In-Vivo Dosimetry for Ultra-High Dose Rate (UHDR) Electron Beam FLASH Radiotherapy Using an Organic (Plastic), an Organic–Inorganic Hybrid and an Inorganic Point Scintillator System

Abstract

Dosimetry is crucial in radiotherapy to warrant safe and correct treatment. In FLASH radiotherapy, where ultra-high dose rates (UHDRs) are used, the dosimetric demands are more stringent, requiring the development and investigation of new dosemeters. In this study, three prototype fiber-optic dosemeters (FODs)—an inorganic, an organic–inorganic hybrid metal halide, and an organic (plastic) scintillator are optimized and investigated for UHDR electron irradiations. The plastic FOD is developed by Medscint, whereas the others are in-house made. The stem signal is minimized by spectral decomposition for the plastic scintillator, and by band-pass wavelength filters for the inorganic and organic–inorganic hybrid metal halide FOD. All prototypes are tested for the dose rate defining parameters. The optimal band-pass wavelength filters are found to be centered around 500 nm and 425 nm for the inorganic and organic–inorganic hybrid metal halide FODs, respectively. A sampling frequency of 1000 Hz is chosen for the inorganic and organic–inorganic hybrid metal halide FODs. The plastic FOD shows to be the least dose rate dependent with maximum deviations of 3% from the reference for the relevant beam settings. The inorganic and organic–inorganic hybrid metal halide FODs, in contrast, show large deviations of >10% from the reference and require more investigation. The current FOD prototypes are insufficient for application in UHDR electron beams, and require further development and investigation.

1. Introduction

Cancer treatment involves radiotherapy for over 50% of patients [1]. This treatment strategy is limited by the toxicity to the healthy tissue surrounding the tumor. The recent rediscovery of the FLASH effect could be a tipping point in the radiotherapy field [2]. In this, the use of ultra-high dose rate (UHDR) deliveries shows a strong decrease in radiation-induced normal tissue toxicity, while maintaining the anti-tumoral effect for a wide variety of organs and tumor types [3,4]. Therefore, FLASH radiotherapy (FLASH-RT) can potentially be used to treat more radioresistant pathologies and/or reduce the adverse effects observed with treatments that already have a good prognosis. In addition, the treatment times in FLASH-RT are negligible compared to organ motion and allow a decrease in the target volume [5]. The use of UHDR requires an update of the current dosimetry, as most dosemeters saturate at such dose rates, or are not capable of measuring all relevant beam parameters. This is crucial since the radiobiological origin and beam parameters needed to obtain the FLASH effect are yet to be determined.

The current state-of-the-art dosemeters used clinically in conventional radiotherapy are ionization chambers. The reading of such device requires correction for, amongst others, ion recombination, which for UHDR beams exceeds 70%. The magnitude of this correction compromises the accuracy of the dose measurement and the safety of the treatment [6,7]. Therefore, forces were bundled in the joint UHDpulse research project to develop dosemeters and redefine dosimetric standards for UHDR dosimetry [8]. This led to the development of the flashDiamond detector [9], an ultra-thin parallel plate ionization chamber [10], and portable calorimeters [11]. However, while suitable for integrated dose determination in pulsed UHDR electron beams, these dosemeters do not allow dose assessment of the individual pulses. Luminescence dosimetry based on scintillation or radio-luminescence (RL) is one of the proposed strategies due to its high temporal and spatial resolution, and its theoretical dose rate independence. The combination of scintillation/RL materials with fibers optics have emerged as a promising technology for dosimetry in radiotherapy. Researchers worldwide have developed and tested a wide range of scintillating/RL materials, both organic and inorganic. In parallel, significant progress has been made in devising reliable methods to eliminate or reduce the Cherenkov/stem effect, a major limitation of fiber-optic dosemeters (FODs). As a result, numerous prototype systems have been created, with some advancing to commercially viable solutions [12,13,14]. Therefore, the stem signal needs to be minimized, and a trade-off between the signal-to-noise ratio and temporal resolution needs to be taken into account. Point scintillator prototypes have been developed with [15,16,17,18] and without [19,20] individual pulse measurement. The capability to measure individual pulses depends strongly on the sampling rate of the readout system. In previous studies, pulse discrimination was limited to pulse repetition frequencies of 180 Hz when no triggering was used and 360 Hz in the case of triggered acquisition. In the work of Ashraf et al. [15], the readout was triggered based on the pulses, allowing individual pulse measurements up to a pulse repetition frequency of 360 Hz. In this work, an inorganic, an organic–inorganic hybrid metal halide, and an organic (plastic) scintillator are optimized and tested as in vivo dosemeters (IVDs) for UHDR electron irradiations. Organic–inorganic hybrid metal halides have gained attention as potential scintillators due to their dosimetric characteristics [21,22,23]. Accurate real-time IVD is needed to describe the radiobiological mechanisms of the FLASH effect and to use it to its full potential [24]. Our aim is to establish suitable dosimetric protocols using FOD-based point scintillator systems, with the focus on exploring the potential advantages of inorganic, organic–inorganic and plastic fiber types for UHDR applications.

