Time‐resolved point dosimetry for spread‐out Bragg‐peak proton FLASH using fibre‐coupled scintillators Authors: Sky R. Steenholdt, Jacob G. Johansen, Eleni Kanouta, Per R. Poulsen Published: 2026 ABSTRACT AbstractBackgroundIn FLASH radiotherapy, the dose is delivered using ultra‐high dose rates (UHDR), which are approximately 100 times higher than those used for conventional (CONV) treatments. This has shown promise in sparing normal tissue while maintaining tumour control. Proton beams, particularly in the spread‐out Bragg peak (SOBP), offer a favourable depth–dose profile for sparing healthy tissue. However, quality assurance for FLASH requires time‐resolved dosimetry to capture the temporal structure of pencil beam scanning delivery. Fibre‐coupled scintillating detectors have been used both in electron and photon UHDR beams, and have been applied in the proton beam entrance plateau. Extending their use to the SOBP requires careful calibration to address quenching and water‐inequivalent response near the Bragg peak.PurposeTo calibrate and validate a fibre‐coupled inorganic scintillator detector system for accurate, time‐resolved point dosimetry in the SOBP for UHDR proton beams, which will enable preclinical and in‐vivo FLASH studies with robust dosimetric and geometric verification.MethodsExperiments were conducted using a clinical proton PBS beam line. A 2D range modulator generated a 5 cm SOBP from a mono‐energetic beam. Four ZnSe:O scintillator probes coupled to optical fibres were read out by silicon photomultipliers at 50 kHz. An ionisation chamber provided reference dose measurements.The calibration included determining a signal dependent saturation factor of the silicon photomultiplier ksat, measuring the absolute calibration factor ND,w,15, and characterising the correction for the depth‐dependent under‐response kQ.A calibration validation was performed in the SOBP across a range of UHDR beam currents, evaluating both dosimetric accuracy and probe positional stability. The calibrated system was then used to characterise SOBP beam spot profiles, in terms of full width at half‐maximum and dose rate variation with depth.ResultsA saturation multiplier of up to 55% was observed across all four probes. The depth‐dependent under‐response reached up to 12% at the distal SOBP edge. Both effects were successfully corrected for through fitting simple functions. The validation in the SOBP demonstrated that the calibration achieved positional stability within ±0.1 mm and agreement between the measured and absolute doses within 0.5% for all probes. Beam characterisation revealed full‐width at half‐maximum broadening from 8.3 mm at shallow depth to 21.5 mm near the range end, with spot profiles comprising two Gaussian cores and a Lorentzian tail. The maximum instantaneous dose rate in the UHDR beam fell from 800 Gy/s in the entrance plateau to 280 Gy/s in the SOBP.ConclusionsThe developed calibration method enables accurate, time‐resolved dosimetry in UHDR proton SOBP beams, allowing for the separation of saturation and quenching corrections. The fibre‐coupled scintillator system demonstrated high precision in both dose and geometry, making it suitable for quality assurance in preclinical FLASH studies. This approach streamlines recalibration, reducing beam time requirements, and supports routine monitoring of PBS‐delivered proton FLASH treatments in complex depth–dose scenarios. INTRODUCTION Radiation therapy at ultra‐high dose rates (UHDR), about 100 times higher than conventional (CONV) levels, has shown a promising capacity to spare normal tissue 1 , 2 , 3 , 4 , 5 , 6 , 7 , 8 , 9 while preserving tumour control. 4 , 6 , 9 This biological phenomenon is called the FLASH effect and has been observed using electrons, 1 , 2 , 3 , 10 , 11 photons, 12 , 13 , 14 and protons. 5 , 7 , 15 , 16 To date, in‐vivo proton FLASH studies have been conducted in both the entrance plateau 5 , 8 and the spread‐out Bragg peak (SOBP) regions, 16 , 17 with the latter leveraging the favourable depth–dose distribution provided by the Bragg peak. 18 In clinical practice, an SOBP is achieved by superimposing multiple energy layers. However, changing the energy of the beam is time‐intensive and reduces the mean dose rate of the field. 19 Therefore, to maintain UHDR, a static 2D range modulator (2DRM) (ridge filter) can be placed in the beam to obtain an SOBP from a mono‐energetic proton beam. 20 In most cases, proton beams are delivered using pencil beam scanning (PBS), where each point in the field is exposed to variable instantaneous dose rates while the beam scans over the field. This means that while the mean field dose rate may be a good parameter to describe FLASH with pulsed electron beams, 21 more detailed metrics, such as the PBS dose rate, 22 may be needed to describe the beam structure more accurately in proton therapy. Measuring the dose rate is essential to fully investigate the defining dose rate quantities. Therefore, highly time‐resolved quality assurance (QA) is necessary to verify the correct dose and temporal structure of PBS delivery. So far, QA in UHDR has mostly relied on dose‐integrating detectors, such as ionisation chambers 7 , 15 or alanine pellets. 