Time-resolved plastic scintillator dosimetry in MR linear accelerators without image distortion

One of the newest developments within radiotherapy is the integration of Magnetic Resonance (MR) scanners and Megavoltage X-rays from linear accelerators into the new MR-Linacs. Some MR-Linacs are able to perform gated treatments based on continuously acquired 2D MR images taken during dose delivery, which has the potential to reduce margins required for tumor coverage. We have developed a dosimetry system that can provide time-resolved, dose-per-pulse, dosimetry without distorting the MR images in order to characterize this technology. The system is based on a plastic BCF-60 scintillation detector (PSD) coupled to an optical PMMA fiber cable. The detector was placed in a plastic tube filled with water and inserted into the piston of a dynamic MRI compatible phantom to be treated on a ViewRay MRIdian 0.35 T MR-Linac. The piston performed a sinusoidal movement to simulate a breathing cycle of either 4 or 8 s. We demonstrate that the detector system is able to provide real-time dose-per-pulse measurements without causing distortions in the MR images. The time-resolved dosimetry system revealed systematic dose rate transients during gated treatments that lasted about 1 s in every gating sequence where the beam turned on and off. Detection of such effects require real time dosimetry which does not interfere with the MRI based gating.


Introduction
Megavoltage (MV) radiotherapy with online magnetic resonance (MR) imaging became clinically available in 2017 (Raaymakers et al., 2017).This technology enables continuous monitoring of patients with high soft-tissue contrast, thereby making it possible to track the tumor during treatment (Corradini et al., 2019).Furthermore, as some linear accelerator systems can process image data in real time and switch the photon beam on and off based on the position of the tumor or the organs at risk, advanced gating techniques is now an option.Consequently, MR-Linacs have the potential to improve motion management compared to conventional linear accelerators, with a prospect of reducing margins required for tumor coverage and lowering the dose to organs at risk (Corradini et al., 2019).To exploit this new technology without compromising safety, it is relevant to characterize the advanced treatments by independent dosimetry methods including time resolved dosimetry with simultaneous MR based gating (Weidner et al., 2022).Dosimetry in a magnetic field is challenging as the field influences the secondary electron trajectories due to the Lorentz force (Jelen and Begg, 2019;O'Brien et al., 2017).For magnetic flux densities larger than about 0.3 T, this can lead to a significant perturbation of the electron fluence: The fluence will be different in the detector compared to the fluence in water in the absence of the detector.This is an issue, as detectors are normally calibrated without a strong magnetic field to give the dose in water in the absence of the detector (IAEA, 2001).The perturbation can be substantial if the density of the detector differs from that of water, and as a result, air-filled ionization chambers are not optimal for use in MR-Linacs (Reynolds et al., 2013;Meijsing et al., 2009).In addition, many of the conventional detectors contain materials such as graphite or metals, which disturb MR imaging by creating object artifacts (Choudhri et al., 2013;Stadler et al., 2007).
To avoid the problems of conventional detectors, different alternatives have been attempted.One option is plastic scintillators.They convert absorbed energy into light through scintillation emission, which is then guided through optical fibers to the optical readout instrument (Archambault et al., 2010;Beddar et al., 1992aBeddar et al., , 1992bBeddar et al., , 1992c)).The all-plastic nature of the detector should not lead to distortion of the MR images.Furthermore, the high degree of water equivalence with respect to electron density and atomic composition means that scintillators are essentially perturbation free in MV photon beams (i.e., the absorbed dose to the scintillator material is almost the same as to water).Unfortunately, scintillator dosimetry is challenged with other non-ideal characteristics.Most importantly, it is difficult to detect the emitted light from the scintillator, as light is also produced during irradiation of the optical fiber cable (stem signal) connecting the scintillator and the readout instrument (Beddar et al., 1992a;Archambault et al., 2005;Therriault-Proulx et al., 2013).It is therefore necessary to be able to isolate the 'pure' scintillator signal from the stem signal induced in the fiber cable.Another challenge is that low-energy electrons (<100 keV) tend to produce less scintillation light compared to high-energy electrons.This problem of ionization quenching (Peralta and Florbela, 2014;Frelin et al., 2008;Santurio et al., 2020) should, however, be of little importance in MV photon beams as long as all measurements are carried out using the same beam quality (Clift et al., 2000), because the fraction of low-energy electrons tends to be fairly constant even if different field sizes are considered (Santurio and Andersen, 2019).Finally, in the context of scintillator dosimetry carried out in strong magnetic fields, it is expected that the magnetic field will have an influence on the signal yield (Blomker et al., 1992;Green et al., 1995;Bertoldi et al., 1997), although this has not been demonstrated conclusively (Therriault--Proulx et al., 2018;Simiele et al., 2018;Stefanowicz et al., 2013) for the scintillator materials normally used in MV dosimetry.As the magnetic field in an MR-Linac is close to uniform this is, however, not a concern if the scintillator system is calibrated directly in the MR-Linac.
For conventional linear accelerators, research has established that plastic scintillation detectors can provide time resolved dosimetry and accurate dose measurements in a moving target (Beierholm et al., 2008(Beierholm et al., , 2015;;Sibolt et al., 2017;Lambert et al., 2006).It has furthermore been shown that plastic dosimeters can provide accurate dose measurements under static settings in MR fields (Therriault-Proulx et al., 2018), and that the amount of Cerenkov radiation generated depends on the direction and strength of the magnetic fields under these settings.To our knowledge, no study has examined the use of plastic scintillators for time resolved dosimetry in moving targets in MR-Linacs.To address this, we have developed and investigated the use of a plastic scintillator dosimetry system (Beierholm et al., 2011(Beierholm et al., , 2013)), where the scintillating detectors are coupled with PMMA optical fibres, for characterization of a 0.35 T MRIdian Viewray accelerator.Our hypothesis is that this system can perform real time dosimetry in the presence of a magnetic field without distorting the MR images.This is essential, as distortion could influence the ability of the accelerator system to analyze images, thereby leading to sub-optimal gating.
The purpose of this study was to demonstrate that plastic scintillators can provide accurate time-resolved dosimetry in moving targets in an MR-Linac, including 1) validation of MR-image compatibility (absence of image distortion), 2) stem correction relevant for the dynamic conditions, and 3) dose accuracy by comparison with alanine dose measurements.This was done through scintillator measurements in a Viewray MRIdian using a moving phantom and a simulated gating treatment.

