Abstract
In vivo range verification methods will reveal information about the penetration depth into a patient for an incident proton beam. The prompt gamma (PG) method is a promising in vivo technique that has been shown to yield this range information by measuring the escaping MeV photons given a suitable detector system. The majority of current simulations investigating PG detectors utilize common scintillating materials ideal for photons within a low neutron background radiation field using complex geometries or novel designs. In this work we simulate a minimal detector system using a material ideal for MeV photon detection in the presence of a significant neutron field based on the Cherenkov phenomenon. The response of this selected material was quantified for the escaping particles commonly found in proton therapy applications and the feasibility of using the PG technique for this detector material was studied. Our simulations found that the majority of the range information can be determined by detecting photons emitted with a timing window less than ∼50 ns after the interaction of the proton beam with the water phantom and with an energy threshold focusing on the energy range of the de-excitation of 16O photons (∼6 MeV). The Cherenkov material investigated is able to collect these photons and estimate the range with timescales on the order of tens of nanoseconds as well as greatly suppress the signal due to neutron.
Introduction
Range verification methods play an important role in proton therapy. Knowledge of where the proton beam comes to rest in vivo reveals information about which intervening structures were irradiated, how much of the target volume was covered, and if the beam distally penetrated into unplanned normal tissue. The need to monitor the range arises from the uncertainties associated with the steps involved in computing the range and is one of the main factors that limit the full potential of proton therapy [1–7]. Specifically, the uncertainties dictate the beam direction, and in certain cases such as head and neck patients where the disease site can be surrounded by critical structures including the brainstem and spinal cord (see Daftari et al. [8]), a conservative treatment plan may not be the most ideal plan (see Knopf and Lomax [5]). However, all patients, including challenging treatment sites, can benefit from knowing the delivered dose distribution for target conformality and normal tissue sparing.
The treatment plan includes multiple steps where every step in the process increases the uncertainty, which arise from both theoretical estimations and clinical implementations [5, 9]. The greatest contributing factors for the overall uncertainty are the daily patient setups and changes in the patient’s anatomy; however, all steps involved with the calculation of the range contribute to a cumulative uncertainty of ∼1.5 cm [10]. Knopf and Lomax present a review article discussing many of the sources of uncertainty in proton therapy treatments [5] while the uncertainties associated with Monte Carlo techniques have been discussed by numerous authors [7, 11]. Knowledge of these uncertainties can give an overall estimation as to how the delivered dose will be distributed within the patient, but a true indication of the accuracy of the delivered dose distribution as compared to the planned dose distribution would be through measurements conducted through in vivo processes. To quantify these effects on the proton beam’s range, many in vivo range verification methods have been developed.
There are many techniques used to verify or monitor the range of the incident proton beam [5]. Among these techniques are methods based on the escaping secondary particles produced from the interaction of the incident proton with the target material. These include the positron-emitting nuclei (PEN) [2, 12–14] and prompt gamma (PG) methods [1, 16]. The PEN method is used in conjunction with a commercially available or custom-built PET scanner that monitors the de-excitation gamma particles arising mainly from the daughter fragments of 15O and 11C with timescales ∼2 –20 minutes. The production yields as a function of range for these daughter fragments (including shorter lived PEN particles), their associated positrons, and resulting annihilation photons, are calculated via Monte Carlo methods for a particular patient anatomy and dose. These calculations are then compared with the measurements produced from the PET scanner where the differences between the planned and delivered dose distributions can be quantified. However, this method only yields a result post-irradiation and is subject to biological effects (“wash out”), long PET acquisition times, and is tissue composition dependent. The PG method is a promising alternative to the PEN method as it has the potential to overcome many of the PEN method’s drawbacks [1, 6], and simulation comparisons between both methods have been conducted [17]. Although the PG method has many advantages compared to the PEN method, the PG method has not been used clinically due to the lack of a commercially available detector system.
Prompt gamma detector systems are being actively studied and many have been modeled with Monte Carlo simulations with some groups constructing experimental detectors for measurements [1, 18–21]. Among these are detector designs based on a Compton camera, a novel knife-edged collimation system, an array-type, and a sequential type. Most utilize a scintillator as the detector medium including CsI(Tl), BGO, LYSO, LaBr3, and NaI, while others use semi-conducting materials such as high-purity germanium. Regardless of the material, these systems all incorporate some type of method, either passive (shielding or collimation design) or active (through time-of-flight data analysis), to reduce the background signal generated from unwanted secondary particles while localizing the high energy photon signal that contains the useful range information. The background suppression is an important aspect to these detector systems and is one of the motivations for this investigation.
