Abstract Book

S1143

ESTRO 37

EP-2083 The Study of effect induced by respiration on CT radiomic feature extraction Y. Lu 1,2 , G. Gong 1 , Y. Yin 1 1 Shandong Cancer Hospital Affiliated to Shandong University, Department of Radiation Oncology, Ji'nan, China 2 Shandong Normal University, School of Physics and Electronics, Ji'nan, China Purpose or Objective To assess the effects of respiration on CT radiomic feature extraction based on four-dimensional computed Thirty-four thoracic cancer patients undergoing 4D-CT were studied retrospectively. 4D-CT scans were sorted into 10 phases according to breath cycle, i.e., 0%, 10%......90%. The 0% (end of inspiration), 20% (middle of expiration), 50% (end of expiration), and 70% (middle of inspiration) phases were selected. The left lung and right lung were selected as regions of interest (ROIs) for feature extraction. Hierarchical cluster methods were used to screen features, compare the features of different inflation and respiration states of the lung tissue, and quantitatively analyze the influence of respiration on radiomic feature extraction. Results Eighty-seven features were extracted from 4 phases of the two lungs, and 10 non-redundant features were selected after hierarchical clustering. The number of features that differed significantly between the 0% and 50% phases was 7 and 4 for left and right lungs, respectively, there were 4 common features and 3 uncommon features. The number of features that differed significantly between the 20% and 70% phases was 5 and 4 for left and right lungs, respectively, there were 4 common features and only 1 uncommon feature. The number of features that differed significantly between the 0% and 50% phases and the 20% and 70% phases was 2. Conclusion The respiration state caused a significant difference in certain CT radiomic features and the influence of breathing motions should be considered in feature extraction. EP-2084 Feasibility of CBCT positioning in an MRI only Prostate Treatment Planning A. Garcia Perez 1 , S. Reigosa Montes 1 , A. Lopez Medina 1 , A.G. Teijeiro Garcia 1 , J. Vazquez Rodriguez 1 , F.J. Salvador Gomez 1 , A. Gonzalez Castro 2 , M. Martinez Agra 2 , V.M. Munoz Garzon 2 1 Hospital do Meixoeiro, Radiofisica y PR, Vigo Pontevedra, Spain 2 Hospital do Meixoeiro, Radioterapia, Vigo Pontevedra, Spain Purpose or Objective The MRI-only workflow aims to base the treatment planning solely on magnetic resonance imaging (MRI), hence excluding the traditional computed tomography (CT) scan. Several studies 1-2 have mainly focused on the equivalence in calculated dose based on CT and synthetic CT (sCT or MRCAT: MR for Calculating ATtenuation, in Philips nomenclature). The aim of this study is to compare if any difference in patient positioning based on conventional CT or based on sCT is found, when the patient is accurately relocated based on Cone Beam CT (CBCT). Material and Methods We performed an MRI scan (Philips Ingenia 3T) and a conventional CT scan (Philips Big Bore) for two prostate patients. The first patient had two different treatment phases and they were considered independently because they had different isocentres. tomography (4D-CT). Material and Methods

into account, nevertheless they need data from imagining systems like CT characterizing the elemental composition of these inhomogeneities. Model-based iterative reconstruction (MBIR) algorithms in dual-energy and spectral CT can improve the accuracy of the predicted elemental composition compared to the currently used single-energy CT. Of interest is, how much should these algorithms be fine-tuned to predict the elemental composition for the RTP of prostate. The purpose of this work is to estimate the effect of zinc in prostatic calculi on the accuracy of elemental composition determined by the MBIR algorithm DIRA (A model-based iterative reconstruction algorithm DIRA using patient-specific tissue classification via DECT for improved quantitative CT in dose planning, Med. Phys. The accuracy of elemental compositions and mass densities of tissues predicted by DIRA was compared using computer simulations for two configurations: (i) a breast calcification according to ICRU report 46 and (ii) a calcification containing 8% of Zn; this value was obtained by the micro-PIXE method. In both cases, the material of the prostate was decomposed using the two-material decomposition method inside the iterative loop of DIRA to tabulated prostate tissue and calcium. Such material doublet was well suited for breast calcifications used in earlier studies. The calcifications were inserted to a mathematical anthropomorphic phantom derived from the ICRP 110 voxel phantom, where the size of prostate, bladder and rectum were adjusted according to a clinical case. Noise in projection data was not simulated. Results The omission of zinc in the base material doublet resulted in larger errors in both the linear attenuation coefficient (LAC) and the mass energy absorption coefficient (MEAC) for the calculi and the prostate tissue outside the calculi. For instance the error in MEAC in the calculi was increased from 3.9% to 6.6% at 30 keV when Zn was present. In the prostate tissue outside the calculi the increase in error was not so strong. 44, 2017, 2345–2357). Material and Methods

Conclusion Errors caused by the omission of zinc in the two-material decomposition of DIRA were present, nevertheless they did not break the algorithm. A more accurate material decomposition method of the prostate is desirable, nevertheless the tabulated prostate tissue and calcium doublet may be used in the meantime.