34.6. Proton Therapy Simulation

Clinical Motivation

Proton therapy is used in radiation oncology to deliver dose to a tumor while sparing healthy tissue. Compared with photon therapy, the key feature is the Bragg peak: protons deposit little dose on the way in, then a sharp maximum near the end of their range, followed by a rapid falloff.

In Geant4, a proton therapy simulation aims to reproduce this depth dose behavior and the lateral spread of the beam in a patient or phantom. You typically simulate a beam delivery system and a water or patient-equivalent phantom, then calculate and analyze dose distributions for different treatment scenarios.

In this chapter the focus is on what is specific to proton therapy. Details such as basic water phantom geometry, generic proton beams, and generic energy deposition scoring are covered in the dedicated example chapter “Practical Example: Proton Beam in Water.” Here, we extend those ideas toward realistic medical applications.

Typical Proton Therapy Geometry

A clinical proton therapy setup consists of two main parts. First, a beam delivery system transports and shapes the beam coming from an accelerator. Second, a patient or phantom volume receives the beam. The beam delivery can be based on passive scattering or pencil beam scanning.

In Geant4 you rarely model the entire accelerator. Instead, you start your geometry at a beamline exit window or at the nozzle exit and define the beam properties there.

In a passive scattering system, scatterers, range modulators and collimators spread a monoenergetic beam to form a broad field and create a spread out Bragg peak. Modeling all of these components in detail is possible, but computationally expensive. Often, you represent only the parts that significantly affect energy spectrum and spatial distribution at the phantom entrance, or you approximate them by a pre-shaped phase space.

In a scanning system, magnetic elements deflect a narrow pencil beam to paint the target volume spot by spot. In Geant4, instead of modeling magnets explicitly, you usually define a sequence of primary proton spots with appropriate positions, directions, and energies, as provided by a treatment plan or as a simplified pattern.

For dosimetric studies, you place a water phantom or a realistic anthropomorphic phantom downstream of the nozzle. The world volume extends far enough to fully contain the treatment field and secondary particles. The geometry detail inside the patient may range from a single uniform box to voxelized CT geometry with tissue-specific materials. The choice depends on the purpose of the simulation and the computing resources.

Modeling Clinical Beam Parameters

A clinically relevant proton beam is characterized by its nominal energy, energy spread, spatial profile, angular divergence, and temporal structure. In Geant4 you encode these properties in the primary particle source.

The nominal energy is usually between 60 MeV and around 250 MeV in therapy. Clinical beamlines introduce an intrinsic energy spread that can be approximated by a Gaussian distribution. You can represent this in Geant4 either by a custom primary generator or by using the General Particle Source with Gaussian energy settings. This spread is essential to reproduce realistic depth dose curves and spread out Bragg peaks.

The transverse beam profile at the phantom entrance is typically Gaussian or approximately Gaussian. You specify the initial beam spot size in position space, and the angular divergence as small angular deviations from the nominal direction. Correlations between position and direction may exist, especially in scanning systems. When available, phase space data measured at the nozzle exit can be used to sample initial particle parameters.

Temporal structure is less critical for pure dose calculations, but it matters for time-resolved simulations, dose rate studies, and models that include detector timing. You can simulate time structure by assigning global times to primaries according to a bunch pattern if needed.

To connect with clinical planning, you may want to read a list of spots or fields from a file, each with energy, position, number of protons, and possibly weighting factors derived from a treatment plan system. Your primary generator then iterates through this list over many events, so the macroscopic irradiation reproduces the planned dose delivery.

Treatment Head and Patient Geometry

Including treatment head components is important when you study out-of-field dose, neutron production, or shielding. Passive scattering nozzles contain materials such as brass, stainless steel, plastic, and air gaps. Scanning nozzles may include vacuum windows, monitor chambers, range shifters, and aperture devices. Each contributes to scattering and energy loss.

In Geant4, you construct these components as regular volumes with appropriate materials, using basic solids or more complex geometry as required. Thin foils and chambers may be modeled as simple plates if you only need their integrated effect on energy and scattering. For detailed detector simulations, you represent their internal layers separately and assign sensitive detectors where needed.

The patient is represented by a simplified water box, a homogeneous tissue volume, or a voxelized phantom. For patient-specific studies, voxelized geometry derived from CT data is common. A separate workflow converts DICOM images into a 3D array of voxels with associated Hounsfield units. These are mapped to Geant4 materials by a predefined calibration curve. The resulting G4VPhysicalVolumes represent the patient anatomy, and Geant4 propagates protons through heterogeneous tissues.

To represent treatment positions, you may rotate and translate the phantom relative to the beam to mimic gantry angles and couch shifts. This allows simulation of multiple fields and complex treatment arrangements while using a single base patient model.

Physics Lists for Proton Therapy

Accurate proton transport in the therapy energy range requires appropriate electromagnetic and hadronic physics models. Geant4 provides reference physics lists that are widely used in proton therapy research and validation.

Electromagnetic processes are primarily responsible for continuous energy loss and multiple scattering. For accurate depth dose and lateral spread, you need a modern electromagnetic option, such as the standard electromagnetic models with an option tailored for medical applications. These models reproduce stopping powers and scattering in soft tissue and high-Z materials with sufficient precision for clinical studies.

Hadronic processes become important for nuclear interactions of protons with tissue and beamline materials. These interactions produce secondary particles, including neutrons and heavy fragments, that contribute to dose outside the primary beam. Physics lists such as QGSP_BIC or specialized lists for medical applications include models tuned for the therapy energy range and are commonly selected.

