The use of apertures combined with pencil beam scanning (PBS) fields has gained attention as an optimization strategy in proton therapy. The apertures serve to sharpen the lateral dose fall-off of a proton field yielding a higher conformality of the target volume irradiation and sparing healthy tissue. Hence, the clinical advantage of proton therapy over conventional radiotherapy can even be improved in certain scenarios, e.g. the irradiation of shallow tumors (Dowdell et al 2012, Moteabbed et al 2016, Yasui et al 2017, Bäumer et al 2018, Ciocca et al 2019, Maes et al 2019, Bäumer et al 2021, Holmes et al 2022). The simplest clinical implementation consists of field-specific, milled brass apertures, which are mounted manually in an accessory slot of the treatment head. Alternatively, the field edges are collimated by multi-leaf collimators (MLCs) (Tominaga et al 2022), which can also be adapted to the individual energy layers (Vilches-Freixas et al 2020, Grewal et al 2021). Eventually, the dynamic trimming of individual spots with rapidly sliding bars was successfully demonstrated (Hyer et al 2014). The need for supplementary collimation of scanned fields scales with the spot size of the pencil beam (Hyer et al 2021).

While supplementary apertures enhance the conformality of scanned proton fields, they come with the disadvantage of excess stray dose mediated by secondary neutrons (Smith et al 2019). In general, secondary neutrons are produced by non-elastic nuclear interactions of protons within the beam line, machine and room components as well as the patient (Smith et al 2019, Hälg and Schneider 2020). The increased stray dose by the use of apertures is even larger for spatial fractionation realized with brass apertures, which is currently under research (Charyyev and Wang 2020, Tobola-Galus et al 2025). Since neutron absorption has a low probability and additional secondaries are produced within the patient and the room components, the neutron stray dose is incident on the whole body far from the edge of the proton field, i.e. out-of-field (Newhauser and Zhang 2015, Hälg and Schneider 2020). Considering the possibly high biological effectiveness of neutrons and the risk of radiation-induced second primary cancers (SPC), the out-of-field neutron dose needs to be assessed precisely (Hälg and Schneider 2020). This requires either a sophisticated dosimetry approach or complex Monte Carlo (MC) simulations (Farah et al 2015, Trinkl et al 2017, Englbrecht et al 2021, Darafsheh et al 2025). There, the corresponding dose calculations are not yet included in commercial clinical treatment planning systems and are not accounted for in the therapeutic dose (Hälg and Schneider 2020, Darafsheh et al 2025). Previous studies indicate that the consequences in terms of SPC are limited (Hälg and Schneider 2020). However, there is a major concern particularly for young patients due to the larger impact on the lifetime attributed risk for SPC. Furthermore, for pregnant patients, the dose to the fetus mediated by stray neutrons is of major interest in cancer management (Wang et al 2016, Weessies et al 2025).

The purpose of the current study was twofold. First, the increase of the out-of-field dose in terms of neutron ambient dose equivalent by supplementary apertures sharpening the field edges was determined experimentally. A measured data set for monoenergetic square proton fields without apertures applied to a water-equivalent target acquired with a similar setup as in Trinkl et al 2017 served as a reference. In addition, the dependence of the neutron production on the aperture material was evaluated for brass and nickel as well as the influence of the order of a range-shifter and an aperture on the neutron dose. Second, the measured data were used for comparison to simulation results obtained with the TOPAS (Perl et al 2012, Faddegon et al 2020) MC simulation framework for collimated scanned fields. This consistency check was valuable since the uncertainties associated with neutron dosimetry inside proton therapy treatment rooms as well as the simulation of the respective experimental conditions are typically high and difficult to determine (Farah et al 2015, Trinkl et al 2017, Englbrecht et al 2021, Darafsheh et al 2025). To narrow down the uncertainties and allow for a more accurate estimation, a direct comparison between simulation and measurement results is crucial.