This material investigation focuses on a simplistic, proof-of-concept, detector system design for use with the PG method based on a material that exploits the Cherenkov phenomenon. In a given medium, relativistic particles traveling faster than light polarize the molecules of the medium along its track. Certain media emit radiation as their molecules return to the ground state in the form of Cherenkov photons. The velocity of the particle must exceed the speed of light in the medium given by the relation [22]:
The glass Cherenkov material (In2O3) included in this investigation has recently been studied to understand the response to low and high energy photons as well as thermal neutrons [23–25]. In these works the authors show that by changing the chemical composition of the materials the response to neutrons can be minimized while still preserving specificity to MeV photon detection. The purpose of the present investigation is to evaluate the feasibility of the Cherenkov material to applications in proton therapy, specifically to range verification using the PG method.
The investigation serves three objectives: 1) To describe the complex radiation field produced from incident protons of varying energies onto a water phantom, 2) To characterize the Cherenkov detector material to relevant MeV photons and neutrons arising from the complex radiation field, and 3) To study the feasibility of the PG method utilizing the Cherenkov detector material with a simple detector setup.
This investigation uses the Geant4 Monte Carlo Toolkit (version 9.4 patch 02) [26, 27]. Geant4 gives flexibility to the user for defining the entire simulation settings from physical models, target geometry, and beam parameters. The physics model settings are based on previous publications [1, 29] where hadron inelastic interactions are described by the binary cascade model implemented in the built-in “QGSP_BIC_HP” physics package. The inelastic interactions are indeed important as they dictate secondary particle production. This package also includes a complete description for the interactions of particles typically encountered in the energy range used for proton therapy. Electromagnetic processes are described by the standard electromagnetic physics package. Included in these simulations are the description of the optical photon that is handled by the G4OpticalPhysics class that models scintillation and Cherenkov light production, transportation, and absorption. The physical parameters for the optical settings are obtained from a publication of Hayward et al. [23]. A cut value of 50μm is used for all simulations that translate to energy values of 990 eV, 57.33 keV, 56.57 keV, and 5 keV for photons, electrons, positrons, and protons in water as well as 105.67 keV for electrons in the Cherenkov glass sample discussed below. The cut value for electrons in glass falls below the threshold of Cherenkov light production, approximately 156 keV. The optical absorption data was obtained through personal communication with Hayward [22] with an absorption edge around 400 nm. Though validating the Geant4 physics processes used in the simulations is not the focus of the present manuscript, extensive work has however been done in the literature [30, 31]. Various electromagnetic and hadronic models available in Geant4 and suitable for proton therapy simulations were compared against experimental data of range measurements of proton in water. A good agreement between measurements and simulated data was achieved by using the standard electromagnetic and binary cascade models, which we have incorporated in the present work [28]. The low energy electromagnetic models used in the simulations are also compared to calculated values provided by NIST [30]. Few papers also exist for comparison of Geant4 modeling of prompt gamma emissions. Verburg et al. compared multiple codes including Geant4 to experimental data of discrete gamma ray cross-sections. Good agreement between data and calculated cross-sections were achieved for some energies of incident protons but large discrepancies indeed existed among other energies. The amount of prompt gamma was compared to experimental data by Dedes et al. [32] where the authors found that the yield was overestimated by Geant4. A similar result was obtained by Schumann et al. [33].
The first objective (secondary radiation field study) is conducted using a water phantom of dimensions 20×20×30 cm3 with proton beams of various energies (70 –200 MeV, pencil like beam with spatial spread of 1 cm radius and no energy distribution, using up to ten million histories per simulation). The yields of secondary particles escaping the phantom as well as the angular distributions and energy spectra of secondary neutrons and photons are quantified. The escaping photons are further studied to quantify their times of emissions as well as their emission amounts as a function of depth. The amounts of photons per depth are compared with the incident protons energy deposition to deduce the range.