Accelerators
The main linac used in this study was a MRIdian (Viewray, USA) with a 0.35 T magnetic field located at the Copenhagen University Hospital-Rigshospitalet, Denmark.This MR-Linac has one beam quality: a 6 MV flattening-filter free (FFF) photon beam.The tissue-phantom-ratio    (IAEA, 2001) was TPR 20,10 = 0.65.Supplementary measurements were performed in photon beams on a Varian Truebeam linac 6 MV FFF TPR 20,10 = 0.632 and 6 MV TPR 20,10 = 0.667 at the Technical University of Denmark.

Scintillator system
The plastic scintillation detectors used in this study were based on BCF-60 green scintillator material (1 mm diameter and 2 mm length, Saint Gobain) attached to a 9 m long 2 mm diameter PMMA optical fiber cable.
Scintillator signals were acquired by a ME40 in-house developed scintillator dosimetry system (Beierholm et al., 2011) which allows for simultaneous dose-per-pulse measurements with two fiber-coupled scintillator detectors.Stem signal and scintillation light were separated chromatically (Frelin et al., 2005) using yellow and magenta 45 • dichroic mirrors (Edmund Optics, USA) and separate light detection by two H5784 photomultiplier tubes (Hamamatsu, Japan).One photomultiplier tube (referred to as the 'green channel') mainly detected green light, and the other mainly detected blue light (referred to as the 'blue channel').For the measurements presented here, the system was synchronized with the linear accelerator such that the dose could be measured for each individual gun pulse using an 80 μs integration of light following each synchronization pulse, and a background correction based on measurements with an identical integration time delayed 200 μs from the gun pulse.The ME40 instrument was placed in a shielded area of the treatment bunker.

