ESTRO 38 Abstract book

S1153 ESTRO 38

however is the requirement of simulation volumetric imaging and image guidance imaging to be acquired in the upright position. In this work, we aim to acquire upright cone beam CT (CBCT) projections on a conventional digital linac and reconstruct images with an open-source toolkit. Material and Methods A phantom imaging study was conducted on a TrueBeam® linac (v2.5, Varian Medical Systems) in developer mode using a Catphan® 500 (The Phantom Laboratory). The kV imaging system was kept stationary at a gantry angle of 0° while the couch was rotated through a 180° arc. An extended field of view (FOV) was achieved by offsetting the kV detector in two half-rotation acquisitions (+13cm,- 13cm) in continuous fluoroscopy mode (125kVp,40mA,11fps) with no bowtie filter. The 4cm wide region where the pair of offset projections overlaps was used for normalisation between image pairs, at each angle the mean intensity of a central patch was matched. A vertical shift correction was calculated by matching vertical line profiles in the overlap region. The corrected offset projections were combined into a single image frame (Fig.1) and header information modified using custom software. The combined projections were reconstructed with the Reconstruction Toolkit (RTK) using an alternating direction method of multipliers with total variation regularisation (ADMM-TV) algorithm. A standard CBCT of the Catphan® taken during routine quality assurance is included for comparison.

Fig.2: Comparison of CBCT and Upright CBCT images of a Catphan® Reconstructed upright images (Fig.2) show a general agreement with the CBCT images. Expected intensity variations are observed in the spatial linearity module (CTP401) and some structures are visible in the high contrast resolution module (CTP528). The upright images suffer from ring artifact which may be reduced by median filtering or other methods. Further image improvements may arise from projection pre-processing methods and reconstruction options. Conclusion We have demonstrated upright CBCT can be acquired on a digital linear accelerator and reconstructed using an open source toolkit. Initial reconstruction results are promising, with future work to focus on improvement of image quality through projection space filtering. EP-2089 Technical implementation of total marrow irradiation (TMI) using VMAT J.M. Perez moreno 1 , C. Rubio Rodriguez 2 , J. Valero Albarrán 2 , P. Fernández Letón 3 1 Hospital HM Puerta del Sur, Radiofisica y Proteccion Radiologica, Mostoles, Spain ; 2 Hospital Universitario HM Sanchinarro, Radiation Oncology, Madrid, Spain ; 3 Hospital Universitario HM Sanchinarro, Radiofisica y Protección Radiológica, Madrid, Spain Purpose or Objective To implement LINAC based total marrow irradiation (TMI) using VMAT in the first clinical case. Material and Methods Our first patient VMAT TMI case is reported. Treatment planning was performed on RayStation 7 using a VERSA HD LINAC. Patient was immobilized using a head and neck mask and BodyFIX system. Two planning CT scans were performed to imaging whole target volume, head first and feet first supine (HFS and FFS). Target volume was defined in both scans and it included all bones in the body, whole brain, supraclavicular, axillary and paraaortic lymph nodes and spleen. A margin of 3 mm was used except in lower extremities and spleen, where 5 mm margin was used. Dose prescription was 12 Gy in 6 fractions twice a day. Four factors were considered in plan design: 1. body and target geometry 2. encompass full target while limiting beam size to 27 cm in head-feet direction 3. force an overlap of 5-8 cm between beams of adjacent sectors 4. geometric viability of treatment

Fig.1: (a) offset projections (b) combined corrected projection (c) line profile of (green) uncorrected and (purple) corrected projection. Results Attenuation correction and vertical shift correction between the two offset projections improved the extended FOV projection (Figure 1c). A gradient from left to right is observed in regions of air and is attributed to anode heel effect and static flood field corrections. This will be addressed using a measured average flood field image in future experiments.