2.1. Experimental setup

Experiments were performed with an IBA ProteusPlus treatment system (IBA PT, Louvain-La-Neuve/Belgium) at the West German Proton Therapy Centre Essen, Germany. Protons were accelerated by an isochronous cyclotron up to about 230 MeV. Kinetic energies down to 100 MeV were obtained with a degrader and the subsequent energy selection system. Proton fields were delivered in step-and-shoot PBS mode with the IBA universal nozzle located in the fixed-beam treatment room. The treatment head (nozzle) comprises a segmented ionization chamber at its entrance for beam centering, a pair of dipole scanning-magnets, and two ionization chambers at its exit for monitoring the fluence and the lateral beam position. The average virtual source-axis-distance (VSAD) is 210 cm. The energy-dependent standard deviation () of the Gaussian-shaped pencil beam spot lies between 3.2 mm (226.7 MeV) and 8.1 mm (100 MeV) at the isocenter (IC). The angular spread of the pencil beam ranges from 2.4 to 5.9 mrad (). The standard deviation of its energy-spread, which is described by the of a normal distribution (Clasie et al 2012), lies between 0.05 and 0.5%.

Owing to its universal design for passive beam delivery and PBS, beam-shaping devices such as range-shifters and collimating apertures can easily be mounted in the snout, i.e. the applicator tray at the nozzle exit. The cylindrical Snout180 of the nozzle (outer diameter: 180 mm) used in this study allows for a maximum inner diameter of 160 mm for clinical apertures. To minimize the air gap between nozzle exit and patient surface, the snout can be translated in beam direction. In case of the application of a range-shifter, the PMMA block is mounted downstream of the apertures at the nozzle exit (figure 1(a)).

Figure 1. Close-up of the collimating aperture in combination with a downstream range-shifter (halfway inserted) prior to the measurements to investigate the impact on the secondary neutron dose distribution inside the treatment room (a). The schematic proton spot distribution relative to the aperture (brass: aperture, white: opening) is visualized on the right (b).

Download figure:

Standard image High-resolution image

For all experiments, primary beam energies of 100, 140 and 180 MeV were selected to deliver quasi-monochromatic proton fields of lateral dimension to a solid water phantom (SP34, IBA dosimetry, Schwarzenbruck/Germany) comprising 30 polystyrene plates (RW3) of 1 cm thickness aligned in beam direction. The geometric center of the phantom was placed at the IC (figure 2). The lateral spacing of the equally weighted spots was 2.5 mm. One monitor unit per spot was applied in each field, which corresponds to an entrance plateau dose to water between 2 and 5 Gy at 2 cm depth, depending on the primary beam energy and range-shifter setup. The measured neutron dose values were normalized to the entrance plateau dose to water at 2 cm depth.

Figure 2. Setup of the measurements investigating the impact of a collimating aperture on the secondary neutron dose distribution at different azimuthal angles and 2 m distance from the isocenter (IC) (0, 45, 90 and ) around a solid water phantom (center) inside the fixed-beam treatment room. The red spheres visualize the detection volume at the corresponding positions of the WENDI-II extended-range rem-meter in the experiments. Position 1 was changed to position 1a (1.12 m distance) in the first experiment. Position 4 was only included in the simulations. The distances to the aperture of positions 1(a) to 4 were 2.25 (1.37), 2.18, 2.02 and 1.83 m, respectively. The aperture emission angles with the initial beam direction at the detection positions were , , and , respectively.

Download figure:

Standard image High-resolution image

In the first experiment, fields of lateral size were applied with and without a rectangular aperture. The PBS field size is defined by the positions of its outermost spots. Employing the aperture with an opening of (figure 1(a)), taking the SAD and its difference for both lateral dimensions into account, the effective field size at the IC of the ‘open beam’ configuration (, spots) can be sustained while the lateral dose fall-off is reduced. To achieve this, the spot pattern and the aperture position, i.e. the air gap, were adapted such that the outermost spots of the spot field would hit the aperture edge with the center of their Gaussian lateral shape (figure 1(b)). To match the opening dimensions of the aperture to the SAD, its downstream edge was set to a distance of 25 cm to the IC, creating a 10 cm air gap. Furthermore, two different polymethylmethacrylate (PMMA) range-shifters were applied with the described fields for proton energies of 140 and 180 MeV in combination with the aperture and open beam configurations. They were mounted at the nozzle exit downstream of the apertures. The RS51 has a water-equivalent thickness (WET) of 5.10 cm, while the WET of the RS74 is 7.43 cm. Their physical thicknesses correspond to 4.44 cm and 6.50 cm, respectively.