2. Materials and Methods

2.1. UHDR Irradiation

All irradiations were performed with the ElectronFlash (SIT, Aprilia, Italy) linear accelerator. This is a research machine developed for preclinical FLASH research. It provides pulsed electron beams of 7 and 9 MeV, of which systematic variation in the dose rate modality (FLASH/standard), number of pulses (Np), pulse repetition frequency (PRF), and pulse length (PL) is available. It comes with a set of applicators with different diameters and lengths for further (instantaneous) dose rate (iDR) variation. The following terminology and definitions are used to characterize pulsed ultra-high dose rate (UHDR) electron beams:

• DPP = ${\int }_0^{PL}$iDR
• D = DPP·Np
• DR = ${\displaystyle \frac{\mathrm{D}}{\frac{\mathrm{Np}-1}{\mathrm{PRF}}+\mathrm{PL}}}$

with DPP being the dose delivered by a single pulse, iDR the instantaneous dose rate or dose rate within the pulse, D the dose, and DR the average dose rate. A representation of a pulsed beam using these definitions is shown in Figure 1.

2.2. Inorganic and Organic-Inorganic Hybrid FODs

The inorganic and organic–inorganic hybrid FODs are in-house-made prototypes, using Y₃Al₅O₁₂:Ce³⁺ (YAG) and (C₃₈H₃₄P₂)MnBr₄ (Organo-Br) as scintillating materials, respectively. The inorganic YAG scintillator emits light with a wavelength peaked at 551 nm, corresponding to the 5d→4f transition of the Ce³⁺ ions [25,26], and has a decay time in the order of ns. The Organo-Br scintillator emits light with a wavelength peaked at 517 nm corresponding to the 4T→6A transition of the Mn²⁺ ion, and has a decay time in the order of 0.4 ms [21]. The active end of each point scintillator is composed of the scintillating material, in powder form, mixed with Norland Optical Adhesive 68. The optical fiber consists of a cladded polymethyl methacrylate (PMMA) core, with a diameter of 1.0 mm, covered with a black polyethylene jack with an outer diameter of 2.2 mm (ESKA™, Industrial Fiber Optics, Tempe, AZ, USA). The active end is covered by a heat-shrinking tube to prevent light leaks. The fiber is polished to ensure optimal light transmission, and is coupled to a readout system that consists of a wavelength filter, a photo multiplier tube (PMT) and a USB-6353 data acquisition card (National Instruments, Austin, TX, USA), housed in a light-tight black box. The readout is controlled via a LabVIEW (National Instruments, Austin, TX, USA) script [13,27,28].

2.2.1. Stem Signal Reduction

Stem signal reduction was performed by optical filtering. Three band-pass wavelength filters (Edmund Optics, Barrington, NJ, USA) were tested to optimize the stem signal removal.

Table 1. The tested band-pass filters to minimize the stem signal.

Central Wavelength	Full Width at Half Maximum (FWHM)	Alias	Stock Number
425 nm	50 nm	WF425	86-937
500 nm	80 nm	WF500	65-734
525 nm	50 nm	WF525	86-951

shows the properties of the tested filters.