The second objective (characterization of the Cherenkov detector material) is conducted based on a glass sample described in [23, 24]. The sample is composed of 90 g of cladding material mixed with 7.2 g of In2O3, has a density of 2.83 g/cm3, an index of refraction of 1.56 (measured with 600 nm photons), a Cherenkov photon production threshold of 156 keV, and a range for 6 MeV electrons of 1.3 cm. The sample has a very fast response and transportation time; the optical photons produced from a fast electron are generated with times on the order of tens of picoseconds while the traversal of these optical photons across the length of the sample are on the order of hundreds of picoseconds, depending on the sample size [23]. The sample is irradiated with beams of secondary photons and neutrons that were generated in the above simulation in order to quantify the response of this material for a common radiation field in proton therapy treatments. The choice of using these beams comes from the results of the first objective. The energy spectrum of the secondary electrons and resulting optical photons were generated in the Cherenkov material. LYSO and BGO scintillators are also used in the simulations for comparison.
A simulation of a simple detector segment consisting of the glass sample coupled to a photo-multiplier tube (PMT) window wrapped with a thin, highly reflective material, where all components are encased in an aluminum sheath is included. The aluminum sheath is used to achieve a light-tight environment to reduce the background noise from the external visible light. The number of optical photons striking the PMT window is counted to investigate the optimal spatial dimensions and to estimate the signal produced from the glass sample. The detector segment is shown in the insert of Fig. 1 which shows the detector system setup.
The final objective is satisfied by simulating a beam of 1 million, 110 MeV protons onto a water phantom with the detector system discussed above. A single detector segment with dimensions of 1 cm in length by 1 cm diameter is placed along the beam direction directly above a 5 cm×5 cm×30 cm water phantom to estimate the range of the incident protons. The resulting Cherenkov optical photons striking the PMT window are counted and a relation between the incident beam’s range and the generated signal is deduced. For this objective, the center of the detector segment is placed at a depth of 4.5 cm to 10.5 cm with 1 cm increments to detect the emitted secondary particles. Between the depths of 8.5 and 9.5 cm we used 0.2 cm increments for better resolution in the BP location. Only secondary particles emitted in the location of the detector segment are simulated. This proof of concept procedure can be realized by implementing a sequential detector setup utilizing the system described in [1].
Results and Discussion
Range verification utilizing the PG method in proton therapy has certain goals to achieve. These would include localizing in vivo the important signal generated from the PG photons while increasing the ratio of signal produced from photons to that from neutrons (the production ratio PR). This investigation aims to characterize the Cherenkov material to the PG method by simulating a beam of PG photons and neutron radiation in order to estimate the PR. We then apply this material to a simple detector setup to assess its localization performance for estimating the range. To accomplish this we first investigate the escaping radiation produced from the target phantom.
Secondary radiation field analysis
Our investigation for the characterization of the Cherenkov material begins with the analysis of the complex secondary radiation field arising from the water phantom study. The purpose of studying the escaping particles is to provide knowledge as to which particles and what energies to consider for the radiation field for the PR estimation.
Escaping particle yields
As the primary proton comes to rest within the target it undergoes several mechanisms to deposit its energy. Among the many ways to deposit the energy include inelastic collisions with the target nuclei thus creating secondary particles (including intact target nuclei in an excited energy state and daughter nuclei fragments). Some generated secondary particles are confined to the phantom. We found that among the many particles produced from the incident proton interactions, the lighter daughter fragments, nucleons, and photons make up the majority of escaping particles. Selected escaping secondary particles are shown in Table 1, which reveals neutrons and photons having the highest emission yields for all proton energies considered in this study (70 MeV, 110 MeV, 160 MeV, 200 MeV). Of all escaping particles, the photons and neutrons are of interest in proton therapy applications; the photon component of the radiation field is used for range verification techniques while the neutron component contributes to unwanted signal, degrades the detector system, and is the main component needed for facility shielding calculations. The remaining escaping particles are not used for any current range identification methods.
For our purpose it’s sufficient to conclude that photons and neutrons comprise the majority of the escaping radiation field encountered in proton therapy applications regardless of the energy of the primary beam. As the photon and neutron components are critical to range verification techniques and detector design, the photon and neutron energy spectrum, as well as their angular distribution, is further analyzed for use in characterizing the Cherenkov detector material.
Secondary neutron and photon analysis
The neutron component for our PR estimation is composed of a background field made from neutrons of various energies where the energies are deduced from the water phantom study conducted above. It should be noted that we have so far only considered neutrons generated from the target phantom. Although ideal for our current investigation, in a real clinical setting there are numerous sources of neutrons (for example, the proton transport system, the particle accelerator, as well as the patient specific aperture and compensator).