1 , 5 However, for time‐resolved measurements, fibre‐coupled scintillating crystals have proven effective for electrons 23 , 24 and photons. 25 , 26 , 27 We extended the in‐house developed fibre‐coupled scintillator system from entrance‐plateau measurements 28 , 29 to depth‐resolved operation in a passively generated SOBP. Because high ionisation density near the Bragg peak reduces scintillation yield, we performed a dedicated depth calibration to correct for quenching. 30 , 31 This study aims to calibrate and validate a fibre‐coupled scintillator detector system for future time‐resolved in vivo point dosimetry to accurately characterize proton SOBP FLASH beams. MATERIALS AND METHODS Dosimeters This study used a detector system (Figure 1) previously characterised and used in murine proton FLASH studies in the entrance plateau. 5 , 6 , 28 , 29 , 32 It consisted of four probes (D1, D2, D3, and D4) of inorganic scintillating crystals (ZnSe:O) with a 0.5 x 0.5 mm2 base. D1 was 1 mm long for increased sensitivity, as this probe was planned to be placed in a lower dose region in the in‐vivo experiments and D2, D3, D4 were 0.5 mm long. This scintillator material was chosen due to high scintillation intensity, 33 , 34 negligible stem effect, 28 radiation hardness and a short decay time of τ= 6μs compared to beam spot durations of at least 3 ms. A schematic drawing of a single detector probe of the system. An inorganic scintillating crystal (ZnSe:O) was connected to an optical fibre, which leads the light to a light tight box. There, the light intensity was measured by a Si photomultiplier (SiPM) and read out with a digitiser at 50 kHz. Each crystal was coupled to a >10 m polymethyl methacrylate (PMMA) optical fibre (Ø= 0.5 mm), directing the light to an optically sealed box with a silicon photomultiplier (SiPM) (MicroFC‐SMTPA‐60035, SensL), which converted the light to an electrical potential. The SiPMs were placed 15 mm from the fibre end and were operated with a gain of 27 V (2 V overvoltage). The signals from all four SiPMs were read out using a multi‐channel digitiser (USB‐1808, Measurement Computing) at a 50 kHz sample rate, allowing for sub‐spot‐duration temporal resolution in both CONV and UHDR beams. Pictures of the system can be found in a previous publication. 29 An ionisation chamber (IC) (Advanced Markus Chamber, TW34045, PTW) was used as a reference dosimeter for the calibration to absolute doses and validation of the calibrated scintillating dosimeters. The IC had been cross‐calibrated against a ROOS chamber, which could not directly be used for UHDR measurements due to saturation. 35 The saturation of the IC was corrected via the recombination correction factor, which was determined at all depths separately for UHDR beams and ranged from 1.001 to 1.002. Radiochromic films (EBT‐XD, Ashland) were used to determine the dose homogeneity at different depths, before commencing calibration experiments. Beam All measurements and irradiations were performed at the Danish Centre for Particle Therapy (DCPT), Aarhus, Denmark. A clinical proton beam facility (ProBeam 4.3, Varian) produced a horizontal monoenergetic scanning pencil beam. Two types of beams were used for the experiments: A 244 MeV proton beam with a beam current ranging from 1.1 to 4.5 nA for CONV irradiations and a 250 MeV beam with requested beam currents ranging from 5 to 215 nA for UHDR irradiations. A 5cm SOBP was generated by using a 2DRM, 20 , 36 provided by GSI Helmholtzzentrum für Schwerionenforschung GmbH (GSI), creating a multienergetic beam from the monoenergetic beam line. The 2DRM was fabricated from 3D‐printed polypropylene, and consisted of a 1 mm base plate with a grid of 44 mm tall pins, each with a 3x3 mm2 base plane. All measurements were performed using a 60 mm by 60 mm uniform calibration field created by 31 by 31 spot positions with 2 mm spacing in the isocentre plane (Figure 2a) This ensured a uniform dose distribution at the detector positions. However, the 2DRM used to create the SOBP caused an inhomogeneous dose distribution immediately after the range modulator, 36 which was smoothed out with depth. (a) The irradiation field that was used for all calibration measurements, showing the positions of the detector probes (stars) and the Advanced Markus ionisation chamber (IC) (red circle), where x is in horizontal and y in vertical direction. (b) Setup with a 2D range modulator (2DRM) in front of the water bath. The isocenter is behind the 2DRM at z = 0. The detector probes are attached to the cap of the IC. Underneath, a sketch of the depth dose curve. Calibration notation The calibration notation in this study follows the notation of TRS‐398 37 and the previous calibration of this system in transmission beams. 