Three irradiations of 20 pulses with pulse length of 4.5 µs, a PRF of 245 Hz and an energy of 9 MeV, in UHDR modality were performed for each filter. The resulting signal was investigated in terms of magnitude and decay time to determine the optimal wavelength filter. Also, the contribution of the stem signal was investigated by varying the portion of optical fiber that was exposed to the beam as shown in Figure 2.

2.2.2. Optimizing Sampling Frequency and Integration Window

Different combinations of the sampling frequency and integration window were tested to optimize the sampling such that the temporal resolution was sufficiently high to dissolve the pulses but not too high to prevent buffer overload and low signal-to-noise ratio (SNR). The integration window was always chosen to be the inverse of the sampling frequency. For the YAG FOD, sampling frequencies of 200, 500, 1000, 2000, and 5000 Hz were investigated. For the Organo-Br FOD, the sampling frequencies were 200, 1000, and 5000 Hz.

2.3. Plastic Scintillator FOD

The plastic scintillator FOD is the HS-RP200 (MedScint, Quebec City, QC, Canada), which is a point-dose measuring device. The system is based on a hyperspectral approach for the stem effect removal, allowing for accurate dose measurements. It is calibrated by obtaining the spectral signatures of scintillation, fluorescence, and Cherenkov radiation. The signature of the scintillation was obtained by irradiation with the kV source on a TrueBeam (Varian, Palo Alto, CA, USA). The fluorescence and Cherenkov signatures were obtained by irradiating the scintillator twice with a different portion of the fiber in the field using the 9 MeV electron beam of a TrueBeam and a field size of 10 × 10 cm². A spatial binning parameter can be used to optimize the signal-to-noise ratio (SNR) and define the dynamic range of the system. For UHDR irradiations, the signal is high, allowing a low spatial binning, resulting in a shift in the dynamic range towards higher dose rates. In this work, a spatial binning of 9 pixels was chosen. The dosemeter is capable of independently reading up to four channels simultaneously. In this work, only one channel was used. The scintillator FOD is made of a 1 mm diameter by 1 mm length cylinder attached to a PMMA optical fiber. The probe optical assembly is also embedded in a Nylon 12 encasing for optimal water equivalency. The system can be calibrated to a conventional dose rate, then used at FLASH dose rates. In this work, a dose of 5 Gy was used for the dose calibration.

2.4. Use in UHDR Electron Beams

All prototypes were tested for their dose linearity, their iDR, DR, and DPP independence, and their ability to resolve individual pulses. The experiments consisted of varying the number of pulses (experiment 1), the PRF (experiment 2), the PL (experiment 3), and the source-to-surface distance (SSD) (experiment 4). The Hyperscint RP-200 was further investigated in a conventional modality and for a 7 MeV beam by repeating experiment 3 (experiment 5) and experiment 1 (experiment 6), respectively. Finally, the possibility for individual pulse detection was investigated for all FODs (experiment 7). All experiments were performed in triplet for all prototypes, and the reported error bars represent one standard deviation. For the YAG and Organo-Br FODs, the results of experiment 1 were used to calibrate the fibers via linear regression with intercept forced to 0. During experiments 1–4, the plastic FOD operated at a sampling rate of 1 Hz, with an integrating window of 1 s. When investigating its pulse discrimination ability (experiment 7), the sampling rate was increased to 8.33 Hz, which is the highest sampling rate for the chosen spatial binning. The integrating window was decreased accordingly.

Table 2. Used system parameters for calibration, and experiments for all FODs. Experiments 5 and 6 are marked with a “*” to emphasize that these experiments were only performed for the Hyperscint RP-200 FOD.