Production cuts deserve special attention in proton therapy simulations. Cuts that are too large can distort dose near interfaces and around the Bragg peak. On the other hand, extremely small cuts increase computation time significantly. In practice, you select cuts that are small compared to the smallest geometry features that influence dose distribution, then verify their adequacy by convergence studies.

Dose Calculation in Proton Therapy

The central quantity in proton therapy is absorbed dose in gray, defined as deposited energy per unit mass. In Geant4, you obtain energy deposition from steps and convert it to dose by dividing by the mass of the scoring volume.

In simple studies, you may score depth dose in a 1D array of slabs along the beam axis. Each slab collects the total deposited energy in an event or run. More realistic simulations use 3D dose matrices, where a phantom is divided into voxels and each voxel accumulates deposited energy. You then convert the 3D energy map into a 3D dose map by applying each voxel’s mass.

To compare with clinical plans and measurements, you often normalize dose distributions. For example, you can normalize to the maximum dose, to the dose at a reference depth, or to the total delivered monitor units. Relative depth dose curves and lateral profiles can be extracted from the 3D dose matrix by summing or averaging over appropriate directions.

For treatment planning validation, you may calculate dose difference maps and distance to agreement metrics between Geant4 results and reference distributions. Although detailed analysis methods are handled elsewhere, from the simulation perspective it is crucial to keep event statistics, bin sizes, and scoring definitions consistent with the intended comparison.

Spread Out Bragg Peak and Range Modulation

Clinical proton treatments rarely use a single pristine Bragg peak. Instead, they superimpose multiple peaks at different ranges to create a spread out Bragg peak that covers the target volume uniformly in depth. In Geant4, you can simulate this either with geometric range modulation devices or with multiple beam energies.

A passive scattering system might include a rotating range modulator wheel made of stepped thicknesses of material. As the wheel turns, the beam passes through different thicknesses, each producing a Bragg peak at a slightly different depth. To model this explicitly, you can represent the wheel geometry and simulate time or angle dependent irradiation. Alternatively, you can approximate the modulation by delivering a sequence of beams with different energies, weighted by the nominal wheel design.

For scanning systems, range modulation is achieved by varying the initial proton energy between spots. Here, the spread out Bragg peak is built by summing dose from multiple energy layers. In Geant4, you realize this by assigning different energies to groups of simulated events and accumulating dose into a common scoring structure. The final SOBP emerges in post-processing as the sum of contributions from all energies.

The quality of the SOBP depends on the accuracy of the depth dose for each individual energy and on the correct weighting. Therefore, it is vital to validate each monoenergetic component against measurements in water, and then verify that the composed SOBP reproduces the clinical field’s measured characteristics.

Range, RBE, and Clinical Considerations

The proton range in tissue is critical for treatment planning. Small errors in stopping power or material assignment can shift the distal edge of the Bragg peak by millimeters. In Geant4 simulations for proton therapy, you typically pay close attention to material definitions, especially densities and elemental compositions, and to the calibration from CT Hounsfield units to tissue materials.

Range is often evaluated by the depth at which the dose falls to a small fraction of the maximum, such as 80 percent or 90 percent. To verify your simulation, you compare the simulated range in a water phantom to published reference data or to institution specific measurements at each nominal energy. Any systematic discrepancy suggests problems with physics choices, material definitions, or beam parameters.

Relative biological effectiveness, or RBE, accounts for biological differences between proton and photon irradiation. Geant4 itself calculates only physical dose and does not include a full biological response model. For many clinical proton centers, a constant RBE factor is applied in treatment planning. In a simulation workflow, you typically compute physical dose in Geant4 and then apply an RBE model externally in the analysis step, possibly as a function of depth or linear energy transfer.

Other clinical considerations include robustness to setup and range uncertainties, and the impact of tissue heterogeneities and motion. While these topics extend beyond basic simulation, they influence how you design your Geant4 studies. For example, you may perform repeated simulations with slightly varied beam energies or shifted geometries to study sensitivity to uncertainties.

Validation for Proton Therapy Simulations

Proton therapy simulations must be carefully validated before you rely on them for research or clinical decision support. Validation usually proceeds in stages, starting with simple configurations and progressing to complex patient cases.

The first stage is basic physics validation in water. You simulate monoenergetic proton beams with realistic beam parameters and compare depth dose curves, ranges, and lateral profiles with reference data and measurements. Agreement within predefined tolerances gives confidence in the selected physics list, materials, and beam models.

Next, you validate the modeling of the treatment head. You simulate standard fields with the same nozzle settings used in the clinic and compare calculated dose distributions in water phantoms with measurements, for example from ionization chambers or 2D detector arrays. Any discrepancies may point to missing geometry components or incorrect material assignments.

For patient specific validation, you may simulate a set of test cases planned in a treatment planning system. You import the CT based geometry and beam parameters into Geant4, compute dose, then compare with the planning system’s predicted dose. Tools such as gamma index analysis measure agreement in both dose and spatial dimensions.

Throughout validation, you must also estimate statistical uncertainties in your Monte Carlo results and ensure that sufficient numbers of events are simulated. Regions of low dose or small volumes often require more histories to reach acceptable precision. By combining careful physics configuration, realistic beam modeling, and systematic comparison with measurements and analytical expectations, you build confidence that your proton therapy simulations faithfully represent clinical treatments.

Views: 9