For the second experiment investigating the impact of different aperture materials on neutron production, two clinically relevant materials were selected. These were brass (CuZn(39)Pb(3), mass density ) and nickel (mass density ). In pursuit of a simple material benchmark in terms of relative neutron stray dose, and due to limitations in nickel milling, fully closed apertures with a thickness of 3.3 cm were mounted. The same monochromatic fields as with the open beam configuration were delivered, applying primary beam energies of 100, 140 and 180 MeV. Since the closed apertures with a high mass density were placed at the nozzle exit, almost no primary dose was delivered to the phantom in comparison to the open beam scenario due to the high absorption of the beam. In order to enable a relative comparison between the neutron production of both materials, the entrance plateau doses of the corresponding open beam scenarios were used for the normalization of the neutron dose values measured with the closed aperture setup.

For the third experiment, the configuration featuring a fully closed brass aperture was extended with an RS74 that can be mounted upstream as well as downstream of the respective aperture (Bäumer et al 2018). Here, equivalent fields were irradiated for each order of range-shifter and aperture, applying primary beam energies of 140 and 180 MeV, to examine the effects of the order on the neutron dose distribution in the room.

2.2. Secondary neutron dose measurements

For measuring the out-of-field neutron dose distribution inside the treatment room, three FHT 762 WENDI-II extended-range neutron detectors (Thermo Scientific, Franklin/USA) read out by FH 40-G measuring units (Thermo Scientific) were positioned in the horizontal plane of the IC, at a distance of 2 m to the IC at azimuthal angles of 0, 45 and (position 1–3) with respect to the initial proton beam direction (figure 2). In the first experiment, position 1 was changed to a distance of 1.12 m (position 1a) to the IC to decrease the distance to the collimating aperture and, thus, enhance the sensitivity to aperture effects. It should be noted that the positions were equidistant from the phantom, but not from the aperture, which was located 25 cm upstream of the IC. The distances to the aperture of position 1(a), 2 and 3 were 2.25 (1.37), 2.18 and 2.02 m, respectively. The aperture-related angles with the beam axis at the positions were close to the ones relative to the IC, i.e. , and for position 1(a) to 3, respectively.

The WENDI-II extended-range rem-meter was designed to measure the operational quantity neutron ambient dose equivalent () on the basis of the fluence-to- conversion coefficients from the ICRU Report 57 (ICRU 1998, Olsher et al 2000). While the is measured at a specific detection position, it delivers a conservative estimate of the effective dose received by a whole body neutron irradiation (Hälg and Schneider 2020). Therefore, is not used to assess the risk for late effects of proton therapy patients. In this study, it rather served to compare out-of-field doses for options of nozzle accessories on a relative basis. The WENDI-II contains a He-proportional counter within a cylindrical boron-doped polyethylene moderator (22.9 cm in diameter and 21 cm in height) that features a 1.5 cm thick tungsten powder shell. The high-Z material improves the response of the WENDI-II to high-energy neutrons up to 5 GeV (Olsher et al 2000). The detector was calibrated with neutrons from a Cf-source. Although its energy response function is not flat, it matches the neutron fluence-to- conversion coefficients from the ICRU Report 57 reasonably well (ICRU 1998, Olsher et al 2000, Farah et al 2015). In the neutron energy domain between around 10–200 MeV, the response function of the WENDI-II exhibits both an overestimation of up to 40% (around 140 MeV) as well as an underestimation by up to a factor of two (around 20 MeV). In contrast to this, in the energy range between 0.1 and 10 MeV, the response function is relatively flat with values within 20% of the calibration (Olsher et al 2000). Hence, considering the clinically relevant energy range of neutrons and the generally high measurement uncertainties associated with neutron dosimetry, the WENDI-II is an appropriate choice for the intended application with an associated uncertainty of 15%–20% on the measured values in experimental scenarios comparable to the ones in this study (Caresana et al 2014, Farah et al 2015, Darafsheh et al 2025). For all experiments in this study, the target was irradiated with an entrance plateau absorbed dose of at least 2 Gy to yield a statistical uncertainty of the measured below 1% at all positions. The count statistics could be estimated with a typical conversion factor for the WENDI-II (0.33 nSvcount) from Caresana et al 2014 and Poisson statistics (). To assess the reproducibility of the measured results, each value was acquired twice. In addition, a cross-calibration between all three WENDI-II detectors was performed prior to the main experiments. Furthermore, in the first run of the experiments, a dose rate dependency check was conducted to evaluate a possible underresponse of the detectors due to dose rate effects. Therefore, the beam current was decreased to 1/8th of the standard current.

