KAHIBARO
12.7. Choosing Physics Models
Table of Contents
Medical imaging
For medical imaging simulations such as PET, SPECT, gamma cameras, and CT, you usually need detailed electromagnetic interactions for photons and electrons, but extreme precision at very low energies or for rare processes is often unnecessary. The aim is to reproduce image quality, contrast, and detector response with reasonable computation time.
In GATE, physics lists are typically chosen from Geant4 “reference physics lists” and optionally extended with specific options. For gamma and x ray imaging, photon interactions must be modeled accurately around typical energies:
PET: annihilation photons at 511 keV, with some higher energy photons from radionuclide decay.
SPECT and planar gamma cameras: radionuclide specific lines, for example, 140 keV for Tc 99m, 159 keV for I 123, or multiple lines for In 111.
CT: tens to hundreds of keV, depending on the x ray tube voltage and filtration.
Common Geant4 based choices, exposed in GATE, include standard electromagnetic physics (for general applications), Livermore models, and Penelope models. Livermore and Penelope models provide more precise low energy electromagnetic cross sections, especially below a few hundred keV, at the cost of more computation. For PET at 511 keV, the standard electromagnetic physics is usually sufficient for most imaging tasks. For SPECT and diagnostic x ray energies, Livermore or Penelope physics are often preferred because they describe photoelectric absorption and coherent scattering at low energies more accurately.
When choosing a physics list for imaging in GATE, the main questions are the photon energy range, the importance of low energy scatter, and whether you are focusing on detector response or patient dose. If you mainly care about detector count rates, energy spectra, and spatial resolution, a standard EM physics list may be enough for PET. If you need detailed modeling of scatter in the patient or collimator, for example in SPECT system design or CT scatter studies, then a low energy physics option is preferable.
Hadronic physics is usually not required for pure gamma and x ray imaging simulations, because there are no primary hadrons. It becomes relevant only if you include neutron or proton backgrounds or activation, which is rare in standard nuclear medicine or diagnostic radiology simulations. In most imaging studies, you can safely disable hadronic physics to improve performance.
The choice of physics models in imaging strongly affects scatter fractions and energy spectra. For example, Compton scattering distributions and photopeak to scatter ratios in SPECT depend on the low energy model. Before finalizing, it is good practice to compare simulated spectra and scatter fractions against experimentally measured data from your scanner. If you see systematic differences around the photopeak or in the low energy tail, a different low energy EM model or updated cross section library may be appropriate.
Finally, consider production cuts and transport thresholds together with the physics list. For imaging, you usually do not need to track very low energy secondary electrons, as long as photon interactions and deposited energies in the detector crystals are correct. Setting realistic range cuts avoids generating a huge number of unnecessary low energy secondaries and speeds up simulations. Production cuts are treated in detail elsewhere, but they should always be consistent with your choice of physics list and with the spatial resolution required for your detector model.
For imaging simulations, choose low energy electromagnetic models (such as Livermore or Penelope) for diagnostic energy photons (typically below a few hundred keV), and use standard electromagnetic physics for 511 keV PET photons when ultra low energy precision is not required.
Dosimetry
In dosimetry, the primary goal is to calculate absorbed dose and related quantities accurately. That requires precise modeling of energy deposition from charged particles, especially electrons, in tissue, bone, and other materials. Even if your primaries are photons or protons, most of the dose is ultimately delivered by secondary electrons, so the electron transport models and production cuts are crucial.
For external beam photon radiotherapy and diagnostic dose assessments, the choice of electromagnetic physics is more demanding than in basic imaging. Detailed low energy electron and photon interactions affect the shape of depth dose curves, build up regions, and surface doses. Physics lists that include advanced multiple scattering and ionization models are recommended. In GATE, this typically means using a physics configuration explicitly intended for medical applications, which relies on Geant4 EM “option” variants that improve step limitation and multiple scattering, especially in the 1 keV to several MeV range.
For internal dosimetry, radionuclides emit multiple radiation types, including beta particles, gamma rays, and sometimes alpha particles or conversion electrons. The dose to small structures, such as organs or small tumors, can be very sensitive to the modeling of short range electrons and low energy photons. Low energy electromagnetic physics (Livermore or Penelope) is often chosen for internal dosimetry, particularly when dose gradients are steep or when voxel size is small in image based dose calculations.
Hadronic physics can become relevant when modeling neutron production, heavy ion components, or specific radionuclide decay schemes that involve hadronic secondaries. However, in most conventional photon and beta emitting radionuclide therapy simulations, hadronic interactions have a minor contribution to dose and are often neglected to save computation time. For radionuclide therapies that include alpha particles, or when simulating proton induced activation for secondary dose, you must include appropriate hadronic and ion physics models. Those details are covered in specialized chapters, so here it is sufficient to note that hadronic physics choices should match the energy and type of any hadronic component that significantly contributes to dose.