28 The terminology has been modified to fit the case of SOBP proton beams in water measured by water‐inequivalent scintillators. The dose rate to water can be expressed as: (1) D˙w,Q,spot(p⃗)=MQ,spot(p⃗)·ND,w,z0·ksat(MQ,spot)·kQ(R−z) Where D˙w,Q,spot(p⃗) is the dose rate in [Gy/s] to a point p⃗=(x,y,z) in water for a given beam Quality (Q) from one spot with beam spot centre at (xspot,yspot) at a given depth z. MQ,spot(p⃗) is the signal in [V] measured by a probe at p⃗ from a spot delivered to (xspot,yspot). ND,w,z0 is a probe‐specific calibration factor at a reference depth z0 that converts the signal [V] to dose rate to water [Gy/s]. ksat(MQ,spot) is the correction factor for saturation of the detector at intense beams. For the scintillator‐based detector, this is from the saturation of the SiPMs. The SiPMs consists of 18 980 pixels working in Geiger‐mode, giving a binary signal when one or more photon hits the pixels. The total signal from the SiPM is proportional to the number of pixels firing and not the number of photons hitting. The ratio between the number of pixels firing and the number of photons hitting the detector will degrade with increased beam intensity, as more pixels will be hit by multiple photons, which cannot be distinguished from a single photon hit. therefore ksat depends on the amplitude of the signal MQ,spot. The factor kQ(R−z) corrects for depth‐dependent under‐response observed in scintillating crystals in proton beams. 31 , 38 This under‐response primarily stems from internal quenching effects and the water inequality of the scintillating crystals. 39 Quenching describes the reduction in scintillation signal in regions of high linear energy transfer (LET). As protons slow down towards the end of their track, they create a high ionisation density. In such regions, the scintillating material's efficiency in converting deposited energy into photons is diminished, leading to a suppressed signal. This effect might be amplified by UHDR, 40 which will be investigated in this study. The water inequality of the crystal further impacts the signal because the stopping power of the proton beam is strongly dependent on the effective charge and density of the material it passes through. Hence, the absorbed dose in ZnSe:O crystals will be much higher than the absorbed dose in water. It is necessary to correct for this discrepancy, in order to relate the measured dose to the absorbed dose to water. Both quenching and water‐inequivalent effects intensify towards the distal end of the proton range. kQ is independent of the initial proton energy if defined in terms of the residual range (R−z), where R is the depth where the dose has decreased to 10% at the distal edge of the SOBP 37 (Figure 2b). Dose homogeneity The homogeneity of the calibration field after the 2DRM was investigated with film dosimetry prior to the calibration measurements. The aim was to find a reference depth z0, where ND,w,z0 could be determined, which requires a homogeneous field with the same dose to the detector probes and the IC, since the detector probes are point dosimeters, while the IC measures dose over an area. The 2DRM was placed in front of slabs of solid water. Calibrated radiochromic films were placed 10, 15, 20, and 25 cm downstream from the 2DRM and irradiated in two sets at each depth with the calibration field (Figure 2a) using 200 monitor units (MU) per spot. The Irradiation was performed once with UHDR with a requested beam current of 35 nA and once with a CONV dose rate with a beam current of 4.5 nA. The films were read out with a flatbed scanner 24 h after irradiation and doses were calculated using the green channel of the scanned images. The dose distribution at 10 cm downstream from the 2DRM showed rows of hot and cold spots, (Figure 3a and b), with the dose distribution showing a standard deviation of 4% within a 2 cm by 2 cm region at the centre of the field. The depth to determine ND,w,z0 was therefore chosen as z0= 15 cm. Here, the hot and cold spots had dispersed and the dose distribution was homogeneous with a standard deviation of only 0.18 Gy within the centre of the field, which translates to 0.9% of the mean measured dose (Figure 3c and d). (a) Film dosimetry for UHDR at depth 10 cm. The dose distribution is inhomogeneous due to the shape of the 2DRM and alternately shows rows of hot and cold spots. A red line extends across the maximum dose, while a green line crosses the local minimum. (b) UHDR Dose profiles along the x axis for y=−1.4 mm (green) and y=0.4 mm (red) at 10 cm depth. (c) Film dosimetry for UHDR at depth 15 cm. The distribution is now homogeneous (d) UHDR Dose profiles along the x axis for depth 15 cm. Calibration procedure Measurements The set‐up for the calibration measurements (Figure 2b) consisted of the 2DRM placed in front of a water bath. The probes were attached to the cap of the IC, which in turn was fixed on a motorised moving stage. This allowed for the simultaneous measurement of the dose using the detector probes and the IC. The effective measurement depth of the IC was 2.4 mm behind the probes in depth, which was corrected for in the analysis using interpolation. The irradiated calibration field had start/end spots on the diagonal, close to the middle of the field (Figure 2a). These spots were added to ensure a clear start and end signal in the measured data (Figure 4). (a) Raw detector signal in V for the entire delivered field for detector probe D3 (Beam current: 215 nA, Energy: 250 MeV, z = 34 cm). The first and last spots on the diagonal of the field mark start and end of the delivery in the signal (arrows). One row of spots is highlighted. (b) is the raw detector signal zoomed