ESTRO 2021 Abstract Book

S587

ESTRO 2021

Conclusion We developed a high quality MVCT to kVCT conversion model which requires a few dozens of unsupervised (unpaired) training images. We showed the stability of the model against the reduced size of training dataset. This model can be applied not only MVCT to kVCT conversion but also other conversions such as MRI to CT and CBCT to Planning CT conversions. PD-0756 In-vivo assessment of tissue parameters with intra- and inter-patient variation using dual-energy CT N. Peters 1 , A. Kieslich 1 , P. Wohlfahrt 2 , C. Hofmann 3 , C. Richter 1,4,5,6 1 OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden - Rossendorf, Dresden, Germany; 2 OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden - Rossendorf, Dresden, Germany, Dresden, Germany; 3 Siemens Healthineers, Forchheim, Forchheim, Germany; 4 Helmholtz- Zentrum Dresden - Rossendorf, Institute of Radiooncology - OncoRay, Dresden, Germany; 5 Department of Radiotherapy and Radiation Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany; 6 German Cancer Consortium (DKTK), partner site Dresden, and German Cancer Research Center (DKFZ), Heidelberg, Dresden, Germany Purpose or Objective The precise and reliable knowledge of radiological tissue properties is essential for a multitude of radiotherapeutic applications, such as input parameters for Monte Carlo simulation, stoichiometric CT calibration as well as for tissue-mimicking phantoms. However, so far only limited data based on ex-vivo experimental studies is available, summarized in ICRU46. Utilizing a clinically validated Dual-Energy-CT (DECT)-based tissue characterization approach (DirectSPR), we present a precise in-vivo assessment of relative electron density (RED), effective atomic number (EAN) and stopping-power ratio (SPR) for organs in the head and pelvis in a patient cohort analysis of clinical DECT scans. Materials and Methods Organ-specific tissue parameters were obtained from clinical DECT scans of 107 brain-tumour and 103 pelvic cancer patients applying the DirectSPR approach. DirectSPR is characterised by a voxel-wise, patient-specific parameter determination based on a specific superposition of low- and high-energy DECT scans. In total, six structures in the head (brain, brainstem, spinal cord, chiasm, optical nerve, lens) and four in the pelvic region (prostate, kidney, liver, bladder) were investigated. To minimise contamination from surrounding tissues, clinical contours were shrunk and smoothed in 2D. Image slices with artefacts (e.g. due to metal implants) were omitted from analysis. Organ tissue parameters were characterised regarding the cohort mean value as well as the variation within each patient 2SD_intra and between the investigated patients 2SD_inter. Results For 10 organs, including 4 organs not listed in ICRU46, the mean RED, EAN and SPR as well as their respective intra- and inter-patient variation were determined (Table 1). Results are exemplarily illustrated for SPR, crucial for proton therapy planning (Figure 1). SPR intra-patient variation was higher than 1.4% (1.4-5.3%) in all organs and always exceeded the inter-patient variation of the organ mean SPR (0.5-2.0%). The largest