Abstract Book

S43

ESTRO 37

Dresden, Germany 8 National Center for Tumor Diseases NCT, partner site Dresden, Dresden, Germany Purpose or Objective Recent studies demonstrate the clinical reliability and improved accuracy of dual-energy CT (DECT) in proton therapy. Still, a generic heuristic conversion (HLUT) of CT number to stopping-power ratio (SPR) is used in clinical routine, since a medical device for patient-specific DECT- based SPR prediction is not yet available. Here, we propose an applicable method for HLUT optimization using information from patient-specific DECT-based SPR prediction on a broad patient cohort. Material and Methods Clinical DECT scans of 102 brain-, 25 prostate- and 3 lung- tumor patients were evaluated in total. Each scan was acquired with a single-source DECT scanner (Definition AS) and processed in syngo.via (both Siemens Healthineers) to generate 79keV pseudo-monoenergetic CT (MonoCTs) and SPR datasets (derived from electron density and photon cross section). Voxelwise correlations of CT number and SPR were determined within the irradiated volume (20% isodose) and expressed as frequency distribution including patient information of all 3 cohorts. A piece-wise linear function was defined minimizing the deviation from the median SPR distribution for each CT number (DECT-based adapted HLUT). The intra- and inter-patient variability was also obtained from the frequency distribution. To assess dose differences and range shifts, proton treatment plans were recalculated in XiO (Elekta) on MonoCT using (A) clinical or (B) adapted HLUT, and (C) patient-specific DECT-based SPR datasets. Results Mean range shifts (±1SD) of 1.2(±0.7)% for brain-, 1.7(±0.5)% for prostate- and 2.3(±0.8)% for lung-tumor patients were determined using the clinical HLUT instead of the patient-specific DECT-based SPR prediction. On average the clinical HLUT predicted larger SPR for brain, muscle and trabecular bone leading to this systematic range deviation. This effect is partially compensated in brain-tumor patients, since the clinical HLUT provides a smaller SPR for cortical bone. Using the DECT-based adapted HLUT (Fig. 1), mean range shifts were significantly reduced (p<<0.001, two-sample t-test) below 0.3% (Fig. 2). Hence, the adapted HLUT achieves a reduction of systematic deviations for all 3 tumor sites while standard deviations remained almost unchanged. Still, range shifts larger than 1% arise owing to the large intra-patient soft tissue diversity of approx. 6% (95% CI) and age-dependent inter-patient bone variation of 5%.

Conclusion We propose a set of elemental I -values, found via an analytical model, that are well suited for the use in human tissues in combination with the Bragg additivity rule. Our model establishes uncertainties on I -values which enables to quantify the resulting uncertainties on RSPs and particle range by taking co-variances into account. OC-0085 Improving CT calibration for proton range prediction by dual-energy CT based patient-cohort analysis P. Wohlfahrt 1,2 , C. Möhler 3,4,5 , W. Enghardt 1,2,6,7 , M. Krause 1,2,6,7,8 , E.G.C. Troost 1,2,6,7,8 , S. Greilich 3,4 , C. Richter 1,2,6,7 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 Helmholtz-Zentrum Dresden - Rossendorf, Institute of Radiooncology - OncoRay, Dresden, Germany 3 German Cancer Research Center DKFZ, Medical Physics in Radiation Oncology, Heidelberg, Germany 4 National Center for Radiation Research in Oncology NCRO, Heidelberg Institute for Radiation Oncology HIRO, Heidelberg, Germany 5 Heidelberg University, Department of Physics and Astronomy, Heidelberg, Germany 6 Department of Radiotherapy and Radiation Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus- Technische Universität Dresden, Dresden, Germany 7 German Cancer Consortium DKTK, partner site Dresden,