2.3. Monte Carlo simulations

MC simulations were performed with the Geant4 wrapper TOPAS (Perl et al 2012, Faddegon et al 2020) (version 3.6) tailored to PT (Zacharatou Jarlskog and Paganetti 2008). The simulation geometry considered the most important room components with their respective materials regarding secondary neutron production. The beam model was implemented as a Fermi–Eyges source (Eyges 1948) in the TOPAS source option, which allows for the definition of all relevant beam parameters. An in-house beam model developed based on measurements was used to obtain all necessary data as described in Verbeek et al 2021. The spot scanning pattern to obtain the field at the IC was implemented via the lateral translation of the source, employing the TOPAS time feature. The beam angle and position with respect to the central beam axis were calculated taking the VSAD in both lateral dimensions into account.

For all simulations in this study, the TOPAS default physics list and the standard physics options were used in agreement with De Saint-Hubert et al 2023. This included setting a default range cut, which corresponds to the minimum energy for the production of secondary particles, to 0.05 mm for all primary particles. The default minimum and maximum energies for the electromagnetic range were 100 eV and 500 MeV, respectively, for all particles.

The TOPAS default physics list features six physics modules comprising the corresponding Geant4 physics lists G4EMStandardPhysics_option4, HadronPhysicsQGSP_BIC_HP, G4HadronElasticPhysicsHP, G4IonBinaryCascadePhysics, G4DecayPhysics as well as G4StoppingPhysics (Perl et al 2012). Processes such as the intranuclear cascade, nuclear evaporation and various neutron interactions are described by the high-energy models featured in HadronPhysicsQGSP_BIC_HP for inelastic interactions of hadrons, including the binary cascade (BIC) model and the data-driven high precision neutron package (NeutronHP) extension, describing neutron interactions from thermal up to 20 MeV (Zacharatou Jarlskog and Paganetti 2008).

Two types of scorers were used to tally the relevant quantities for the evaluation of the out-of-field neutron dose distribution. An absorbed dose scorer with 300 bins in beam direction with a resolution of 1 mm was placed inside the solid water phantom to tally the depth-dose distribution in each simulation. The purpose was to acquire the corresponding absorbed dose values used for the normalization of the neutron contribution to . The was tallied with four different neutron ambient dose equivalent scorers inside air-filled spheres with a radius of 10 cm at distances of either 1.12 m or 2 m from the center of the irradiated phantom, i.e. the IC (figure 2). The scorer calculates the by tallying the incident neutron fluence binned by the respective neutron energy and folding the fluence values with energy-dependent fluence-to-ambient dose equivalent conversion coefficients. For this study, the conversion coefficients were taken from Pelliccioni 2000. The simulations included an additional scoring position at with the primary beam and 2 m distance to the IC (position 4), i.e. and 1.83 m distance to the aperture, for an enhanced evaluation of possible angular dependencies (figure 2).

Concerning the investigation of neutron generation in the aperture, its direct contribution to the out-of-field neutron dose was discriminated from the ones from secondary interactions in the phantom and the surrounding room components. Hence, in the corresponding simulations, the scorers were extended by component filters for the neutron origin in the aperture and the phantom.

Besides ambient dose equivalent scoring, an energy-resolved neutron fluence spectrum was scored inside the air-filled spheres at each position. The scorer was defined to tally the neutron spectrum in 120 logarithmic-equidistant bins from 1 meV to 1 GeV. The simulated spectra deliver additional information on the energy-dependent fluence contributions to the calculated neutron ambient dose equivalent at each position.

