ESTRO 37 Abstract book

ESTRO 37

S533

PO-0967 Quality assurance of 4DCT in the EORTC Lungtech trial on SBRT for patients with NSCLC M. Lambrecht 1 , J.J. Sonke 2 , U. Nestle 3 , H. Peulen 1 , D. Weber 4 , M. Verheij 2 , C.W. Hurkmans 1 1 Catharina Hospital, Department of Radiation Oncology, Eindhoven, The Netherlands 2 The Netherland Cancer Institute, Department of Radiation Oncology, Amsterdam, The Netherlands 3 Kliniken Maria Hilf, Department of Radiation Oncology, Mönchengladbach, Germany 4 Paul Scherrer Institute- ETH Domain, Center for Proton Therapy, Villigen, Switzerland Purpose or Objective To provide an overview of the 4DCT acquisition methods and achievable accuracy to image tumour volumes in institutions participation in the EORTC SBRT Lungtech trial. Material and Methods 3DCT and 4DCT images were acquired of a 008A CIRS phantom with 2 spherical inserts of 7.5 and 12.5mm diameter placed inside a cylinder of lung equivalent material using the institutional scan protocols. Regular asymmetric (cos 6 (t)) tumour motion was simulated with amplitudes A=15mm and periods t=3s and 6s and with A=25mm and t=4s. All CT scans were imported centrally in a treatment planning system to enable auto-contouring of the reconstructed sphere. Knowing the exact volume of the sphere, an institution dependent HU threshold was determined on the static CT images. This threshold was thereafter used for auto contouring the 4D-CT datasets. Acquisition parameters were saved and volume and amplitude deviation were assessed in comparison to the static volume. Results Acquisition parameters were rather comparable over the 11 institutions analysed (table 1). However, substantial inter-institution variations were found in the observed volume deviations. Average volume deviations for the 12.5mm sphere were 14.6%(-3.5% to 68.5%) at end of inspiration, 12.2%(0.05% to 36.0%) at mid-ventilation and 1.6%(-1.6% to 9.0%) at end of expiration. For the smaller 7.5mm sphere deviations were 13.1%(-98.7% to 64.8), 0.7%(-12.9% to 20.2%) and 16.3% (-33.8% to 66.4%), respectively. The amplitude deviation was in average within 2 mm (figure 1). However, underestimations up to 6.3mm were observed. No correlations were found between acquisition parameters and volume or amplitude deviations.

further improvements in 4DCT imaging are possible and local 4DCT QA could be improved.

PO-0968 Dual energy and iterative CBCT reconstruction allows reducing patient dose M. Gilles 1 , D. Benoit 2 , J. Bert 2 , A. Iborra 2 , U. Schick 3 , O. Pradier 3 , H. Fayad 1 , N. Boussion 4 , D. Visvikis 2 1 UBO - INSERM UMR 1101, LaTIM, Brest, France 2 INSERM UMR 1101, LaTIM, Brest, France 3 CHRU Morvan, Radiotherapy, Brest, France 4 CHRU Morvan - INSERM UMR 1101, LaTIM, Brest, France Purpose or Objective CBCT are increasingly used during radiotherapy treatment for patient positioning and adaptive treatment. These images induce a dose increase to the patients. The idea of this work is to reduce the delivered dose with no loss of information using dual energy CBCT acquisitions in combination with an iterative reconstruction algorithm. Material and Methods We first explored iterative reconstructed CBCT images (maximum likelihood gradient ascent optimization and patchwork-based reconstruction from Van Slambrouck and Nuyts, 2 iterations, 360 projections) to determine how much we can decrease count statistics while keeping distinguishable phantom inserts. We then used the dual energy method described by Saito et al. (2017) on these images in order to verify if the materials’ electron density (ρ e ) and effective atomic number (Z eff ) can be accurately retrieved from such low dose images. We used 80kVp and 140kVp (38kV and 65kV) for the CBCT acquisitions, in combination with materials mimicking CatPhan phantom inserts and body tissues. Results The number of counts considered varied from 40 to 2000 per pixel. Although 1000 counts resulted in correct insert distinction by plot profiles, we chose to use the 2000 counts images (10 to 20 times less than an actual state of the art clinical CBCT) to reduce overall image noise. We obtained similar errors for ρ e and Z eff than Saito et al.: 0.5% ± 0.4 vs. 0.4% ± 0.4 for ρ e and 2.0% ± 1.2 vs. 2.2% ± 1.6 for Z eff considering all inserts (calibration and body tissue like). The details of the results are summarized in the table. We obtained smaller errors for our adipose common insert, with ρ e = 0.1% and Z eff = 3.4% compared to ρ e = 0.8% and Z eff = 6.0% for Saito et al.

Conclusion Iterative reconstruction CBCT combined with dual energy acquisitions allows for a significant dose reduction (at least a factor 10) by decreasing the required count statistics. The ρ e and Z eff values required for segmentation and dose calculation can be obtained without a loss of accuracy. Further investigations will focus on the use of these images for treatment planning and direct CBCT based dosimetry applications.

Conclusion Our data suggest that the expiration phase is the most accurate phase to define the tumour volume and should therefore be preferred for GTV delineation when using a mid-position, gating or tracking strategy. The large variation found among the institutions indicated that