Alanine
Alanine dosimetry ensured traceability for the scintillator measurements in magnetic fields.The alanine pellets (1.15 mg/cm 3 mass density, 4.8 mm diameter, 2.8 mm thickness, Harwell, UK) consisted of 91% L-α-alanine powder and 9% paraffin wax, acting as binder.During measurements in water, the pellets were wrapped in parafilm.The pellets were read out using electron paramagnetic spin resonance spectroscopy with an EMXmicro system (Bruker, Germany) following the procedure outlined in (Ankjaergaard et al., 2021).Alanine measurements were calibrated in cobalt-60 and corrected for beam quality (Anton et al., 2013) and influence of magnetic field (Billas et al., 2021)

Other detectors
In the study of influence of the scintillator detectors on MR images, comparative measurements were carried out using a 3D semiflex ionization chamber 31021 (PTW, Germany), and two Al 2 O 3 :C crystals (a 6 mg TLD 500 crystal, (Ural Polytechnic Institute, Russia), and a 2.2 mg Stillwater Crystal Growth, Landauer USA) coupled to optical PMMA fibers.Supplementary measurements were carried out using EBT3 film.(Gafchromic, Ashland, US), which were read out on an Epson flatbed scanner and analyzed using the Risøscan software (Helt-Hansen and Miller, 2004).

Calibration
The calibration of the plastic scintillator detector was carried out in Solid Water (Gammex, USA) in the MR-Linac using the two-channel chromatic removal technique described by (Therriault-Proulx et al., 2013;Guillot et al., 2011).In this method the scintillator and fiber are irradiated twice, where the amount of fiber irradiated differs between the setups.Alanine pellets were used as a reference.

Temperature dependence
The temperature dependence of the scintillator was investigated in a temperature-controlled water bath operated at 10, 15, 20, 30, and 37 • C during irradiations in a Varian Truebeam accelerator.Measurements were also carried out for a blank PMMA fiber without a scintillator at 10, 20, and 37 • C. For each temperature point, the detector was irradiated multiple times with 6 MV FFF square fields with sides ranging from 1 cm to 20 cm.

Dose profiles
In plane dose profiles at 80 cm source-to-surface distance and 5 cm depth of water were acquired perpendicular to the vertical beam axis of the MR-Linac accelerator using a Beamscan MR 3D water phantom (PTW, Germany) with the plastic scintillator placed directly on top of a model 31021 ionization chamber (Fig. 1).EBT3 film dose profiles were measured in the MR-Linac using a Solid Water phantom with 5 cm as buildup and 14 cm as backscatter.

Simulated treatments
We conducted measurements during simulated treatments using an MRI compatible dynamic phantom (Model 008M, CIRS, USA), which contains a moveable piston intended to simulate a breathing cycle.The control software was set to move the piston sinusoidally with an amplitude of 12.83 mm and a given period (4 s or 8 s).In this setup, three types of simulated treatments were made: (i) static (no piston movement during the beam delivery), (ii) sinusoidal cylinder movement (4 s period) without MR-Linac gating, and (iii) sinusoidal cylinder A simple treatment plan consisting of a single rectangular 6 MV FFF field of size 13.28 cm × 19.92 cm (577.4MU) was delivered in all three cases.The treatment planning system (TPS) calculated a mean dose of 5.38 Gy/fraction for the target volume (a cylinder of 1 cm diameter, 2.5 cm length) of the static CIRS phantom assuming all components to be water.The minimum and maximum dose calculated by the TPS for the target volume was 5.24 Gy and 5.53 Gy, respectively.For each simulated treatment, either a plastic scintillator or a stack of six alanine pellets were placed in a water filled plastic tube, which was then inserted into the target volume of the moveable piston of the phantom (Fig. 2).
Due to the lower sensitivity of alanine, the treatment plan was run two or three times for these irradiations corresponding to 1154.8 MU and 1732.2MU respectively.The alanine pellets used with the gating function active received a total of 1154.8MU while the other pellets received 1732.2MU.All results were subsequently scaled to reflect the dose for one treatment fraction (577.4MU).