In addition to the scorers with their geometry and material, a number of other relevant components with their corresponding materials were defined within the simulation environment. All beam-shaping devices downstream of the modeled proton source, i.e. range-shifters and collimating apertures, were included into the simulation geometry. In addition, components such as the solid water phantom, the Kevlar treatment table below as well as its mount made of steel and aluminum, iron magnets of the beam line (simplified cuboids with the magnets’ outer dimensions) at the respective positions and the concrete walls, floor, ceiling and maze were modeled. The material of the world was set to air. Further information on the material definitions can be found in the supplementary material.

2.4. Dose energy response correction of the WENDI-II measurements

The simulated fluence energy spectra tallied at each detection position were used to correct for the imperfect energy response of the WENDI-II rem-meter as alternative to the standard readout of the device. This served to assess the measurement uncertainties caused by the spectral response of the WENDI-II. The readings of the WENDI-II (in terms of ) were multiplied by the correction factors described in the following. The correction method is based on the neutron fluence energy response data simulated for the WENDI-II with MCNPX 2.7 from De Smet 2017. The energy response values in terms of fluence were converted to response values in terms of by bin-wise division by the respective fluence-to- conversion coefficients to obtain the simulated energy-dependent response function of the WENDI-II. In addition, the average energy response of the WENDI-II for Cf was calculated with the Cf energy spectrum from the ISO 8529 standard. The simulated energy response function of the WENDI-II was normalized with the calculated average energy response for Cf to obtain the energy-dependent response function denoted as . The average response for Cf of this approximated normalized response function is equal to unity, corresponding to the calibration of the detector with Cf in terms of . For each simulated value in this study, the corresponding simulated fluence energy spectrum was integrated together with the aforementioned fluence-to- conversion coefficients and the normalized energy response function interpolated to the energy bin structure of the spectra from the simulations. The resulting integrals over the full energy spectrum were divided by the corresponding simulated total values yielding the inverse of the desired correction factors.

3.1. Impact of the aperture on the out-of-field neutron dose

Figures 3 and 4 show the results for the comparison of the aperture with the open beam configuration in the scenarios with and without range-shifter. The standard deviation () of all simulated results calculated by TOPAS was below 1%, thereby matching the experimental statistical uncertainty (see section 2.2). Moreover, the cross-calibration of all three WENDI-II detectors as well as the dose rate dependency check conducted during the measurements yielded insignificant deviations within the reproducibility, which was below 5%. In all following figures, the measurement data are presented as unprocessed WENDI-II readings (colorbars) as well as values obtained from the WENDI-II response correction (data points) depicted in section 2.4. All measurement data presented in the text are referring to corrected data.

Figure 3. Comparison of the results of simulations and measurements to investigate the impact of a rectangular static aperture on the neutron ambient dose equivalent normalized to the respective plateau absorbed dose at 2 cm depth in RW3, for various beam energies and positions around the RW3 phantom at 1.12 m/2 m distance from the isocenter. The statistical uncertainty of all simulated and measured results was below 1%. Measurement data are presented as colorbars for the unprocessed WENDI-II readings and data points for the corrected obtained from the WENDI-II response correction. Simulation data are presented as stacked bars of contributions to the total from neutrons originating directly from the aperture (black hatched) or the remaining room components (colored), including the phantom.

Download figure:

Standard image High-resolution image

Figure 4. Comparison of the results of simulations and measurements to investigate the impact of a rectangular static aperture combined with two range-shifters with a water-equivalent thickness (WET) of 5.1 cm/7.43 cm (‘RS51/RS74’) on the neutron ambient dose equivalent normalized to the respective plateau absorbed dose at 2 cm depth in RW3, for 180 MeV and various positions around the RW3 phantom at 1.12 m/2 m distance from the isocenter. The statistical uncertainty of all simulated and measured results was below 1%. Measurement data are presented as colorbars for the unprocessed WENDI-II readings and data points for the corrected obtained from the WENDI-II response correction.

Download figure:

Standard image High-resolution image

The direct contributions from the aperture and the remaining room components, including the phantom, to the total