Parameters for Calibration and Characterization
Experiment/ Calibration	Modality	Np	Dose [Gy]	Nominal Pulse Length [µs]	PRF [Hz]	Energy [MeV]	SSD [cm]
Calibration Plastic	Conventional	120	9.207 ± 0.049	1	10	9	106
Exp. 1	UHDR	2–35	2.663–47.493 ± 0.071	1.5	10	9	106
Exp. 2	UHDR	7	9.347± 0.025	1.5	1–245	9	106
Exp. 3	UHDR	7	1.83–30.20 ± 0.37	0.5–4.5	10	9	106
Exp. 4	UHDR	7	2.247–9.249 ± 0.045	1.5	10	9	106–175.5
Exp. 5 *	Conventional	120	9.02 ± 0.22	1	1–245	9	106
Exp. 6 *	UHDR	2–35	1.543–47.493 ± 0.071	1.5	10	7–9	106
Exp. 7 Plastic	UHDR	8	10.634 ± 0.015	4.5	4	9	106
Exp. 7 (In)-organic	UHDR	7	8.6	4.5	245	9	106

gives an overview of the used beam parameters for the different experiments.

For both the in-house and the Hyperscint RP-200 prototypes, an RW3 slab was made with a dedicated cutout or insert to position the point scintillator in an RW3 water equivalent phantom. The buildup was 7 mm for the YAG and Organo-Br FODs and 15 mm for the plastic FOD. At least 5 cm of water equivalent RW3 was used for backscatter. A flashDiamond detector (PTW, Freiburg, Germany) was positioned 5 mm distal to the point scintillator in the phantom and served as a reference dosemeter after depth correction [9,29]. An applicator with a diameter and length of 10 and 73.5 cm, respectively, was used.

3. Results

3.1. Stem Signal Reduction: Optimizing Wavelength Filters

The peak signals for the inorganic and organic–inorganic hybrid FODs are shown in

Table 3. The peak signal for the YAG and Organo-Br FOD for the different band-pass wavelength filters for nominal pulse lengths of 4.5 and 1.5 µs, and the ratio between the peak signals from the different pulse lengths using the same wavelength filter.

Point Scintillator	Wavelength Filter	Peak Signal for 4.5 µs Pulse [Counts]	Peak Signal for 1.5 µs Pulse [Counts]	Ratio
YAG	WF425	7.9·${10}^3$	2.1·${10}^3$	3.84
	WF500	614.1·${10}^3$	150.1·${10}^3$	4.09
	WF525	97·${10}^3$	36.2·${10}^3$	2.68
Organo-Br	WF425	4.3·${10}^3$	2.3·${10}^3$	1.83
	WF500	141.8·${10}^3$	277.9·${10}^3$	0.51
	WF525	471.2·${10}^3$	593.5·${10}^3$	0.79
Ratio between effective pulse lengths				3.31

for nominal pulse lengths of 4.5 and 1.5 µs, corresponding to effective pulse lengths of 3.91 and 1.18 µs, respectively. The YAG FOD has the highest signal for the WF500, and the ratio of the effective pulse lengths is matched best when using the WF425. For the Organo-Br FOD, the highest signal is obtained with the WF525, and the ratio of the effective pulse lengths is matched best when using the WF425. It should be noted that the response ratio between the 4.5 and 1.5 µs nominal pulse lengths for the Organo-Br is <1 for the WF500 and WF525, suggesting lower response for a longer pulse.

Figure 3 shows the response of the YAG (B) and Organo-Br (C) FODs with various portions of the fiber in the field (A). The envelope of the YAG FOD is similar for all positions. It shows a narrow peak (FWHM < 0.26 ms), followed by a smaller, broader peak with a decay time of >3 ms. The magnitude of the broader peak in position 3 is less than that for positions 1 and 2, albeit not significant. The envelope of the Organo-Br FOD in position 3 differs from the one in positions 1 and 2. When a relevant portion of the optical fiber is exposed to the beam, the pulse results in two overlapping peaks, one narrow (FWHM < 0.31 ms) with high magnitude, and one broader with lower magnitude. For position 3, where the stem signal is minimized, only the lower, broader peak remains.

3.2. Optimizing Sampling Frequency and Integration Window

The response of a single pulse irradiation using different sampling frequencies is shown in Figure 4. Higher sampling frequencies result in higher peak values. This comes with higher noise levels, both in the signal and in the tail. Only for 2000 and 5000 Hz sampling frequencies can the smaller, broader peak be observed, as it is dissolved in the lower sampling frequency curves. The YAG FOD shows a longer decay time than the Organo-Br one, as the tail of the curves does not return to zero. In addition, the signal is more noisy compared to the one of the Organo-Br FOD.