in on the highlighted row of spots. Additionally, each of the spots have been assigned a mean value MQ,spot represented by a horizontal line across the respective signal. ksat for each probe was determined at z=10cm despite the inhomogeneous dose distribution at this shallow depth. ksat was, therefore, determined from the largest possible range of instantaneous dose rates, as is required for an accurate calibration. This was done in two measurement series: The first (Series 1a, Table 1) used UHDR, where the beam current was steadily increased from 5 to 215 nA in 10 nA steps and the MU per spot adjusted accordingly from 100 to 800 MU to satisfy machine requirements of a minimum 3 ms spot duration. To extend the saturation curve to CONV dose rates, a second set of measurements was taken where CONV beams were delivered with 3 different currents of 1.1, 2.9 and 4.5 nA with 100 MU per spot (Series 1b). TABLE 1Summary of all the measurements taken as part of this study.Meas. SeriesNo of irrad.Depth [cm]Beam current [nA]Beam Energy [MeV]MU/spotQuantity determined1a21105–215250100–800 ksat 1b3101.1–4.5244100 ksat 2115200250800 ND,w,z0 32010–38200250800 kQ 42010–36.54.5244100 kQ 521345–215250100–800SOBPvalidationJohn Wiley & Sons, Ltd. Summary of all the measurements taken as part of this study. The calibration factor ND,w,z0 was measured at z0=15 cm with an UHDR beam at 200 nA requested beam current and 800 MU per spot (Series 2). The depth‐dependent response correction kQ(R−z) was mapped across the entire depth‐dose curve using a step size of 5 cm in depth in the entrance region of the beam and 0.5 cm around the SOBP. For UHDR the requested beam current was kept constant at 200 nA with 800 MU per spot (Series 3) and for CONV the current was 4.5 nA with 100 MU per spot (Series 4). Finally, a set of irradiations was performed with the detectors at z=34 cm (Series 5). This irradiation was used for validation of the calibration. The requested UHDR beam current was steadily increased from 5 to 215 nA in 10 nA steps, and the MU per spot were adjusted from 100 to 800 MU per spot. Spot signal and probe position The depth‐dependent lateral position of the spot (xspot,yspot) was determined from the planned coordinates in the isocentre plane (xplan,yplan) corrected for beam divergence at the measurement depth z. (2) xspot(z)=xplan·200cm+z200cm (3) yspot(z)=yplan·256cm+z256cm The start and end times of each spot were determined based on the marked start and end time of the irradiation (Figure 4a) and the spot duration from machine log files. 29 From this, the mean signal per spot MQ,spot(p⃗) was determined for each spot and each of the four probes, taking advantage of the sub‐spot‐duration resolution of the detector system (Figure 4b). The MQ,spot of a given probe were used to determine each probe's lateral position (x,y) in the field: a Gaussian distribution G (Equation 4) was fitted to the signal amplitude MQ,spot at the planned spot positions (xspot,yspot). A,x,y and σ were free fitting parameters, representing the maximum signal amplitude [V], the probe's x and y position [mm] and the approximate spot size [mm]. (4) G(p⃗)=A·e−rspot(p⃗)22σ2 (5) rspot(p⃗)=(x(z)−xspot(z))2+(y(z)−yspot(z))2 Determination of ksat For a measured signal MQ,spot, ksat(MQ,spot) is the ratio between a hypothetical measurement without saturation of the SiPM and the actual measurement. To determine the hypothetical unsaturated measurement, the spot dose profile close to a probe, that is, when rspot≤σ, was assumed to be well described by the single Gaussian distribution in Equation 4. It gives the following hypothetical unsaturated measurement at close distances rspot from a probe: (6) Munsaturated(p⃗)=aMUspottlog,spot·e−rspot(p⃗)22σ2 Here, MUspot is the number of MU in the spot, tlog, spot is the logged irradiation time of the spot, and a is a constant that transforms dose rate (in the unit of MU/s multiplied by the Gaussian profile) to the SiPM signal measured in V. For a given spot, rspot(p⃗) and σ were both known from the Gaussian fit in Equation 4, where the lowest beam current (5 nA, Series 1a, Table 1) was used for the UHDR beam to avoid effects of SiPM saturation. Similarly, the constant a was found from the lowest beam current measurements (5 nA) by fitting of Equation 6. The assumption, that the actually measured signal MQ,spot for this current was unsaturated and thus equal to Munsaturated in Equation 6 (Figure 5a), was verified using lower CONV currents (1.1–4.5 nA, Series 1b, Table 1) (Figure5b). (a) Dotted: The measured signal MQ,spot for all detector probes plotted against Munsaturated/a in Equation 6. The linear fits (solid lines) represent the hypothetical unsaturated measurements Munsaturated with slope a (b) Zoom of the measured signal MQ,spot for 5nA (dotted) and 1.1nA (crosses) with the linear fit (solid lines). After determination of the constant a, ksat was calculated for all spots with rspot≤σ in Series 1 as the ratio between the hypothetical unsaturated measurement (Equation 6) and the actual saturated signal MQ,spot: (7) ksat(MQ,spot)=MunsaturatedMQ,spot The resulting ksat values were fitted to a common third‐degree polynomial across all four probes, which spans all dose