The results show that neutrons are emitted with a wide spectrum of energies covering several orders of magnitude. The majority of escaping neutrons have energies ranging from about 10 to 100 MeV with the exception of the incident 70 MeV proton simulations that shows a higher amount of escaping neutrons between 1 and 10 MeV (Fig. 2). Although the neutrons >10 MeV have the highest emission amount for therapeutic proton energies, the neutrons≤10 MeV would be a better representation for constructing the background field as these neutrons are emitted quasi-isotropically primarily due to the pre-equilibrium and equilibrium models used in the simulation (pre-compound model, nucleon evaporation, low energy nuclear breakup models, etc.) as shown in Fig. 3. The contribution from the higher energy neutron component, which is predominantly forward peaked due to the cascade model mechanism, could be minimized by careful placement of the detector at angles orthogonal to the beam direction.
The photon component is also analyzed for the energy spectrum and angular distribution but not included in our radiation background spectrum. Instead we focus on the characteristics of the higher energy photons that would generate the majority of the signal used in the PR estimation. The resulting escaping photon energy spectra show various peaks including the de-excitation 12C (∼4.4 MeV) and 16O (∼6 MeV) photons as shown in Fig. 4. These photons are the main photons of interest in the PG method as their emission yields reach a maximum near the location of the Bragg Peak (BP) (shown for the 16O photons in Fig. 8, discussed later). The production of 12C arises from fragmentation reactions of the incident proton with the target 16O nuclei of the form 16O(p, x)12C. The photon energy spectra are independent of the incident proton energy for all primary beams considered. This result makes sense because the peaks represent the final de-excitation products from the interaction of the primary proton with the target material and is thus material dependent. The emission amounts, however, are influenced by the energy of the primary beam. The photon angular distribution is quasi-isotropic regardless of the energy as shown in Fig. 3. This isotropic behavior is due to the culmination of multiple processes (de-excitation mechanisms of energetic residual daughter nuclei fragments and Compton scattering). The low amount of forward escaping photons as compared to large angle escaping photons is attributed to attenuation through the phantom. Angular distributions for the remaining incident proton energies show similar results for both neutrons and photons escaping the water phantom.
Our results conclude that the PR estimation is composed of two components: A photon component representing the useful PG signal and a neutron component with energies≤10 MeV. The PG method considers both the 12C and 16O de-excitation photons as the primary signal. We therefore chose to simulate a mono-energetic beam of 6 MeV photons (about the de-excitation energy of 16O) due to its higher energy to mimic the signal and deduce the Cherenkov material’s response.
Secondary photon characteristics / Prompt gamma analysis
In this section we discuss additional properties of emitted photons by studying their timing and depth location characteristics. The emitted photon yields show a distinct difference in their emission times as shown in Fig. 5 with the majority of photons being emitted with times less than 10 ns after the initial proton interaction with the target nuclei. There is also a distinct difference in the energy spectrum for the photons emitted in≤10 ns and >10 ns as shown in Figs. 6 and 7, respectively. The photons emitted≤10 ns include the high energy photons (>2 MeV), such as those arising from 12C, 16O, and 2H (via neutron capture), with the majority of photons having energies less than ∼50 keV. The prominent peak shown for the photons emitted after 10 ns is mainly the annihilation peak at 511 keV. Lower energy photons (having energy <511 keV) are believed to be populated by the annihilation photons reduced by Compton scattering mechanisms with some additional contributions from 12C and 16O de-excitation mechanisms. The photons emitted in≤10 ns are considered to be the prompt gammas. Selected prompt gammas have maximum emission yields that lie in close proximity with the incident protons BP. These photons are used in the PG method for proton range verification.
The high energy de-excitation photons arising from 16O has its maximum emission yields in close proximity to the Bragg Peak. Fig. 8 displays the incident proton’s depth-dose distribution and the 16O de-excitation photon emission yields as a function of depth. The position of the Bragg Peak (∼9.0 cm) lies a couple of millimeters deeper compared to the peak of 16O de-excitation curve. This is due to the insufficient energy of the protons at these depths to interact in-elastically with the target oxygen nuclei to produce these photons. The 16O de-excitation photon curve has low emission yields until a few millimeters superficial to the location of the Bragg Peak, when it sharply rises to the maximum with no yields past the depth of the Bragg Peak. Also 16O is a nuclei abundant in most materials encountered in the patient. The above reasons make the de-excitation photon of 16O as one of the main photons of interest in the PG method for range verification in proton therapy. Despite the peaks not directly coinciding, the relationship between the maximum locations of the depth-dose curve and the 16O de-excitation photon yields (or any other depth metric) can be used for range verification.