3.3. Application in UHDR

3.3.1. Experiment 1: Linearity with Number of Pulses

The Hyperscint RP-200, YAG, and Organo-Br FODs show a linear response with the delivered pulses with an R² of >0.999, 0.999 and >0.999, respectively. The resulting graphs are shown in Figure 5A. Figure 5B shows the deviation of the dose measured with the point scintillators from the reference dose, obtained with the flashDiamond. The calibration factors for the YAG and Organo-Br FOD, obtained from this experiment are given in

Table 4. Calibration factors obtained via linear regression from experiment 1.

FOD	Calibration Factor [Gy/Counts]	Uncertainty [Gy/Counts]
YAG	2.967·${10}^{-6}$	3.6·${10}^{-8}$
Organo-Br	7.603·${10}^{-5}$	5.1·${10}^{-7}$

.

3.3.2. Experiment 2: Pulse Repetition Frequency Dependence

The response of the Hyperscint RP-200 shows small variations with varying PRF, with readouts ranging between 9.18 and 9.31 Gy, and being within 1% of the average. The YAG and Organo-Br FOD on the other hand, as shown in Figure 6, show larger deviations with varying PRF. Also, the uncertainties associated with the dose assessment of these detectors are larger. The YAG FOD shows a trend of increasing response with increasing PRF, with readouts ranging between 8.70 and 11.94 Gy, where the reference detector gave 8.35 Gy on average. A maximal discrepancy with the reference detector of 43.3% is observed at a PRF of 200 Hz. The Organo-Br FOD shows no trend but does show a large variation in readout, ranging between 6.94 and 8.62 Gy, where the reference detector gives 8.24 Gy on average. Especially at 1, 200, and 245 Hz, large discrepancies (>10%) with the reference detector are observed.

3.3.3. Experiment 3: Pulse Length Dependence

The dose per pulse increases linearly with increasing pulse length for the Hyperscint RP-200 and YAG FOD, albeit with a different slope, as shown in Figure 7. The dose per pulse response of the Organo-Br FOD increases monotonically with pulse length, but deviates from linearity for pulse lengths between 3 and 4 µs.

3.3.4. Experiment 4: Source to Surface Distance Dependence

The response of the FODs is shown against the 1/SSD² in Figure 8 because of the inverse square law. Both the Hyperscint RP-200 and the YAG FOD show a linear behavior, while the Organo-Br FOD does not. For SSD > 136 cm, the response is as expected, whereas for shorter SSD, a decreasing response with decreasing SSD is observed.

3.3.5. Experiment 5: Conventional Pulse Repetition Frequency Dependence

The dose reading of the Hyperscint RP-200 decreases with increasing PRF from 9.31 to 9.18 Gy and from 10.73 to 9.69 Gy in UHDR and conventional modality, respectively (Figure 9). For UHDR modality, this leads to an increasing deviation from the reference from −0.28 to −1.58%, whereas a fluctuating deviation from reference is observed in conventional modality with a maximum of −2.09%.

3.3.6. Experiment 6: Energy Dependence

Both in the 7 and 9 MeV beam, the Hyperscint RP-200 shows a linear response with the number of delivered pulses (Figure 10). For both energies, this FOD shows a trend in the deviation with the reference detector (Figure 10B). In the 7 MeV beam, the dose is overestimated for a low number of pulses, up to 2.7%, achieves better agreement with the reference detector for the 20 pulse irradiation, and underestimates the dose for a high number of pulses, up to −1.71%. In the 9 MeV beam, the dose is underestimated for all data points, with the magnitude of the underestimation increasing with the number of pulses from −0.48 to −2.95%.

3.3.7. Experiment 7: Pulse Discrimination

Figure 11 and Figure 12 show the pulse discrimination capabilities of all prototypes. The Hyperscint RP-200 is able to discriminate pulses up to a PRF of 4 Hz, due to the sampling rate limitation associated with the spectral binning. For a spectral binning of 9 pixels, this is 8.33 Hz. The YAG and Organo-Br FODs, on the other hand, are able to discriminate pulses up to the maximum PRF tested (245 Hz). The YAG FOD shows a longer decay time, resulting in overlapping pulses for this PRF.