rates and can now be applied to all spots in the field to correct for the saturation. Determination of ND,w,15 ND,w,15 was determined for each probe by measurement at reference depth z0=15 cm (Series 2, Table 1). The effective measurement depth of the IC was accounted for by using interpolation. ND,w,15 was then calculated as the ratio between the total IC dose and the total integral signal after correction for saturation with ksat (Equation 8) (8) ND,w,15=D∑spotMQ,spot(x,y)·ksat(MQ,spot)·tlog,spot Combined with ksat, the calibration factor allows for determining the instantaneous dose rate D˙Gys directly from the probe signal MQ,spot in the case of no quenching. Determination of kQ To evaluate the depth‐dependent correction factor for under‐response kQ, the total measured dose was determined with Equation 1 assuming a kQ=1 at various depths from 10 to 38 cm. This reduced measured dose was then compared to the absolute dose D measured with the IC (Figure 6), where a smoothing spline was used to interpolate between the measurement depths to take into account the difference in depth between the detector probes and the IC. The under‐response of the scintillating crystals as a function of depth was characterised by a reduction in the measured dose relative to the absolute dose (Figure 6). A large spread was seen at 10 cm depth, due to inhomogeneous dose distribution. Towards the end of the SOBP, the steep dose gradient leads to high sensitivity to positioning uncertainties, which also causes an increased spread in the scintillator measured dose. These points were therefore omitted for the characterisation of quenching, due to the previously described inhomogeneous dose distribution. Accordingly, the correction factor was calculated: (9) kQ,meas(z)=D(z)DGys(z) The absolute dose as measured by the IC over the depth (black squares) with a smoothing spline function (black line) and the reduced Dose measured by the probes (crosses) for (a) UHDR and (b) CONV. To achieve a general function across all probes, the correction factors from the four probes were fitted with a function (Equation 10) inspired by Bortfeld's Bragg curve equation 41 and Birk's law. 31 (10) kQ(R−z)=F·e−b(R−z)(R−z)c F, b and c are fitting parameters and the term (R−z)c describes stopping power simplified from the Bethe–Bloch equation, the exponential decay factor e−b(R−z) describes range straggling and F scales the result to the data. The range R of the beam was determined from a smoothing spline fit to the IC dose. This analysis was conducted for both UHDR (Series 3, Table 1) and CONV beams (Series 4). Altogether, this resulted in a calibration (Equation 1) that can determine D˙w,Q,spot(p⃗) for all depths z, accounting for depth‐dependent effects. Calibration validation in the SOBP For validation across the SOBP, the completed calibration was first applied to Series 3 and 4 (Table 1). Then, the calibration was applied to the validation measurements taken in the SOBP with UHDR (Series 5) with the purpose of investigating both dosimetric and geometric accuracy. After integrating the signal over all spots, the measured total dose Dw,Q(p⃗) was compared to the absolute dose D measured with the IC. For the geometric accuracy, the position (x,y) of the detector probes in each measurement was calculated using Equation 4. Since these positions were not changed between measurements, any variation of position was either due to variation in the spot positions delivered by the machine, or measurement uncertainty. Beam properties The calibrated and verified system was applied to measure the properties of the beam in two steps. First, the exact spot dose profile, including the shape of the tail, was determined both at z=10 cm (Series 1a, Table 1) and at z=34 cm (Series 5). Each of the two series consisted of multiple measurements with varying dose rates. To account for this, the measured dose rates from each spot D˙w,Q,spot were normalised and plotted against rspot(p⃗), determined using (Equation 5). The dose profile was then fitted with a distribution function S(rspot,p⃗) consisting of two Gaussians and one Lorentzian term (Equation 11). (11) S(rspot,p⃗)=A1·e−rspot(p⃗)22σ12+A2·e−rspot(p⃗)22σ22+1π12Γrspot(p⃗)2+12Γ2 Secondly, the amplitude of the spot dose profile Dmax and the full‐width half‐maximum (FWHM) were determined as a function of depth for both UHDR (Series 3) and CONV (Series 4). MQ,spot(p⃗) was determined and translated into D˙w,Q,spot(p⃗) per the determined calibration. The maximum of the dose rate was integrated over the irradiation time per spot and divided by MU (Equation 12). This compensates for output variations between measurements. The FWHM was determined as the full width of the fitted function S(rspot,p⃗) at half the amplitude. (12) Dmax(z)=max(D˙w,Q,spot(p⃗)·tlog,spot)MUspot RESULTS Calibration The non‐linear response of the SiPMs resulted in significant signal attenuation, reaching up to 55%, particularly evident at elevated instantaneous dose rates. The saturation correction factor ksat(MQ,spot) was consistent across all four probes (Figure 7a) and thus modelled as a single function using a 3rd‐degree polynomial with a root mean square error (RMSE) of 0.0269: (13) ksat(MQ,spot)=1+0.7874V−1MQ,spot+2.556V−2MQ,spot2−0.189V−3MQ,spot3 The calculated slope a of the unsaturated reference signal for each detector probe can be found in Table 2 along with the calibration factor ND,w,15. TABLE 2Calibration properties: a is the hypothetical unsaturated SiPM signal and ND,w,15 is the calibration factor with uncertaintyProbe no. aμVsMU ND,w,15[Gy/Vs] 15.92717±3523.301359±1034.42946±943.691230±16John Wiley & Sons, Ltd. Calibration properties: a is the hypothetical unsaturated SiPM signal and ND,w,15 is the calibration factor with uncertainty (a) ksat data from all probes (red, green, blue, yellow) together with the fitted third degree polynomial (black) (b) kQ(R−z) plotted as a function of the distance from the end of the range (R−z) for UHDR (red) and CONV (blue). Each point represents the mean and 1SD across all four probes and the line represents the fitted function (Equation 10). The correction for the depth‐dependent under‐response kQ(R−z) was measured with all probes in the range from z=10 cm to z=38 cm, covering the range of the SOBP. The effect was observed as an increased signal loss with depth of up to 12%. kQ(R−z) was fitted for both UHDR (RMSE = 0.0221) and CONV dose rates (RMSE = 0.0176), and the two functions were found to follow a similar shape. However, there was an increase of up to 2% in kQ for UHDR compared to CONV. (Figure 7b). Calibration validation As a first test, the completed calibration was applied to the data from Series 3 and 4 (Table 1, Figure 8a and b). Across all 20 measurement depths, the system measured a dose that deviated from the IC dose D by −1.3%±4.3% (μ±1SD) with larger deviation towards the distal end of the proton range. Dosimetric validation: (a) The measured dose at UHDR (Series 3, Table 1) for all probes (red) after calibration compared to the absolute dose measured with the ionisation chamber (IC) (black). The proton range R is marked at 38.25 cm. (b) The measured dose (D1–D4) at CONV (Series 4) for all probes (blue) after calibration compared to the absolute dose (black). The proton range R is marked at 36.75 cm. (c) The measured dose at z=34 cm (Series 5) from the calibrated system compared to the absolute dose for a range of doses from 10 to 100 Gy with UHDR. A unity line (grey) shows where the two are equal. (d) The measured dose in per cent relative to the absolute doses for the same measurements. Then, the calibration was applied to the validation dataset Series 5 (Table 1). The eight different doses correspond to the number of MU per spot increasing from 100 to 800 MU in steps of 100 MU (Figure 8c). Across all measurements, the probe measured dose increased linearly with the IC measured dose. The relative deviation between integrated probe doses and IC measured dose D (Figure 8d) was 0.2±0.5% (μ±1SD). The standard deviation of the measured probe positions across 21 measurements was ≤ 0.05 mm in horizontal and ≤ 0.1 mm in vertical direction (Table 3, Figure 9). The measured values MQ,spot(p⃗) for repeated measurements were very stable with a RMSE of 7×10e‐4 V. TABLE 3Measured horizontal and vertical positions of the detector probes and the relative mean dose for validation, all with standard deviation.Probe no.Horizontal position [mm]Vertical position [mm]Relative mean Dose %1−5.28 ± 0.040.36 ± 0.0999.8 ± 0.42−3.25 ± 0.05−2.9 ± 0.0999.7± 0.534.53 ± 0.04−4.3 ± 0.0899.8 ± 0.246.31 ± 0.031.09 ± 0.1099.9 ± 0.4John Wiley & Sons, Ltd. Measured horizontal and vertical positions of the detector probes and the relative mean dose for validation, all with standard deviation. Geometric validation (a) The spot positions of the field (black) at z=34 cm overlayed with the measured positions of the detector probes for all measurements in the validation set. (b) a zoom of the plot to show the variation in determined probe positions. Beam properties Using the fully calibrated system, the detailed spot dose profile was determined both in the entrance plateau and in the SOBP with UHDR beams (Figure 10a and c), where measurements with multiple beam currents were conducted (Series 1 and 5, Table 1). The data from all probes were used simultaneously, as the alignment indicated they were well calibrated relative to each other. At z=10 cm, the spot dose profile was found to be well described by two Gaussian distributions with σ1=3.4 mm and σ2=6.3 mm and a Lorentzian with Γ=19.0 mm. The spot's FWHM was 8.3 mm. Detailed spot dose profile for UHDR at (a) z=10 cm and (c) z=34 cm on a logarithmic scale with the normalised data from all probes in black, two Gaussians in red and green respectively, the Lorentzian distribution in blueand the total fitted function in yellow(b) Detailed spot dose profile for UHDR at z=34 cm fit with a single Gaussian (red) (d) the residuals at z=34 cm relative to the value of the fitted data for the combined function (top) and a single Gaussian (σ=8.05 mm) (bottom). The same fitting procedure was repeated in the middle of the SOBP at z=34 cm. Here, the spot width had increased to be described with σ1=7.6 mm, σ2=11.4 mm, and Γ=18.7 mm. FWHM was 18.8 mm. The residuals of the combined function and a single Gaussian fit (Figure 10b) were analysed (Figure 10d). It was found that the single Gaussian only