Figure 8 also displays the yields of all photons (referenced as the total photon curve, shown as open squares) excluding the 511 keV annihilation photons. These photons have high emission yields that increase with increasing depth reaching a well-defined peak just before the location of the Bragg Peak where it sharply decreases. There is also photon yields extending past the Bragg Peak, which decrease with increasing depth. It is tempting to use the total photon curve for range verification due to the well-defined correlation between its peak and the depth-dose distribution. It must be mentioned that, with the exception of 16O de-excitation, these other photons included in the total photons curve may arise at lateral distances along the primary proton track and are not necessarily the products from the interaction of a proton onto the target nuclei (for example the capture gamma from the reaction n(X,Y)γ).
PR approximation study
We now focus our attention to the characterization of the Cherenkov material to photons and neutrons that mimic the emitted radiation field described above. We specifically study the interaction of neutrons produced from our background field by quantifying the average amount of optical photons generated in each material, and the total amount of optical photons striking the PMT window. Using this information we can estimate the PR for the Cherenkov glass sample that would be an indication of the viability of this material for use in the PG method.
The neutron results are shown first which would represent the unwanted signal in our PR approximation. This quantity is a measure of the amount of optical photons produced in the materials by a direct interaction from a neutron (which has deposited energy in the material). The background field comprised of a hundred thousand incident neutrons with energies between 1 and 10 MeV. The interaction efficiency of the Cherenkov glass sample, LYSO, and BGO materials to these neutrons are shown in Fig. 9 for a 4 cm length×3 cm diameter cylinder. The interaction efficiency represents the percentage of neutrons that deposit any amount of energy in the material divided by the total amount of incident neutrons, which is < 2% for the glass sample for all neutron energies considered. In comparison, the LYSO scintillator material had neutron interactions ranging from ∼75% to ∼50% depending on the incident neutron energy with similar results for the BGO scintillator. This result suggests that the background signal generated by the glass sample is 0.04 to 0.03 times less likely to be contaminated by neutrons for the neutron energies studied. The low interaction efficiency for neutrons in the glass sample is an advantage for proton therapy applications due to the intense neutron field especially with neutron energy≤10 MeV, which are highly abundant and emitted isotropically.
One hundred thousand 6 MeV photon beam are used to produce the signal in the PR approximation. The response of both the glass, LYSO, and BGO materials to this beam is given in terms of total interaction efficiency defined as the number of incident gamma rays that deposit any amount of energy in the material divided by the total number of incident gamma rays. The energy deposition can be from any process initiated by the incident 6 MeV photon (such as the photoelectric effect, Compton scattering, or pair production and related phenomenon). Our simulations average yield interaction efficiency for a 3 cm length×3 cm diameter glass sample is ∼20% that is consistent with published results of ∼19% [23]. The interaction efficiency for the 4 cm length×3 cm diameter cylinder for both the glass, LYSO, and BGO samples are ∼26%, ∼66%, and ∼67%, respectively, which suggests that the scintillator materials are better suited for the 6 MeV gamma ray detection by about 2.5 times. This is probably due to the higher density of the scintillator materials as compared to the Cherenkov glass material (∼7.1 g/cm3 compared to 2.83 g/cm3).
We define the ratio of the average amount of optical photons produced in the glass sample and the LYSO scintillator by the incident beam as which gives an estimation of the desired signal. The ratio for a 4 cm length×3 cm diameter cylinder for 6 MeV photons is 0.0023 which suggests that the amount of optical photons produced in the glass sample is ∼.2% compared to the LYSO scintillator. This result makes sense considering that the majority of optical photons produced in the LYSO material come from the scintillation process (∼32,000 optical photons per deposited electron energy in MeV) that is non-existent for the Cherenkov material investigated in this study.
The data collected from the above simulations are used in the estimation of the PR. For this approximation the ratio of the amount of optical photons produced from 6 MeV photons to that of neutrons () is multiplied by the interaction efficiency of both the 6 MeV photons and the various neutron energies resulting in for each material investigated. The amount of optical photons produced is normalized to equal amounts of incident particles of 6 MeV photons and neutrons. From our target phantom study as shown in Table 1, the ratio of yields for secondary neutrons to photons has a range of ∼.2 – ∼.6, respectively. We choose a 1:1 ratio of gamma to neutrons as a good approximation of the emitted particle field from realistic clinical scenario.
