ESTRO 38 Abstract book

S1102 ESTRO 38

This approach can result misleading in clinical situations where the inter-fraction variability, the different organs filling and the changes in patient’s anatomy (i.e. weight loss) can lead to considerable variations in RED map. A possible approach to overcome this variability consists in the daily generation of a synthetic CT (sCT) by segmenting the dMRI in 5 density levels (air, lung, fat, tissue and bone) and assigning a RED bulk value to each level according to ICRU 46 recommendations. Aim of this study is to evaluate the dose calculation accuracy of this approach in MRgART and to evaluate if assigning patient specific RED values can improve such accuracy. Material and Methods 26 patients treated in the pelvic and abdominal sites were retrospectively enrolled. For each patient, a planning CT (pCT) was acquired and segmented in 5 density levels, then median RED (mRED) values were calculated for fat, soft tissue and bone. Correlation between mREDs and clinical parameters (age, sex, body mass index) was investigated by using the Pearson Correlation Coefficient (PCC). Two sCTs were generated by assigning RED bulk values to the segmented levels on pCT: the sCT ICRU uses the RED values recommended by ICRU 46, the sCT tailor uses the median patient specific RED values. The sCTs were generated from the CT image and not from MRI in order to avoid any potential bias due to the different MR and CT patient positioning. The same treatment plan was calculated on the sCTs and compared to those calculated on pCT in terms of 1%/1mm gamma analysis and dose volume histogram (DVH) parameters.The statistical significance of the difference between the sCTs was evaluated using Wilcoxon Mann Whitney test. Results A significant correlation between the clinical parameters and the mRED values was observed only between bone and age in women (PCC=-0.71), probably due to osteoporosis and menopausal status of the enrolled women (Figure 1).

Conclusion The use of bulk synthetic CT achieves a high level of dose calculation accuracy, allowing its reliable use for online adaptive MR-guided Radiotherapy. In particular, assigning patient specific RED values to sCT allows to improve the accuracy of this approach. EP-2012 MR-guided online adaptive radiotherapy for pancreatic cancer: where are we and where are we going? L. Placidi 1 , D. Cusumano 1 , L. Boldrini 2 , V. Chiloiro 2 , S. Teodoli 1 , A. Capotosti 1 , S. Manfrida 2 , F. Cellini 2 , L. Azario 1 , M. De Spirito 1 , V. Valentini 2 1 Fondazione Policlinico A. Gemelli IRCCS - Università Cattolica Sacro Cuore, Medical Physics, Rome, Italy ; 2 Fondazione Policlinico A. Gemelli IRCCS - Università Cattolica Sacro Cuore, Radiation Oncology, Rome, Italy Purpose or Objective Magnetic Resonance guided -Radiotherapy (MRgRT) represents to date the most suitable IGRT technique for Online Adaptive (OA) RT, due to the superior soft tissue contrast and the real-time gating approaches achievable using cineMRI images. Aim of this study is to perform a process optimization analysis for the evaluation of the robustness and reliability of the OA workflow implemented in our institution for locally advanced pancreatic cancer (LAPC) patients (pts) undergoing low tesla MRgRT. Material and Methods Our OA workflow is characterised by different steps: for each fraction, OaR re-contouring is performed within a distance of 3 cm from the PTV surface on the daily acquired MR image scan. The dose distribution of the original plan is firstly calculated on the anatomy of the day and, if the dose constraints are not met, treatment plan is re-optimised. Online Quality Assurance (QA), based on an independent Monte Carlo (MC) calculation, is finally performed before delivery starts. Ten (pts) affected by LAPC were retrospectively enrolled after being treated with the OA workflow, for a total amount of 50 delivered fractions. Several parameters have been registered for the analysis: single fraction couch shifts after daily MRI registration, OaRs and target volumes variation, number of delivered OA fractions with a predicted or re-optimised dose, beam-on time for each adaptive fraction, online QA 3%/3mm gamma passing rate, DVH metrics (target coverage as PTV V95% and CTV V98%, and OaRs specific dose-volume constraint) and single fraction distance between CTV and duodenum/stomach centre of mass. Results Lateral, longitudinal and vertical single fraction couch shifts result in (mean±SD (range min/max) for all patients, for all 50 fractions) -0.18±0.58cm (-1.49/0.96cm), - 0.11±0.72cm (-2.81/1.38cm), -0.02±0.31cm (- 0.80/0.68cm) respectively. OaRs and targets volume variation are summarised in Figure 1. Out of a 50 fractions, 34 fractions were re-optimised and 16 delivered using the

For the dosimetric analysis, high agreement was found between dose calculated on sCTs and pCT: γ passing-rate was 91.2% ± 6.9% for sCT ICRU and 93.7% ± 5.3%for sCT tailor . A statistically significant gain in using sCT tailor respect to sCT ICRU46 was found (p = 0.0013). As shown in Figure 2, in 8 cases out of 26 the use of a personalised approach lead to an improvement in dosimetric accuracy higher than 4%.