ESTRO 2020 Abstract book

S998 ESTRO 2020

of the DECT data is visibly noisier than the SECT. The noise also causes the red dots in the surroundings of the phantom, which represent pixels that lack a real solution. The iterative DECT solution was tuned to show a comparably noisy image to SECT; this was achieved without degrading the sharpness of transitions between materials. The analytically non-solvable pixels can also be handled. The standard deviations within a central ROI in the insert were 0.0166, 0.0498 and 0.0169 for the SECT, the DECT analytic and the DECT iterative, respectively.

Fig. 2 a) Dosimetric impact of MR inaccuracy on liver and the spinal cord relative to the size of the target. Conclusion The current study is a preliminary study aimed at putting in place an procedure allowing addressing the dosimetric impact of image deformation with any anatomy of interest. Further investigations involving more liver cancer patients will be carried out. In addition, using this procedure, the impact of the MR geometric accuracy will be evaluated for different treatment techniques and on patients with different cancer sites. PO-1713 An iterative approach to reduce noise in dual-energy CT (DECT) for radiotherapy treatment planning J. Zimmerman 1 , G. Poludniowski 2 1 Karolinska University Hospital, Radiotherapy Physics and Engineering, Stockholm, Sweden ; 2 Karolinska University Hospital, Medical Radiation Physics and Nuclear Medicine, Stockholm, Sweden Purpose or Objective DECT has been a promising technique to improve dose calculation accuracy in radiotherapy for several years. The aim is to get a more accurate representation of a patient’s composition, compared to single-energy CT (SECT). Several publications (Bär et al. 2017 gives a good overview) have shown the potential to improve range prediction for proton beams within an object, but artifacts and noise can degrade the benefits. In this work we demonstrate how iterative reconstruction of DECT may help overcome the noise issue. Material and Methods A Siemens Defintion AS+ scanner was used, calibrated for proton therapy treatment planning using a stoichiometric approach. All calibration measurements were done with a CIRS 062M phantom with 10 different tissue substitute inserts, including a vial of water. A parameter model with two free parameters was used, establishing a relationship of CT-number HU as a function of electron density ρ e and effective atomic number Z eff relative to water. For the DECT work, 140 kVp and 80 kVp protocols were fitted to the model to find the respective pair of parameters. A DECT image set could then be solved for ρ e and Z eff through the fitted equations. Although this problem is analytically tractable, here it was also solved with iterative methods under the constraint of a penalty function to suppress the noise. The results presented are for a Total Variation penalty, which is known to suppress noise while preserving edges between materials. Results Image 1 shows stopping-power ratio (SPR) images, for a proton beam of 100 MeV, of the CIRS phantom with the “Bone 200” insert for a) theoretical calculations from tabulated chemical composition b) SECT stoichiometric calibration c) analytic solution of the DECT problem and d) iterative solution of the DECT problem with penalty. In image 2 the line profiles are plotted, following the red lines in each of the plots in image 1. The analytic solution

Conclusion No firm conclusion should be drawn from the difference in average SPR between the SECT, the DECT and the composition derived data, since both the SECT and DECT stoichiometric calibrations are made for theoretical human tissues whereas the CIRS materials are not closely tissue-like. Further, the geometry studied is simplified compared to a human body. However, the present results demonstrate the feasibility of the approach to solve the DECT problem. Reference Bär E, Lalonde A, Royle G, Lu HM, Bouchard H, Med Phys. 2017;44(6):2332-2344 PO-1714 Multimodalities imaging PET/CT-MR for radiotherapy: study of positioning uncertainties P. Hinault 1,2 , I. Gardin 1,3 , P. Gouel 1,3 , P. Decazes 1,3 , H. Souchay 2 , P. Vera 1,3 , D. Gensanne 4 1 University of Rouen Normandie, Laboratoire d'informatique de traitement de l'information et des systèmes- LITIS EA 4108, Rouen, France ; 2 General Electric, Healthcare, Buc, France ; 3 Centre Henri Becquerel, Medical Imaging and Nuclear Medicine Department, Rouen, France ; 4 Centre Henri Becquerel, Radiotherapy Department, Rouen, France Purpose or Objective CT scan is considered as the reference imaging for radiotherapy treatment planning, giving the right geometry with a high spatial resolution electronic density information need for dosimetry calculation. Other image