This corrected ratio is our estimation of the PR and is shown in Fig. 10. When corrected for the interaction efficiency of the incident particles, the glass sample generates at a minimum ∼40% more total optical photons from 6 MeV photons as compared to neutrons whereas the LYSO scintillator generates a maximum of ∼15% more total optical photons for the PR. Our interpretation of this result suggests that the glass sample would have much lower neutron compared to the LYSO scintillator. As mentioned earlier, current research to suppress the background neutron signal for scintillating materials is being conducted by using active or passive shielding techniques. We believe that using the glass material the shielding can be minimized and research can be dedicated for localization techniques.
The glass sample investigated thus has a high PR. The signal produced is orders of magnitude lower as compared to the signal from the LYSO due mainly to the low interaction efficiency of the 6 MeV photons. This is believed to be due to the lower density of the material. It was reported by Hayward et al. [23] that these glass materials could have higher densities and Zeff that would increase the performance and thus viability for use in proton applications. Nevertheless, the signal generated by the glass sample is due almost entirely to the incident 6 MeV photons; the neutron signal has been greatly suppressed. This cannot be said of the LYSO sample simulated where our PR approximation is considerably lower. All of these results and comments are from our very simplistic PR estimation and is provided as a proof of concept. Further research is needed before any definitive conclusions can be reached.
Cherenkov material characteristics
This section investigates the properties of the generated electrons and resulting optical photons from the interactions of the incident 6 MeV photons. This information is used to optimize the spatial dimensions for the simplistic detector segment mentioned above.
Secondary electrons are generated from the interaction of the incident 6 MeV photons by three main processes; photoelectric, Compton scattering, and pair production. As these secondary electrons traverse the glass they polarize the material. The depolarization of the medium back to the ground state results in the Cherenkov optical photons. This chain of events is initiated by the energy deposited by the 6 MeV photons. We quantified the generated secondary electron spectrum and the resulting optical photon energy spectrum as shown in Fig. 11. The secondary electron spectrum is quasi-constant with an upper limit threshold of production at ∼5.7–5.8 MeV and has a higher yield for lower energies due to Compton scattering processes. The optical photon wavelength spectrum has the highest yield at lower photon wavelengths (∼300 nm) and continuously decreases to the highest wavelength (∼725 nm), which is consistent with Hayward et al. [23]. The upper limit of the optical photon wavelength generated in our simulations is higher as compared to Hayward et al. [23] which can be explained by the parameters used in defining the properties of the glass sample in our simulation. Although the values of optical wavelengths range from 300 to 725 nm, the lower wavelength photons would be used in future experiments due to the increased amount of photons produced and practical reasons. This information is used for pairing the glass sample with a PMT for optimal transmission, thus minimizing signal loss.
The last characteristic of the glass study concerns the spatial dimensions of the glass sample. This is conducted as these dimensions dictate the amount of optical photons produced and absorbed. Simulations are conducted to quantify the amount of optical photons generated and traversing the glass material to strike the PMT window while varying the length of the glass as shown in Fig. 12. The amount of optical photons produced in the glass sample increases as the length of the sample increases. This is different as compared to the amount of optical photons striking the PMT window that show a maximum in photon collection with a length of 1.5 cm. This radius was chosen to match commercially available PMT tubes but could also be varied for future studies. Once the optical photons traverse the glass sample and enter the coupled PMT window the photons are terminated and no further tracking of these photons are made. Further investigations are planned for simulating the optical photons transportation through the PMT and other components of the detector system chain.
As the length of the glass sample increases, the probability of interaction increases, thus we would expect a greater amount of optical photons generated with increasing length. Likewise, as the length of the glass sample increases, the probability for an optical photon to be absorbed also increases, which is consistent with our results. There is little difference between the amount of optical photons that is collected when the length is 1 –3 cm but considerably more optical photons is produced at these lengths. Small sizes are favorable in commercially available detector systems thus we expect these spatial limits to be of use for production sizes.
Range verification with the Cherenkov material
The final objective is to validate if the Cherenkov glass material can verify the range of incident protons using the PG method. The setup was described above where the spatial dimensions were chosen to be 1 cm in length×1 cm in radius; small dimensions used only for signal localization and only as a proof of concept. This setup is expected to have very low efficiency.