MR compatibility
The MR image compatibility for various detectors (plastic scintillators, Al 2 O 3 :C, alanine, and the 31021-ionization chamber) was investigated in the MR-Linac by placing each detector in turn for imaging in the CIRS phantom.The ionization chamber was not connected to an electrometer in order to minimize artifacts caused by electric fields.Imaging was carried out with a torso coil placed on top of and below the phantom using a True Fast Imaging with Steady State Precession (TRUFI) sequence, which is a type of balanced steady-state free precession (bSSFP) sequence, yielding a T2/T1-weighted contrast.All detectors were both imaged in a static setup and during movement with a frame rate of four images per second.

Uncertainty budget for dose calculations
As shown in Table 1, the uncertainty for the accumulated dose of the plastic scintillator measurements in this study is dominated by temperature corrections and positioning.The calibration part of the scintillator uncertainty was transferred from the uncertainty of the alanine pellets used to calibrate the scintillator signal to dose.The dosimetric impact of positioning uncertainty in the CIRS phantom was assessed from the dose gradients in the treatment plan.As both 10 and 15 Gy were used to irradiate the alanine pellet as an example Table 2 shows the uncertainty budget of a single alanine pellet irradiated with 15 Gy.The calibration part of the uncertainty is slightly higher for a 10 Gy pellet.Fig. 4. Dose profile comparison between an ionization chamber, EBT3 film, and the plastic scintillator for (a) a 10 cm × 5 cm field and (b) a 5 cm × 5 cm field.The ionization chamber and the scintillator enter from the negative side and measure along the longitudinal direction of the field (inline).The signal in the blue channel is dominated by the stem signal (gray), which increases linearly with the amount of fiber cable inside the beam.The stem corrected scintillator signal is shown in blue.

Temperature dependency
The temperature dependence of the signal from the blue and green channel are shown in Fig. 3(a) and (b), respectively.The stem-corrected scintillator signal is shown to decrease with temperature, Fig. 3(c), for temperatures between 10 • C and 37 • C. The data are well described by a linear regression (not shown) with a temperature coefficient of (− 0.547 ± 0.010) %/ • C, independent of the field size.The uncertainty (k = 1) of the temperature coefficient includes all known sources of errors.The temperature effect is associated with the scintillator itself, as the stem signal is dominated by Cerenkov light, which is generated by differences in the refractive index in the fiber and thus relatively temperature independent.This is qualitatively consistent with the data in Fig. 3, as the blue channel is dominated by the stem signal (Fig. 3(a)) whereas the green channel is a mixture of scintillator light and stem signal (Fig. 3 (b)).Since stem signal is present in both the blue and the green channel, the temperature coefficient found is for the total system and not the BCF-60 scintillator alone.