4. Discussion

Point scintillators are promising dosemeters for UHDR beams. However, optimization is needed before they can be used routinely. One of their biggest drawbacks is the presence of a non-negligible stem signal. A well-known strategy to reduce the contribution of the stem signal is optical filtering [14]. For the YAG and Organo-Br FODs, optical filtering was performed and optimized within the available set of filters. From the peak emission wavelength of the YAG FOD, WF525 was expected to return the highest signal. However, the highest signal was observed for the WF500 filter due to its larger FWHM. Since also the peak ratio between the 4.5 and 1.5 µs nominal pulse length is acceptable for this filter, the WF500 was chosen as the optimal wavelength filter within the tested set of filters. For the Organo-Br FOD on the other hand, the highest signal was, as expected, observed for the WF525. However, as the peak ratio between the 4.5 and 1.5 µs nominal pulse length was <1 for this filter, the WF425 was chosen as the optimal wavelength filter within the set of tested filters. There is no physical reason explaining why the peak ratio is <1 for WF500 and WF525. More investigation is needed to be able to understand and explain this behavior. This requires tests with a spectrometer and shows the benefit of spectral decomposition or analysis, as used in the Hyperscint RP-200, over optical filtering.

The impact of the stem signal was investigated by irradiating different portions of the optical fiber while keeping the scintillating head in the field. The signal from the YAG FOD is minimally affected by the stem signal. The amplitude of the narrow peak does not differ significantly, regardless of which portion of optical fiber is irradiated. There is a slight decrease in magnitude for the broader peak and the decay when a smaller amount of optical fiber is in the field. The opposite can be seen for the Organo-Br FOD. The position where a minimal amount of optical fiber is in the field shows that only the broader peak remains, with a decrease in magnitude. This suggests that the signal is dominated by the stem signal. However, in position 2, the amount of irradiated optical fiber is almost halved compared with position 1, which is not reflected in the response curves. Again, more investigation, including spectrometer measurements, are needed to determine the impact of the stem signal.

The choice of sampling frequency is crucial to maximize the SNR, while keeping a sufficiently high temporal resolution. The signal is highest for the highest sampling frequency, despite being associated to the lowest integration window. The fact that the shortest integration window should result in the lowest signal, assuming that the full pulse is delivered within a single integration window, indicates that a gain factor is applied to compensate for the differences in the integration window. Indeed, when investigating the noise levels, a discrete set of values is observed, which are proportional to the integration window. Therefore, the signal for the highest sampling rate is artificially increased. Also, high sampling frequencies rapidly fill the buffer, allowing only short recordings. This can be solved by triggered acquisition. As a trade-off, it was opted to use a sampling frequency of 1000 Hz for the YAG and Organo-Br FODs since this resulted in a smooth signal and sufficient temporal resolution to discriminate pulses, even at the highest PRF that can be delivered by the ElectronFlash.

4.1. Application in UHDR

4.1.1. Plastic FOD

The Hyperscint RP-200 showed to be the best-performing FOD for this UHDR electron beam. Deviations from the reference detector were within 5.25% for all measurements where the amount of fiber in the beam, and hence the proportion of the stem signal, was approximately equal (experiments 1–3, 5 and 6). This includes the measurement with 0.5 µs nominal pulse length, which results in an effective pulse length (FWHM) of 0.278 ± 0.022 µs. The output of such short pulses is unstable and results in a relatively low dose per pulse in terms of UHDR irradiations. Also included was the measurement for a PRF of 1 Hz. The used sampling frequency and integration window of 1 Hz and 1 s are, respectively, too low and too long for an accurate measurement of the dose. This low PRF results in a relatively low average dose rate. Both beam parameter settings are not expected to be relevant in FLASH-RT. When excluding these data points, the deviations from the reference detector remained within 3%, which is currently acceptable for UHDR dosemeters under investigation. This is in agreement with the findings of Baikalov et al. [16].

The response of the Hyperscint RP-200 appears to drift, as the irradiations were always performed in increasing magnitude of the varying beam parameter, and a growing underestimation of the dose was observed. Since this underestimation resets in every new experiment, it is not expected to be due to (non-temporary) radiation damage.