Production cuts are tightly coupled to the physics models in dosimetry. To compute accurate dose in small voxels or thin structures, secondary particle production must be allowed down to short ranges, often on the order of or smaller than the voxel size. This usually implies shorter range cuts than in imaging. For example, if your dose grid has a voxel side of 2 mm, you typically want electron and photon range cuts that are significantly smaller than 2 mm, for example 0.1 mm or 0.2 mm, to capture energy deposition correctly. Using too large cuts can artificially smear out dose gradients and underestimate peak doses.
The trade off is computation time. Lower energy thresholds and smaller cuts create more secondaries and require more tracking steps, which slows simulations. In practice, you need to balance accuracy and performance by testing how sensitive your dose results are to the chosen cuts. You can run small simulations with progressively smaller cuts and see when the dose distributions stabilize within your required uncertainty.
When configuring physics for dosimetry in GATE, think first about the clinical or research quantity you will compare against. For example, if you compare depth dose curves in water with experimental measurements, you must ensure that electron transport and energy loss are reproduced correctly over the relevant depth range. If you compare organ doses with a treatment planning system, your physics choices must support accurate energy deposition in voxelized patient geometries. In both cases, using a consistent, medical adapted EM physics list and carefully chosen production cuts is crucial.
For accurate dose calculations, always use detailed electromagnetic physics for electrons and photons and choose production cuts significantly smaller than the smallest relevant dose scoring dimension.
Proton therapy
Proton therapy simulations place different requirements on physics models compared to photon based imaging or dosimetry. In proton therapy, protons deposit energy through continuous slowing down and produce a characteristic Bragg peak near the end of their range. Accurately predicting the position and shape of this Bragg peak, and the dose tail beyond it, depends on both electromagnetic energy loss and hadronic interactions of protons with nuclei.
The electromagnetic part describes the gradual energy loss of protons as they traverse matter, governed by ionization and excitation processes. The hadronic part describes nuclear reactions, such as inelastic scattering, production of secondary protons, neutrons, and heavier fragments. These hadronic secondaries contribute to dose in the entrance region and beyond the Bragg peak and are important for calculating out of field doses and secondary neutron fields.
In GATE, proton therapy physics configurations are built on Geant4 hadronic physics lists that are validated for medical energies, typically from about 50 MeV up to several hundred MeV. These lists combine electromagnetic models for charged particles with specialized hadronic models for protons and light ions. The selection of hadronic models affects nuclear interaction cross sections, secondary spectra, and therefore the low dose halo around the high dose region.
When choosing proton physics models, the main criteria are the intended application and the required level of accuracy. If the goal is to study depth dose curves and Bragg peak positions in simple phantoms, you must ensure that the mean range and energy loss are correct to within a millimeter or better. This requires an up to date proton stopping power parameterization, which is already included in modern Geant4 based physics lists that are recommended for therapy. If the goal includes estimating neutron dose to healthy organs or shielding requirements, then accurate modeling of high energy hadronic secondaries is more important, and a physics list with extended high energy hadronic models becomes necessary.
As in other applications, transport parameters and production cuts complement the choice of physics models. For proton therapy, you must track protons down to energies low enough that the full Bragg peak is resolved. Stopping the transport too early will truncate the peak and distort depth dose curves. Similarly, secondary electrons produced by protons must be modeled with enough detail to capture the lateral spread of dose, although electron transport thresholds can sometimes be slightly relaxed away from sharp gradients to save time.
Simulation speed is a particular concern in proton therapy, because each event can involve complex hadronic cascades and many secondaries. If you only need central axis depth dose in a homogeneous water phantom, you might disable or limit certain physics processes that have minimal impact on that one dimensional quantity but large impact on computation time. However, for realistic patient specific planning or for neutron dosimetry, you should keep the full relevant set of hadronic processes.
Validation plays a central role in deciding whether your chosen physics configuration is appropriate. For proton beams, comparing simulated and measured depth dose curves for various energies and field sizes is standard practice. Any systematic shift in Bragg peak range or differences in distal fall off shape can indicate that the physics configuration or material definitions need adjustment. Once you find a physics configuration that reproduces measurements to within your required tolerance, you should keep it consistent across related studies to ensure comparability.
For proton therapy simulations, always use a proton capable hadronic physics list combined with detailed electromagnetic models for protons, and verify that the Bragg peak position and shape agree with experimental depth dose measurements.
Views: 11