Dose profiles
Fig. 4 shows the results of the profile measurements for two different beam sizes.The ionization chamber and the scintillator entered from the negative side.Consequently, the stem signal was minimal at position − 100 mm and maximal at +100 mm, where the electrical cable or optical fiber extended over the full width of the beam.An insufficient stem correction for the plastic scintillator leads to an asymmetric dose profile resembling the uncorrected signal from the blue channel as shown in gray in Fig. 4.After correcting for the stem signal (blue curve), the resulting scintillator dose profiles are comparable with the profiles made with the ionization chamber and the EBT3 film.The upper panels in Fig. 4 show the relative difference between the scintillator and the ionization chamber.These residuals are used as a reference, as it is assumed that the ionization chamber stem signal is minimal, and that the magnetic field does not affect the ionization chamber significantly with respect to asymmetry.The residual plots reveal that the scintillator profile is not completely symmetric but has a 0.1% increase in apparent scintillator dose per cm of fiber in the primary beam.
Fig. 5 compares the symmetry measured by the different detectors.The ionization chamber measurement is highly symmetric around the center of the beam for the full profile.For the scintillator, the asymmetry Fig. 5. Dose profile comparison of symmetry between the side with a high amount of fiber in the beam and the side with a low amount for (a) a 10 cm × 5 cm field and (b) a 5 cm × 5 cm field.The ionization chamber highly symmetric whereas the scintillator is affected by the stem effect.This leads to an overestimation of the dose outside the field where only the fiber and not the scintillator is irradiated directly.The penumbra region is indicated by the vertical dashed lines.The top panel shows the percent differences between the two sides.
was insignificant inside the beam, but increased to about 20% in the low-dose region 10 cm from the beam center.The EBT3 film showed a higher degree of asymmetry and was generally more unstable than the other detectors.

Image distortion
To test the degree of image distortion caused by the different detectors, both static and dynamic MR images of the phantom were obtained for all the detector setups, i.e., a 31021-ionization chamber, two Al 2 O 3 :C crystals, two plastic scintillators and an alanine pellet (Fig. 6).
Alanine and the two plastic scintillators (Fiber 134 and 135) did not generate any noticeable image distortions compared to the reference image.The ionization chamber generated the largest image disturbance of all the detectors tested, whereas both Al 2 O 3 :C crystal detectors showed image distortions, but with the smaller crystal causing lesser artifacts.position relative to the gating window, shown as a red square.The gating window was placed in the center of the sinusoidal movement with a margin of 5 mm to each side.Fig. 8 shows accumulated doses measured with the plastic scintillator and alanine pellets in four different motion and gating conditions.A high degree of reproducibility between repeated measurements was found for the scintillator.The six alanine pellets were placed in a plastic tube with pellet 1 furthest inside the piston.The maximum differences (±1 standard deviation) between the mean alanine dose and the scintillator measured dose was (0.17 ± 0.11) Gy.

Time-resolved measurements
Fig. 9 shows the dose-per-pulse during single sinusoidal periods of either 8 or 4 s duration.The dose-per-pulse was slightly below 0.9 mGy (indicated with a dashed line) when the piston was inside the gating window and zero otherwise.The MR-Linac delivered approximately 100 pulses per second.The durations of the beam-on periods are shown in the panels.There are two beam-on events per period as the piston enters the gating window both in the forward and reverse movement.
Three repeated simulated treatments with sinusoidal periods of either 8 or 4 s duration (identified as run 1-3) are shown in Fig. 10, where the duration of each beam-on period is plotted as a function of the beam-on start time.The data groups into clusters separated by 0.25 s (indicated by dashed horizontal lines), which equals the inverse frame rate (4 Hz) of the MR-Linac during these measurements.Under ideal conditions (perfect image registration, no latency and infinite frame rate), the duration of the beam-on period is the time it takes the piston to pass the gating window, i.e., move from position − 5 mm to position +5 mm.The corresponding theoretical durations are 0.51 s and 1.02 s for the 4 s period and the 8 s period, respectively.Deviations from the calculated durations reflect the detection and response latencies of the MR-Linac, i.e., the fact that position is always detected with a delay due to finite frame rate and image acquisition and processing time.Imperfect image processing adds to this discrepancy by adding uncertainty to the position detected.
Fig. 10 shows, that there was no correlation between the event number and the length of the beam-on events within a treatment, indicating that the image recognition was consistent throughout the treatment.However, between treatments this was not consistent for the 4 s period; run 1 has significantly shorter-than-expected beam-on events compared to the other two runs.