It should be noted that these results were obtained with separate dosimetric calibrations for the UHDR and conventional modalities. When using the conventional calibration in a UHDR beam, deviations of 19% were observed. Also, in experiment 4, where the SSD, and thus iDR, was varied, the Hyperscint RP-200 showed larger underestimation of the dose for higher SSD (lower iDR). This is, however, more likely due to the increased portion of the fiber that is exposed to the beam. The need for modality-specific dosimetric calibration and the larger deviations at higher SSD might be caused by an iDR dependence, which can be overcome by re-calibration or correction factors.

4.1.2. Inorganic FOD

The YAG FOD is not ready for use in UHDR experiments. Large discrepancies (>10%) with the reference detector were observed in all experiments. These discrepancies grew bigger with increasing magnitude of the varying beam parameter. An attempt was made to minimize the contribution of drift in the signal when investigating the YAG FOD. This was performed by repeating the sequence of a beam parameter variation rather than repeating a single setting before starting the next one (i.e., [X, Y, Z, X, Y, Z, X, Y, Z] iso [X, X, X, Y, Y, Y, Z, Z, Z], with X, Y and Z different beam parameter settings). Regardless, the response increased with increasing magnitude of the varying beam parameter. Moreover, the increasing discrepancy appeared cumulative over the different experiments, with the exception of experiment 4, where also different portions of the fiber were being irradiated. Because of the cumulative nature of the discrepancy, it is plausible that it was caused by radiation damage. Similar to the Hyperscint RP-200, the YAG FOD experiment 4 showed a decreasing response with increasing SSD (and decreasing iDR), which is opposite to the other experiments. In experiment 7, the pulse discrimination properties of the YAG FOD were shown. Despite being able to distinguish individual pulses at the maximum PRF, a long decay time was observed, resulting in increasing peak response, with the signal not returning to the background between pulses. The findings for this YAG FOD are unexpected, as the material used as scintillating head showed negligible dose and dose rate dependence when used in a 2D coating [30]. In that work, the ImageDosis system was rigorously tested for real-time 2D UHDR dosimetry across two separate measurement campaigns, using two different research ElectronFlash linacs from S.I.T., which facilitated the systematic variation in the pulsed electron beam parameters, including dose rate modality (both conventional and UHDR), energy levels, PRF, number of pulses, and pulse duration. The ImageDosis system is an in-house developed system, consisting of scintillating coatings (varied according to radiotherapy application) and a scientific camera [23,31]. The major difference between the current work and the one using the ImageDosis system is the readout process. The stem effect is a common challenge in real-time dosemeters that use signal carriers to transmit dose-response signals from sensitive volumes to the readout system. Upon irradiation, the signal carrier itself can generate a measurable signal, which interferes with and offsets the dose-response signal produced by the dosemeter’s sensitive volume. These perturbations are typically small for energies and dose rates used in conventional radiotherapy but probably play an important role for UHDR. A recent work using a plastic scintillator in electron UHDR observed a Cherenkov contribution around 10–11%, before corrections [17]. When looking at 250 MeV proton UHDR beams, the stem effect from fiber-coupled ZnSe:O inorganic scintillators contributed up to 0.2% to the total dose, which is considered negligible [32].

4.1.3. Organic–Inorganic Hybrid FOD

The Organo-Br FOD was previously presented in [13] (PS06), where a lower sampling rate (200 Hz) and larger integration window (5 ms) were used. The hardware of this system has been updated with a new data card allowing faster sampling. Similar results have been obtained with linearity with the number of pulses >0.999 and deviation from linearity with increasing pulse length. The PRF data fluctuated around 0% difference from the reference in both studies, where larger deviations are observed in this study. However, for both studies, the error bars (1 sigma) are large and overlap. New to this study is the response with varying SSD. The large increasing deviation from the reference should be further investigated but is in line with the fact that different calibrations were needed for UHDR and conventional dose rates by Vanreusel et al. [13]. There, also a higher response per unit of dose was observed for lower iDR. Using the data from experiment 4, the effect of the iDR becomes extremely relevant starting from an SSD of 136 cm, in which an iDR of 571.8·${10}^3$± 6.8·${10}^3$ Gy/s was reached. The increase in response per unit of dose suggest that the Organo-Br FOD saturates at a certain iDR. Since the same readout system is used as for the YAG FOD, for which more counts were detected for the same iDR, it is most probable that the saturation does not originate in the readout system. More investigation is needed and ongoing to understand what is causing this effect. Despite the poor results in experiment 1–6, the Organo-Br FOD showed the best performance for pulse discrimination in experiment 7. Even for the maximum PRF, individual pulses could be distinguished, with the signal returning to baseline between pulses.

