arXiv subjects

Miriam Schwarze

Publications and source records attributed to Miriam Schwarze.

8 recordsLinked to original sources

arXiv2025

Active learning-based variance reduction for Monte Carlo simulations: A feasibility study for the nanodosimetry around a gold nanoparticle

Objective: This work presents a data-driven importance sampling-based variance reduction (VR) scheme inspired by active learning. The method is applied to the estimation of an optimal impact-parameter distribution in the calculation of ionization clusters around a gold nanoparticle (NP). Here, such an optimal importance distribution can not be inferred from principle. Approach: An iterative optimization procedure is set up that uses a Gaussian Process Sampler to propose optimal sampling distributions based on a loss function. The loss is constructed based on appropriate heuristics. The optimization code obtains estimates of the number of ionization clusters in shells around the NP by interfacing with a Geant4 simulation via a dedicated Transmission Control Protocol (TCP) interface. Main results: It is shown that the so-derived impact-parameter distribution easily outperforms the actual, uniform irradiation case. The results resemble those obtained with other VR schemes but do still slightly overestimate background contributions. Significance: While the method presented is a proof-of-principle, it provides a novel method of estimating importance distributions in ill-posed scenarios. The presented TCP interface described here is a simple and efficient method to expose compiled Geant4 code to other scripts, written for example, in Python.

physics.med-ph↗

arXiv2025

Cluster Dose Prediction in Carbon Ion Therapy: Using Transfer Learning from a Pretrained Dose Prediction U-Net

The cluster dose concept offers an alternative to the radiobiological effectiveness (RBE)-based model for describing radiation-induced biological effects. This study examines the application of a neural network to predict cluster dose distributions, with the goal of replacing the computationally intensive simulations currently required. Cluster dose distributions are predicted using a U-Net that was initially pretrained on conventional dose distributions. Using transfer learning techniques, the decoder path is adapted for cluster dose estimation. Both the training and pretraining datasets include head and neck regions from multiple patients and carbon ion beams of varying energies and positions. Monte Carlo (MC) simulations were used to generate the ground truth cluster dose distributions. The U-Net enables cluster dose estimation for a single pencil beam within milliseconds using a graphics processing unit (GPU). The predicted cluster dose distributions deviate from the ground truth by less than 0.35%. This proof-of-principle study demonstrates the feasibility of accurately estimating cluster doses within clinically acceptable computation times using machine learning (ML). By leveraging a pretrained neural network and applying transfer learning techniques, the approach significantly reduces the need for large-scale, computationally expensive training data.

physics.med-ph↗

arXiv2025

Cross-Section-Based Scaling Method for Material-Specific Cluster Dose Calculations -- A Proof of Concept

Cross-section data unavailability for non-water materials in track structure simulation software necessitates nanodosimetric quantity transformation from water to other materials. Cluster dose calculation transformation initially employed mass-density-based scaling - an approach resulting in a physically unrealistic material-independence of the cluster dose equation. This study introduces an alternative scaling method based on material-specific ionization cross-sections. The mean free path ratio of the materials for both the primary particles of the track structure simulation and for the secondary electrons served as the scaling factor. The approach was demonstrated through a cluster dose calculation for a carbon ion beam in a realistic head geometry and compared to the previous scaling method. The proposed cross-section-based scaling method resulted in a physically expected increase in cluster dose values for denser materials, which was not visible in the original scaling approach. The introduced scaling approach can be used to determine cluster dose distributions in heterogeneous geometries, a fundamental requirement for its integration into radiotherapy treatment planning frameworks.

physics.med-ph↗

arXiv2025

Exploring Machine Learning Models for Physical Dose Calculation in Carbon Ion Therapy Using Heterogeneous Imaging Data -- A Proof of Concept Study

Background: Accurate and fast dose calculation is essential for optimizing carbon ion therapy. Existing machine learning (ML) models have been developed for other radiotherapy modalities. They use patient data with uniform CT imaging properties. Purpose: This study investigates the application of several ML models for physical dose calculation in carbon ion therapy and compares their ability to generalize to CT data with varying resolutions. Among the models examined is a Diffusion Model, which is tested for the first time for the calculation of physical dose distributions. Methods: A dataset was generated using publicly available CT images of the head and neck region. Monoenergetic carbon ion beams were simulated at various initial energies using Geant4 simulation software. A U-Net architecture was developed for dose prediction based on distributions of material density in patients and of absorbed dose in water. It was trained as a Generative Adversarial Network (GAN) generator, a Diffusion Model noise estimator, and as a standalone network. Their performances were compared with two models from literature. Results: All models produced dose distributions deviating by less than 2% from that obtained by a full Monte Carlo simulation, even for a patient not seen during training. Dose calculation time on a GPU was in the range of 3 ms to 15 s. The resource-efficient U-Net appears to perform comparably to the more computationally intensive GAN and Diffusion Model. Conclusion: This study demonstrates that ML models can effectively balance accuracy and speed for physical dose calculation in carbon ion therapy. Using the computationally efficient U-Net can help conserve resources. The generalizability of the models to different CT image resolutions enables the use for different patients without extensive retraining.