described the centre of the spot dose profile, within ∼15 mm, whereafter the fitted function fell off too steeply, leading to the residuals to increase towards 100%. The combined function S(rspot,p)⃗ produced residuals centred around 0% and was, therefore, better suited to describe the spot dose profile. The larger residuals further away from the spot centre can be explained by the decreased intensity of the signal. The FWHM of the pencil beam was determined for all depths (Figure 11a). At UHDR the FWHM increased from 8.28 mm at z=10 cm up to 21.5 mm at z=37 cm right before the end of the SOBP. Toward the end of the range, the width of the beam dropped slightly. The CONV beam diverged slightly less and had an FWHM of 19.3 mm at z=36 cm, at the end of the SOBP. (a) Full width at half maximum (FWHM) of the beam along the depth of the beam for UHDR (red) and CONV (blue) (b) Amplitude of the spot dose profile Dmax(z)/100MU measured over the depth of the beam for UHDR (red) and CONV (blue). The two dose rates follow a similar profile, except at the beginning and end of their respective SOBP. (c) comparison of the spot dose profile in Gy/s between z = 10 cm (green) and z = 34 cm (yellow) for UHDR (d) comparison of the spot dose profiles in Gy/s for CONV. The maximum amplitude of the spot dose profile Dmax(z) in UHDR with a requested beam current of 200 nA (Series 3, Table 1) was measured over the depth of the beam and decreased from ∼0.42 Gy per 100 MU in the entrance plateau to ∼0.18 Gy per 100 MU at the start of the SOBP and 0.15 Gy per 100 MU at z=34 cm (Figure 11a). This translated to ∼ 800 Gy/s at the z=10 cm and ∼ 280 Gy/s at z=34 cm (Figure 11c). This reduction in amplitude with depth was a result of the pencil beam broadening. This makes it harder to reach high instantaneous dose rates in the SOBP, but leads to less spatial variation in the dose rate during the delivery of a PBS field. The maximum amplitude analysis was repeated for CONV beams (Figure 11b) with a beam current of 4.5 nA (Series 4, Table 1) and Dmax was similar to the UHDR values, with exception of start and end of the respective SOBPs. In this case this translated to maximum instantaneous dose rates of ∼ 25 Gy/s at z=10 cm and ∼ 9 Gy/s at z=34cm (Figure 11d). The instantaneous dose rate was therefore reduced by a factor of ∼30 compared to UHDR. DISCUSSION A scintillator‐based, fibre‐coupled detector previously deployed for proton FLASH entrance‐plateau measurements 28 , 29 was characterised, calibrated, and validated for time‐resolved dosimetry in a passively generated SOBP. The detector response factorised into two physically distinct components: (i) a probe‐independent saturation term of the SiPM, ksat(MQ,spot), and (ii) a depth‐dependent correction kQ(R−z) that captures signal loss due to quenching and water‐inequivalence near the Bragg peak. A single third‐degree polynomial in the measured signal MQ,spot described ksat across all probes, while kQ rose towards the distal SOBP as expected from increased ionisation density. Within our measurement range, kQ functions for CONV and UHDR were overall similar, although the data suggest a modest dose‐rate–dependent increase in quenching at UHDR up to ∼2% (Figure 7b). However, this study was not designed to determine the difference between UHDR and CONV with high precision, therefore, more research is needed to support a definitive conclusion. The observed depth‐dependent under‐response is consistent with the well‐established quenching behaviour of scintillators at high LET. Similar correction factors have been reported for organic plastic scintillators 38 , 40 and for inorganic scintillators. 42 However, no studies have examined how quenching in inorganic scintillators depends on dose rate, representing a critical gap in the current literature. The factorised description enables a streamlined calibration procedure. ksat and kQ are established once for a given system and set‐up configuration and are independent of each other. Thereafter, changes in the absolute signal, for example from fibre re‐coupling, 28 radiation damage 43 moderate optical ageing, or mechanical handling, can be compensated by re‐determining a single absolute factor at a homogeneous reference depth, ND,w,15. In our implementation, two short reference irradiations (one low and one high dose rate) might suffice to re‐establish ND,w,15 when the scintillator probes and 2DRM are unchanged. The two different dose rates serve as a simple check for the calibration and should yield the same ND,w,15. If the 2DRM is replaced or substantially modified, kQ should be re‐measured, as it is specific to the beam energy distribution generated by the 2DRM. Since ksat is a material constant of the SiPM, it is independent of beam quality and could therefore be measured in an alternative set‐up without the 2DRM, solving the issue of inhomogeneity in the dose distribution. All in all, this approach aligns with the spirit of TRS‐398 calibration practice and reduces beam time compared with prior multi‐step routines. 29 , 37 After calibration, the system measured the integral dose in a point with a relative standard deviation of 0.5% against the IC reference and maintained geometric stability of ≤0.1 mm (1 SD). Other calibrations of scintillating dosimeters for proton beams have reached comparable measurement accuracy. 