Dose transients
As shown in Fig. 11, the dose-per-pulse within each gating window increased by approximately 10% during the initial 1 s, after which it was relatively stable.This transient behavior prevailed consistently throughout all measurements but was found to be largest during the initial part of the treatment.The accelerator monitor chamber is expected to appropriately account for the transients, such that the total delivered dose is correct, whereas the decreased dose rate at entries may impact where the dose is delivered.We found that the dose-per-pulse increased for both the continuous beam and the gated beam throughout the full treatment.In case of the non-gated treatment, the increase was about 2% from the first to the last period.

Suitability of the detector system
The purpose of this study was to present a measuring system capable of performing time-resolved dosimetry without image distortion in an MR-Linac.Our results show that neither the plastic scintillator nor the PMMA fiber cause any artifacts on the MR images, which is a requirement for the system to be useful for testing advanced gating treatment plans.Furthermore, the images of the alanine pellets were also without image distortion, which enhances their usefulness as reference dosimeters.Any effect of the magnetic field on the scintillator's light production, such as ionization quenching, was included in the calibration procedure.
The dose profiles in Figs. 4 and 5 show that the stem signal is reduced significantly, when the scintillator is inside the beam.This indicates that the chromatic removal method is a sufficient stem removal technique even in a dynamic setting with an MR field included.Previous studies have found that the Cerenkov component of the stem effect changes with an added magnetic field, but that the spectral distribution stays the same (Maraghechi et al., 2020;Simiele et al., 2020Simiele et al., , 2021)).Other studies have successfully corrected for the stem effect in static setups (Therriault--Proulx et al., 2018;Stefanowicz et al., 2013), while this study is the first, to our knowledge, that shows the removal of the stem effect for a dynamic setup in an MR-Linac where the amount of stem (in beam fiber) changes during the scan.
We saw a residual stem effect after the application of the chromatic removal in the order of 0.1% per cm PMMA fiber inside the beam.This caused up to a 20% overestimation of dose when the scintillator was in the low-dose region outside the beam where only PMMA was directly irradiated.While this residual uncertainty should be investigated further, it is of little importance for applications, such as the present study, where all subsequent measurements were carried out with the scintillator in the primary beam.During measurements the temperature of the system and the photo multiplier tubes (PMT) were monitored.For the PMTs the temperature was stable within 0.02 • C. Since the PMTs have a temperature sensitivity of about 0.1%/ • C no correction was added.
The temperature dependence of the system was quantified in order to dismiss its influence on the residual stem signal or the calibration.We found a significant temperature dependence of the system of (− 0.547 ± 0.010) %/ • C independent of field size.This temperature coefficient matches the results found by (Buranurak et al., 2013;Wootton and Beddar 2013).
Lastly, we saw minor to no difference between the dose measured with the alanine pellets and the integrated plastic scintillator dose, where the difference between the mean alanine dose and scintillator Fig. 9. Example of a time-resolved measurement using the ME40 system.One sinusoidal period is illustrated for both the 8 s duration (top panel) and the 4 s duration (lower panel).One period consists of two beam-on events.dose was within (0.17 ± 0.11) Gy.This is most likely due to differences in the exact positioning of the two detector types, as the difference is within the expected percentage difference for different positions in the field.The volume of the pellets was much larger than the plastic scintillator and, therefore, dose was integrated over a larger area compared to the plastic scintillator.Because of this, the pellets are differently affected by dose gradients compared to the plastic scintillator.

