S1744
Physics - Dose prediction/calculation, optimisation and applications for particle therapy planning
ESTRO 2026
Trento, Trento, Italy. 4 Department of Medical Physics, MedAustron Ion Therapy Center, Wiener Neustadt, Austria. 5 Department of Physics & Astronomy, Louisiana State University (LSU), Los Angeles, USA. 6 Division Radiation Oncology, Karl Landsteiner University of Health Sciences, Krems an der Donau, Austria. 7 Division Medical Physics, Karl Landsteiner University of Health Sciences, Krems an der Donau, Austria Purpose/Objective: Linear energy transfer (LET) significantly influences RBE (relative biological effectiveness) weighted dose (DRBE) of carbon-ion therapy (CIRT). The RBE varies spatially (voxel-by-voxel) substantially within mixed- field of particles, and is described by the RBE models. Two RBE models (LEM-I: Local Effect Model-I, mMKM: modified Microdosimetric Kinetic Model) are commonly used for the CIRT dose calculation. LEM-I model compared mMKM tends to underestimate RBE near the distal portion of the beam (high-LET region) and overestimate it in proximal and mid-target regions (low-LET region) resulting in significant underdosage of the tumor in mMKM-based plans [1-5]. In this study, we evaluated the influence of different multi-RBE dosimetric (LEM-I, mMKM and LETd) parameters and clinical factors in predicting local relapses in non- squamous cell head and neck cancers (NSHNCs) treated with CIRT. Material/Methods: Fifty-five patients of NSHNCs treated with CIRT to a prescription of 68.8 (60.8-76.8) Gy (RBE)/16 fractions using pencil-beam algorithm and LEM-I optimization, with recomputed mMKM [multi-RBE optimization/evaluation]. The local relapses (LR: local progression/ local recurrence within PTV) were contoured by registering MRI (at first LR detection) with planning CT. Voxel-wise LETd and (LEM-I, mMKM) DRBE distribution was evaluated for GTV and LR (Fig 1). A predictive model was developed based on dosiomic features [PyRadiomics (v3.0.1) [6] of GTV and Random Forest [7-8] classifier (Fig 2). SHAP analysis [9] used to elucidate each feature's contribution to the model's predictions, and input-output relationship. Results: The median follow-up period was 24 (3-62) months. Nine out of 55 patients developed LR at median time of 20 (12-38) months. The actuarial 2-year LR-free survival was 82% (CI: 70-97%). GTV volume was larger in LR-group: 73.4±65.4 cm3, compared to non- relapsed group (No_LR-group):49.2±41.4 cm3, however, on univariate analysis none of the clinical factors were significant predictors of LR. In LR-group, 40% of GTV voxels received mMKM doses <57 Gy (RBE) with steep dose gradient and 10% of GTV voxels received concurrent low mMKM-doses and low LETd <42 keV/µm (Fig 1), despite adequate LEM-I doses. The
measures both the time-of-arrival and impact position of individual ions, enabling precise 4D track reconstruction and TOF-based energy-loss determination. Ion imaging experiments were conducted at the MedAustron research and therapy centre in Austria using helium ion beams, CIRS tissue- equivalent slabs, and a sacrificed mouse as imaging objects. System performance was evaluated in terms of timing precision, accuracy of the energy loss measurement and tracking efficiency. Results: The LGAD-based TOF-iCT system demonstrated excellent 4D-tracking capabilities, with an intrinsic time resolution of 30–50 ps per detector and a spatial resolution well below 100µm. By reconstructing ion trajectories and estimating their energy loss from their TOF, the first experimental TOF-based ion images of small animals were obtained using a custom image stitching algorithm. Conclusion: We developed and characterized a TOF-iCT demonstrator based entirely on ultra-fast LGAD detectors. The results demonstrate the potential of TOF-based ion imaging to improve treatment planning accuracy in ion beam therapy. Future work will focus on full tomographic reconstruction, in-vivo imaging experiments, and scaling the system toward clinically relevant geometries. References: [1] Schaffner, B et al. (1998). “The precision of proton range calculations in protonradiotherapy treatment planning: experimental verification of the relation betweenCT-HU and proton stopping power”. In: Physics in Medicine and Biology 43.6,pp. 1579–1592. DOI: 10.1088/0031-9155/43/6/016.[2] Ulrich-Pur, Felix et al. (2022). “Feasibility study of a proton CT system based on4D-tracking and residual energy determination via time-of-flight”. In: Physics inMedicine & Biology, 1361-6560. DOI: 10.1088/1361- 6560/ac628b. Keywords: Ion computed tomography, Time-of-flight Poster Discussion 2882 Predictive multi-RBE & LETd model for local relapse in non-squamous cell head and neck cancers treated with carbon-ion therapy using dosiomic features Ankita Nachankar 1,2 , Maddalena Paghera 3 , Francesco Cordoni 3 , Mansure Schafasand 4 , Eugen Hug 1 , Marta Missiaggia 3,5 , Domenico Attilio Romanello 1 , Antonio Carlino 4 , Piero Fossati 1,6 , Markus Stock 4,7 1 Department of Radiation Oncology, MedAustron Ion Therapy Center, Wiener Neustadt, Austria. 2 Research and Development, ACMIT Gmbh, Wiener Neustadt, Austria. 3 Department of Mathematics, University of
Made with FlippingBook - Share PDF online