28 , 44 The system resolved instantaneous dose rates over more than four orders of magnitude, from 0.1 Gy/s up to ∼800 Gy/s, at 50 kHz temporal sampling. These characteristics match the requirements for time‐resolved QA of PBS beams where local, instantaneous dose‐rate metrics are increasingly used alongside or instead of field‐mean values for FLASH characterisation. 22 Taken together, the achievable precision is comparable to established reference systems for absolute dosimetry (e.g., calibrated ionisation chambers 45 and alanine pellets 46 ), while additionally offering sub‐spot‐duration temporal resolution. With the calibration in place, we measured how spot properties evolve with depth. The spot tails clearly deviated from a single Gaussian: outside approximately 2σ2 the profile was dominated by a Lorentzian component (Figure 10a and c). This behaviour is consistent with measurements at the same facility for mono‐energetic beams and with findings from other proton centres. 29 , 47 , 48 The determined UHDR spot size at z= 34 cm with a single Gaussian (σ= 8.05)(Figure 10b) agrees well with 2D measurements of the spot profile of a study conducted in the same setup: 49 The spot size was measured to be σx= 8.1 mm and σy= 8.0 mm with radiochromic film and σx= 7.7 mm and σy= 7.6 mm with a scintillating sheet. The beam broadened with depth: the FWHM increased from the entrance region towards the distal SOBP (Figure 11a), with a modest reduction in spot size at the end of range, which agrees well with the literature. 50 Consistent with the broadening, the measured maximum instantaneous dose rate decreased from ∼800 Gy/s at z=10 cm to ∼280 Gy/s at z=34 cm for UHDR settings (Figure 11c), which leads to reduced intra‐field dose‐rate variations during field delivery. These depth trends are directly relevant for FLASH dose‐rate planning in preclinical SOBP experiments: the broadened, Lorentzian‐tailed profiles at depth require more conservative assumptions for the minimum spot‐level dose rate achievable in the distal SOBP, and they underscore the value of high‐bandwidth, point‐wise verification in the actual delivery geometry. Several simplifying assumptions made during calibration were tested a posteriori. First, the difference in measurements depth of the IC and the scintillators were accounted for, by assuming a 2.4 mm difference, and then correcting for this with interpolation. There is a sub‐millimetre uncertainty on this depth difference, which could lead to a few percentage uncertainty, when comparisons are made in steep dose gradients, such as the distal edge. However, for the calibration factor ND,w,15, which is measured at the entrance plateau, this uncertainty is negligible. Second, the spot core was assumed approximately Gaussian within r≲σ; at z=34 cm a single Gaussian (σ=8.05 mm) indeed described the central profile well up to ∼15 mm radius (Figure 10b and d), supporting its use in determining ksat. Third, the absolute factor ND,w,15 was determined at z=15 cm, where film confirmed field homogeneity to within ∼0.9% in the central region. Further, the change in kQ between z=10 cm (where ksat was extracted) and z=15 cm (where ND,w,15 was measured) was <1% and therefore negligible for absolute scaling. Finally, treating lateral spreads in x and y as equal did not introduce appreciable bias, as evidenced by the tight collapse of 2D data onto the 1D radial profile (Figure 10). Compared with earlier implementations, the present procedure condenses detector behaviour into a compact, transferable parameter set (ksat, kQ, ND,w,15). This makes the method generally applicable to any time resolved point like detector, while also reducing recalibration effort after re‐assembly. The method assumes that ksat is a detector constant for a given SiPM/readout chain and that kQ depends on scintillator material and LET (i.e., on residual range R−z rather than initial beam energy), which was consistent with our data. Changes to the 2DRM or the scintillating crystal require re‐measuring kQ; minor optical changes can be absorbed by updating ND,w,15. The system is compatible with PBS delivery of an SOBP generated by a static 2D range modulator from a mono‐energetic beam, a configuration that mimics clinical delivery and yet maintains UHDR for preclinical studies. In upcoming work, the calibrated detector will be applied for in‐vivo murine measurements under both CONV and FLASH conditions 17 to provide QA for both dosimetric and geometric aspects of the planned irradiations. CONCLUSIONS In this study, a scintillator‐based dosimetry system was calibrated specifically for time‐resolved proton FLASH in‐vivo measurements in an SOBP. The calibration demonstrated high accuracy and precision in both dosimetric and geometric measurements. The properties of the SOBP UHDR and CONV beams created from the 2DRM were measured in preparation for the preclinical murine FLASH studies. This shows that the system can perform QA for preclinical studies in SOBP proton FLASH beams. CONFLICT OF INTEREST STATEMENT This study was partly funded by Varian, a Siemens Healthineers Company.