Application of the ME40 system
The ability of the ME40 system to perform dose-per-pulse measurements and the possibility of tracking the dose locally and continuously during a treatment.Two applications were presented in this study: In the first case, the ME40 system was used for measuring beam-on periods in a gated treatment as shown in Fig. 10.The number of beam-on events was significantly different among the three 4 s period treatment runs, but the total dose measured for each run was the same (5.10Gy, Fig. 10).One would expect that beam-on time for the 4 s period would be twice the size of the 8 s period, but it is even longer.This is due to the shorter period being relatively more affected by an additional frame in the gating window.As one frame is 50% of the expected time in the gating window for the 4 s period, it is also 50% of the expected dose per beam-on event.However, it is only 25% for the 8 s period.
The second use case is the measurement of dose rate transients at beam-on during gated treatments.As shown in Figs. 9 and 11, each beam-on is followed by an increase in dose rate.Furthermore, it can be seen in Fig. 11, that the beam occasionally stayed on after the piston left the gating window.This is most likely due to the frame rate not being in sync with the piston movement.The transients should not affect the amount of monitor units delivered as the accelerator monitor chamber will compensate for this.It was not investigated whether the systematic transients could lead to hot or cold spots in the dose delivery, or if it would have other clinical consequences.The accumulated treatment doses obtained with alanine and scintillators in Fig. 8 did not indicate any such systematic deviations, but this does not exclude systematic deviations at positions closer to the beam edge.
The changes in dose amplitude for the continuous beam in Fig. 11 was caused by a dose gradient across the beam, as the scintillator moved back and forth in the beam.These results hold great promise for the ME40 system as all requirements for the system to be used dynamically in a MR-Linac were met.This system could pose as a serious contender for online dosimetry in MR-Linacs.The absence of image distortion along with the ability of time-resolved dosimetry is of high value for clinicians, who want to test the accuracy of the advanced gating plans that are now made possible by the MR-Linacs.

Conclusion
We demonstrated a plastic scintillation detector-based dosimeter system, ME40, which can provide real time-resolved dose measurements without disturbing the clinical imaging system.The system revealed systematic dose rate transients during gating measurements.Detection of such effects requires real time dosimetry, which does not interfere with the clinical gating method.The clinical impact of these dose rate transients in the form of cold or hot spots was not examined in this study and are likely depending on the exact gating procedure.For challenging cases with low duty cycle, it should be considered to what extend warmup or latency effects could perturb the delivered dose.

Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: C. P. Behrens and I. R. Vogelius reports research and teaching contracts with Varian Medical Systems, ViewRay, and Brainlab outside of the submitted work.

Fig. 1 .
Fig. 1.Illustration of the scintillator and ionization chamber placement during dose profile scans in the MR-Linac.

Fig. 2 .
Fig. 2. Alanine pellets wrapped in parafilm and placed in the target volume of the CIRS phantom piston.The pellets were numbered 1 to 6.The target material is made of polyurethane which is visible on MRI images.

Fig. 3 .
Fig. 3. Temperature dependence for the blue channel signal (a), the green channel signal (b), and the stem-corrected scintillator signal (c).All measurements are normalized to the respective field size at 10 • C.

Fig. 7
Fig.7illustrates the gating setup, with the upper panels showing the sinusoidal movement of the piston, and the lower panels the piston

Fig. 6 .
Fig. 6.Comparison of MR images for different detectors versus a reference image obtained without any detector.All images were obtained for static conditions.Significant distortion is indicated with a red box, a larger box indicates a larger distortion.A green box centered at the detector indicates absence of distortion.

Fig. 7 .
Fig. 7. Illustration of the gating setup.The top panels show three different piston position (red) within half a period of the sinusoidal function.The lower panels show the corresponding physical position.

Fig. 8 .
Fig. 8.Comparison of accumulated doses obtained using repeated plastic scintillator measurements and six alanine pellets.The error bars reflect one standard uncertainty (k = 1) without common uncertainties.Repeated scintillator measurements were obtained without changing the position of the scintillator.

Fig. 10 .
Fig. 10.Top panels show the beam-on duration for each time the beam was turned on during the gated treatments, with the treatment using an 8 s period, and a 4 s period in the lower panels.The black dashed lines are separated by 0.25 s.

Fig. 11 .
Fig. 11.Differences in dose-per-pulse measured with and without gating (continuous beam) for a 4 s period.At the start of every beam-on, a dose transient lasting 1 s or more can be observed.The magnitude of the transient decreased over time during the treatment, which is evident from the gated beam when comparing the first three beam-on periods with the last three beam-on periods.

Table 2
Summarized alanine uncertainty budget.