The results obtained are shown in Fig. 13. The shown curves represent the depth-dose distribution of the incident protons and the emission yields of optical photons striking the PMT for different secondary particles categorized by their emission times. The depth-dose and the emission yields produced from secondary photons emitted < 50 ns show well defined peaks located at ∼9 cm and ∼8.7 cm, respectively, where the range could be calculated by suitable metrics. Optical photons striking the PMT window generated from all other secondary particles exhibit a low, flat response across all depths simulated by the detector segment. These include all neutron energies. These proofs of concept simulations are to illustrate that the Cherenkov detector segment could be used in conjunction with the PG method.
This is a desirable outcome since the detector segment has not been optimized in any regards but warrants comments on the utility of this result. The conservative timing window of 50 ns was chosen for time-of-flight discrimination which is a technique of synching the detector to the cyclotron beam pulse [34], other time-of-flight techniques have also been proposed (see [6]). The signal generated by the photons emitted in this timing window includes all prompt gammas. Again, it must be stated that this signal would include photons not necessarily generated directly on the proton path, but does include the important 16O de-excitation photons that would contribute to the signal in close proximity to the BP. A better range approximation can be deduced with the use of an energy threshold set to discriminate all but the 16O de-excitation photons combined with this timing window. The timing characteristics of the glass sample would be ideal for this type of application as it has a very fast response time paired with a suitable PMT, of the order of nanoseconds. These results seem very promising and motivate further research of this material for PG range verification of proton therapy treatments.
Conclusions
This study investigates the feasibility of a Cherenkov material for use with the PG method for proton therapy range verification. The radiation field escaping a water phantom from proton beams of various energies is examined where neutrons and photons have the highest yields. This information is used to characterize the Cherenkov material to these escaping particles and quantify the response.
The Cherenkov glass sample is ideal for the collection of the range information obtained from the emitted secondary photons. We found that the escaping 16O de-excitation photon, ∼6 MeV, is well correlated with the incident proton’s range, emitted in < 10 ns after the interaction that produced it, and emitted isotropically. The glass sample is capable of detecting and discriminating the 6 MeV photon, as well as process the information within a few hundred picoseconds (with a sample length of 3 cm [23]), suggesting that individual incident photon events can be counted.
Of the photons emitted in < 10 ns, which includes the important 16O photon, the majority have energies less than ∼50 keV. These photons would only contribute to signal noise. However, a benefit of the Cherenkov material is the ability to inherently discriminate these photons due to the high threshold energy for producing optical photons (∼150 keV for the sample studied), which can be varied by different chemical compositions.
The neutron component of the emitted secondary particle field was also greatly suppressed by the Cherenkov glass sample. Of the neutrons observed, the neutrons emitted with energies≤10 MeV are quasi-isotropically emitted. The sensitivity to these neutrons by the glass sample is found to be < 2%, suggesting that additional shielding to decrease the signal produced from these neutrons can be minimized or neglected completely.
The Cherenkov glass sample’s response to particles mimicking the secondary radiation field is studied and compared to both LYSO and BGO scintillators. The results suggest that the glass sample has an overall lower efficiency for detecting 6 MeV gamma rays. However, the efficiency can be increased for the glass sample by increasing the density and Zeff as mentioned by Hayward [23]. The ratio of the signal generated by the 16O photons compared to neutrons with energies between 1 and 10 MeV is increased with the glass sample compared to the LYSO scintillator. This suggests that the signal produced from both photons and neutrons can be discriminated greater with the glass sample as compared to the scintillator materials.
Proof of concept range verification simulations are conducted with a simple glass detector system segment onto a water phantom in order to characterize the viability of using this material to the PG method. It must be stressed that the simulations conducted for this purpose are artificial in that a sequential discrimination is implemented. Future work includes conducting the measurements and characterizing the detector system to more realistic scenarios.
Overall, this study suggests a novel material for use in proton therapy range verification utilizing the PG technique. Additional investigation on exploring the benefits of the Cherenkov material to range verification methods as well as improving the detector design for a clinical scenario is currently being planned.
Footnotes
Acknowledgment
The authors would like to acknowledge Dr. Jason Hayward for providing us with the optical photon settings of the simulations as well as data related to the Cherenkov glass samples used in this investigation.