ESTRO 38 Abstract book

S241 ESTRO 38

computed tomography imaging and patient motion, particles stop at different geometric locations during treatment than predicted during dose calculation. This may heavily compromise the quality of the computed treatment plan. Conventional radiation therapy planning does not explicitly model this range uncertainty, usually relying only on a single “best-guess” scenario for dose optimization. Probabilistic optimization, in contrast, may incorporate explicit assumptions about the probability distribution over range uncertainty. Thereby it is possible to reduce a potentially deteriorating impact of uncertainties within a structured probabilistic optimization process. The first part of the presentation will give a thorough introduction to the fundamentals of probabilistic optimization in radiation therapy planning. This includes (1) intuitive visualizations on how uncertainty in particle ranges translate into uncertainty in the dose, depending on different correlation and fractionation assumptions, and (2) a brief overview of facilitated objective functions for probabilistic optimization. Within the second part of the presentation we will discuss advantages and disadvantages associated with probabilistic optimization in radiation therapy as compared to conventional optimization and robust (“worst case”) optimization. In particular, this includes an illustration of the challenges associated with probabilistic and robust optimization regarding the exact control of the robustness level of the optimized treatment plans. In the third part of the presentation we will elaborate on prerequisites for exact uncertainty control within probabilistic radiation therapy optimization with regard to (1) the underlying uncertainty models, (2) the facilitated uncertainty quantification pipeline, and (3) the mathematical formulation of the uncertainty minimization algorithms. We will show how this issue is currently addressed within the academic community using e.g. Gaussian Processes, Polynomial Chaos Expansion, and Analytical Probabilistic Modeling. V. Taasti 1 , G.J. Michalak 2 , A.J. Deisher 3 , J.J. Kruse 3 , C.H. Mccollough 2 , L.P. Muren 4 , J.B.B. Petersen 4 , D.C. Hansen 4 1 Memorial Sloan Kettering Cancer Center, Medical Physics, New York, Usa; 2 Mayo Clinic, Radiology, Rochester, Usa ; 3 Mayo Clinic, Medical Physics, Rochester, Usa; 4 Aarhus University Hospital, Medical Physics, Aarhus, Denmark Dual energy CT (DECT), Multi energy CT (MECT) and Photon Counting Detector CT (PCD-CT) have been investigated for their ability to improve the accuracy of the stopping power ratio (SPR) estimation, which is used in the treatment planning process for proton therapy. Especially DECT has been intensively investigated. Most DECT based methods have been shown to provide better results for the SPR estimation than the conventional method based on single energy CT (SECT). The CT number and the SPR are both proportional to the electron density of the tissue whereby they are also proportional to each other. This is exploited in the standard SECT based SPR estimation. However, the proportionality factor is not constant. This has been considered by making a piecewise linear fit between the single energy CT number and the SPR, approximating the proportionality factor with a local constant. Using DECT or even MECT and PCD-CT this approximation is not needed, directly, but approximations are still exploited. Some of these SPR methods are more empirical than others, but often their parameter fitting is Abstract text Introduction SP-0470 Multi-energy CT for improved SPR determination: Proposed methods and their experimental validation

outcome. So far, measurements have been performed in ten institutes. With the QA procedure we were able to identify inaccurate results and optimize the sequences such that consistent results were obtained. Similar results were obtained with measurements on MR-linac Unity systems, where we compared the repeatability and reproducibility in comparison to standard diagnostic systems. SP-0468 Quality assurance and validation for Quantitative PET in Multicenter Trials M. Lubberink 1 1 Uppsala University, Department Of Surgical Sciences- Radiology, Uppsala, Sweden Abstract text In recent years, a lot of effort has been put into harmonization of clinical PET procedures with 18 F- fluorodeoxyglucose (FDG) in oncology. For example, the European Association of Nuclear Medicine’s EARL accreditation program aims to ensure comparable quantitative results among patients and sites, and as such provide quality assurance for multi-center trials. This lecture will cover the use quantitative imaging with PET- CT and PET-MR, especially related to the use of PET in multi-center trials concerning dose planning and treatment response monitoring. First, the concept of quantitative PET will be discussed. PET is unique amongst medical imaging techniques in its ability to absolutely measure a wide range of molecular and physiological biomarkers. These quantitative measurements are based on time-varying measurements of radioactivity concentrations and the tracer kinetic models that describe the in-vivo behaviour of PET tracers. This will be illustrated using its most common application, measurement of glucose metabolism using FDG. The simplifications necessary to use these methods clinically in radiation oncology will be addressed, such as standardized uptake values (SUV), and the challenges associated with them both in terms of biology and instrumentation. Then, the harmonization efforts to ensure comparable SUV measurements between different scanners and hospitals will be discussed. PET-MR provides additional challenges in quantification associated with the limitations of MR-based attenuation correction. Improvements in quantification using recent developments in instrumentation, such as digital PET detectors and time-of-flight, and novel image reconstruction methods, will be reviewed in the context of multi-center studies in oncology. Finally other tracers than FDG and their possible quantitative applications in radiation oncology will be presented, such as those measuring hypoxia, proliferation, blood flow and oxygen consumption. SP-0469 Mitigation of range uncertainties with probabilistic IMPT optimization M. Bangert 1,2 , N. Wahl 1,2 , H. Wieser 2,3,4 1 German Cancer Research Center Dkfz, Medical Physics In Radiation Oncology, Heidelberg, Germany ; 2 Heidelberg Institute Of Radiation Oncology Hiro, Heidelberg, Germany; 3 German Cancer Research Center, Medical Physics in Radiation Oncology, Heidelberg, Germany; 4 Heidelberg University, Faculty For Medicine, Heidelberg, Germany Abstract text At the time of planning it is impossible to estimate the range of charged particles within a patient with absolute certainty. Due to, among others, inevitable limitations in Symposium: Advanced methods to account for proton range uncertainties in treatment planning