4.2. Future Work

More accurate stem signal removal is the first requirement for the YAG and Organo-Br FODs. A spectral analysis is needed to improve the choice of the wavelength filter and reduce the FWHM of passing wavelengths. In addition, other fiber materials should be investigated to minimize the spectral overlap between the scintillator and stem signals. A twin fiber approach could also be considered but requires an update of the hardware and software since a double recording is needed. Alternatively and preferentially, a spectrum-based calibration as used in the Hyperscint RP-200 could be implemented for better stem signal characterization and removal. However, this approach also requires a hardware and software update.

The YAG FOD has the highest light yield, which is beneficial for the SNR but might limit the dynamic range of this FOD. It first requires further investigation and removal of the apparent trend, after which the dynamic range needs to be determined. The Hyperscint RP-200 and Organo-Br, on the other hand, have lower light yield but are more tissue equivalent. The Hyperscint RP-200 showed to be in best agreement with the flashDiamond. However, further investigation is needed to (1) validate its potential drift, by reversing the order of acquisition and determining the origin of this behavior, (2) reduce its uncertainty, and (3) validate the dosimetric calibration with varying instantaneous dose rate, especially when varying between conventional and UHDR modality. The hardware of the Organo-Br and YAG FODs needs thorough investigation to reduce the uncertainties on the measurements and allow systematic investigation of the different parameters. The Hyperscint RP-200 and YAG prototype FODs require more investigation on their increasing dose underestimation for higher SSD (lower iDR) by utilizing additional collimation to keep the amount of exposed fiber constant, and for the Hyperscint RP-200, by validating the spectral decomposition.

5. Conclusions

FLASH radiotherapy has the potential to revolutionize the field of radiotherapy. The requirements of dosemeters for such high dose rate beams become more stringent, especially regarding their temporal features. Moreover, the call for in vivo dosimetry for patient QA and research on the radiobiological mechanisms underlying the FLASH effect grows stronger. Point scintillating dosemeters offer themselves as good candidates due to their high temporal resolution, theoretical dose rate independence and high spatial resolution with possibilities for in vivo use. The YAG and Organo-Br prototypes tested in this study were first optimized in terms of optical filtering and choice of sampling frequency. Unexpected results were obtained, where longer pulse lengths resulted in less detected scintillation light for the Organo-Br FOD. Band-pass wavelength filters centered at wavelengths of 500 nm and 425 nm were respectively chosen for the YAG and Organo-Br FOD. A spectral investigation was needed for these prototypes to further determine and minimize the stem signal. A trade-off was made in the choice of sampling frequency, taking into account the buffer size and pulse discrimination, resulting in a sampling frequency of 1000 Hz with an integration window of 1 ms. The Hyperscint RP-200 was the only prototype that proved useful for UHDR electron beams when investigating its dose, dose rate, dose per pulse, instantaneous dose rate, and energy dependence. It showed to be stable within 3% for all relevant dose rate settings within the FLASH-RT range. A drift was observed in the dose response, which requires more investigation, as it can further reduce the stability. The YAG and Organo-Br FOD both proved unfeasible for UHDR measurements. The YAG FOD appeared to be affected by a strong drift, which should be excluded. The Organo-Br FOD showed less dependency with the relevant beam parameters but had uncertainties of >10%, which is unacceptable. The benefits of the YAG and Organo-Br prototypes are their temporal resolution and pulse discrimination. Therefore, it is worthwhile to further investigate them or search for valid alternatives.