28.7±39% showing a tendency for a smaller bladder volume during treatment than simulation CT. Figure 1 shows a box plot of the percentage variation of volume of bladder for each patient while Table 1 reports mean values and standard deviation of DSC and HD indexes for all structures for both algorithms; r