physics.med-ph↗

arXiv2024

Nanodosimetric investigation of the track structure of therapeutic carbon ion radiation. Part 1: Measurement of ionization cluster size distributions

At the Heidelberg Ion-Beam Therapy Center, the track structure of carbon ions of therapeutic energy after penetrating layers of simulated tissue was investigated for the first time. Measurements were conducted with carbon ion beams of different energies and polymethyl methacrylate (PMMA) absorbers of different thicknesses to realize different depths in the phantom along the pristine Bragg peak. Ionization cluster size (ICS) distributions resulting from the mixed radiation field behind the PMMA absorbers were measured using an ion-counting nanodosimeter. Two different measurements were carried out: (i) variation of the PMMA absorber thickness with constant carbon ion beam energy and (ii) combined variation of PMMA absorber thickness and carbon ion beam energy such that the kinetic energy of the carbon ions in the target volume is constant. The data analysis revealed unexpectedly high mean ICS values compared to stopping power calculations and the data measured at lower energies in earlier work. This suggests that in the measurements the carbon ion kinetic energies behind the PMMA absorber may have deviated considerably from the expected values obtained by the calculations. In addition, the results indicate the presence of a marked contribution of nuclear fragments to the measured ICS distributions, especially if the carbon ion does not cross the target volume.

physics.med-ph↗

arXiv2024

Nanodosimetric investigation of the track structure of therapeutic carbon ion radiation. Part 2: Detailed radiation transport and track structure simulation

Previously reported nanodosimetric measurements of therapeutic-energy carbon ions penetrating simulated tissue have produced results that are incompatible with the predicted mean energy of the carbon ions in the nanodosimeter and previous experiments with lower energy monoenergetic beams. The purpose of this study is to explore the origin of these discrepancies. Detailed simulations using the Geant4 toolkit were performed to investigate the radiation field in the nanodosimeter and provide input data for track structure simulations, which were performed with a developed version of the PTra code. The Geant4 simulations show that with the narrow-beam geometry employed in the experiment, only a small fraction of the carbon ions traverse the nanodosimeter and their mean energy is between 12 % and 30 % lower than the targeted values. Only about one-third or less of these carbon ions hit the trigger detector. The track structure simulations indicate that the observed enhanced ionization cluster sizes are mainly due to coincidences with events in which carbon ions miss the trigger detector. In addition, the discrepancies observed for high absorber thicknesses of carbon ions traversing the target volume could be explained by assuming an increase in thickness or interaction cross-sections in the order of 1 %. The results show that even with strong collimation of the radiation field, future nanodosimetric measurements of clinical carbon ion beams will require large trigger detectors to register all events with carbon ions traversing the nanodosimeter. Energy loss calculations of the primary beam in the absorbers are insufficient and should be replaced by detailed simulations when planning such experiments. Uncertainties of the interaction cross-sections in simulation codes may shift the Bragg peak position.

physics.med-ph↗

arXiv2024

Radial dependence of ionization clustering around a gold nanoparticle

This work explores the enhancement of ionization clusters around a gold nanoparticle (NP), indicative of the induction of DNA lesions, a potential trigger for cell-death. Monte Carlo track structure simulations were performed to determine (a) the fluence of incident photons and electrons in water around a gold NP under charged particle equilibrium conditions and (b) the density of ionization clusters produced on average as well as conditional on the occurrence of at least one interaction in the nanoparticle using Associated Volume Clustering. Absorbed dose was determined for comparison with a recent benchmark intercomparison. Reported quantities are normalized to primary fluence, allowing to establish a connection to macroscopic dosimetric quantities. The modification of the electron fluence spectrum by the gold NP is minor and mainly occurs at low energies. The net fluence of electrons emitted from the NP is dominated by electrons resulting from photon interactions. Smaller NPs cause noticeable peaks in the conditional frequency of clusters at distances around 50 nm to 100 nm from the NP surface. The number of clusters per energy imparted is increased at distances of up to 150 nm, and accordingly the enhancement in clustering notably surpasses that of dose enhancement. This work highlights the necessity of nanodosimetric analysis and suggests increased ionization clustering near the nanoparticles due to the emission of low energy Auger electrons. Whereas the electron component of the radiation field plays an important role in determining the background contribution to ionization clustering and energy imparted, the dosimetric effects of nanoparticles are governed by the interplay of secondary electron production by photon interaction (including low energy Auger electrons) and their ability to leave the nanoparticle.

physics.med-ph↗

arXiv2023

Article Commentary on "Microdosimetric and radiobiological effects of gold nanoparticles at therapeutic radiation energies" [T.M. Gray et al., IJRB 2023, 99(2), 308-317]

In the recently published article by T.M. Gray et al. "Microdosimetric and radiobiological effects of gold nanoparticles at therapeutic radiation energies" (IJRB 2023, 99(2), 308-317) results of Monte Carlo simulations and radiobiological assays on the dosimetric effects of gold nanoparticles were presented. This commentary points out that the results of the two parts of the study are in contradiction and that the predicted magnitude of dose enhancement and its dependence on the shape of the nanoparticle appear implausible. Possible reasons for these observations are discussed.

physics.med-ph↗