Dosimetric characterization of a novel commercial plastic scintillation detector with an MR-Linac

Background: Plastic scintillators have been used as radiation detectors for the past few years,as they are water-equivalent and independent of the dose,dose rate, and angle of incidence. In addition, they are also independent of the presence of a magnetic ﬁeld and could be used for in vivo dosimetry in an MR-Linac. With the advent of a new commercial scintillation detector, Blue Physics Model 10,its characterization has been performed on an MR-Linac with a view to future applications. Purpose: To perform the dosimetric characterization and study potential applications of a novel commercial plastic scintillation detector in a MR-Linac. Methods: Scintillation detector description, calibration procedure, short-term repeatability, dose-response linearity, dose-rate dependence, angular dependence, and temperature dependence have been studied. Percent-depth-dose (PDD) and beam proﬁles were measured for small ﬁelds and a standard ﬁeld, as well as output factors, for comparison with other PTW detectors: a diamond diode and PinPoint and Semiﬂex 3D ionization chambers. The suitability of the plastic scintillator for in vivo dosimetry in a magnetic ﬁeld has also been studied measuring the dose to a point in an anthropomorphic phantom while acquiring MR imaging. This measured dose was compared with that calculated with Monaco planning system and with that measured with a PTW Semiﬂex 3D chamber, the latter without acquiring MR images. Results: Short-term repeatability presented negligible variations ( < 0.4%) for 100 and 20 MU. Similar results were obtained for dose-response linearity and dose-rate dependence.A

resolution allows independent radiation pulses to be measured and visualized, which could be used in future applications.

INTRODUCTION
In recent decades, technological advances have developed new detectors for ionizing radiation, 1 each with its own advantages and disadvantages. The current practice for dosimetry in linear accelerator (linac) commissioning employs ionization chambers, diodes, and diamond detectors. 2 The choice between detectors depends on several factors, such as the type of application, field size, and resolution. This is also the case for the magnetic resonance image guided linear accelerators (MR-Linac), which integrate a linac with a magnetic resonance. 3 When an air-filled ionization chamber is employed in a magnetic field, the secondary electrons depositing the dose are deflected by the Lorentz force while they are moving. This effect is most noticeable at tissue-air interfaces 4 and it is not negligible inside the air cavity of the ionization chamber, and its dose response is influenced by the magnetic field. 5 The effect of the magnetic field in the dose response has less importance in a homogeneous medium, where there is equilibrium in the charged particles. 6 Solid state detectors, such as diodes and diamond detectors show near water equivalence and small active volume, but their dose response is affected by the magnetic field. 7 Certain materials produce scintillation light in response to ionizing radiation. 1 Detectors based on scintillation properties can be classified into inorganic and organic scintillators. Inorganic scintillators include alkali halide crystals like sodium iodide (NaI), and are mostly used in gamma-ray spectroscopy, while organic scintillators include liquids and plastics that although generally faster in their response time, yield less light. The use of plastic scintillators for the dosimetry of high-energy beams in radiotherapy has been used mainly since the beginning of the 90's following two publications by Beddar, Mackie, and Attix. 8,9 . In their work, the main characteristics of these dosimetry detectors were emphasized, such as their water equivalence for photon and electron beams, which avoids the usual conversion of dose measurements from one medium to another. Furthermore, they are almost independent of the temperature, present less radiation damage than diode detectors and better energy dependence than other systems (ionization chambers, LiF thermoluminescent dosimeters, film, and Si diodes). In addition, plastics represent an extremely useful form of an organic scintillator due to their ease of shaping and manufacturing. 1 However, until now, there has not been any available commercialized dosimetry system with multiple scintillator sensors and fully compatible with MR-Linac's.
Plastic scintillation detectors are mainly composed of a doped plastic attached to an optical fiber. Their output is composed of the light produced by the scintillation detectors and the undesirable light produced by the Cerenkov effect (also known as Cherenkov effect) and fluorescent radiation emission. The amount of light produced by the Cerenkov effect depends on the length of fiber exposed to the radiation. 10 Although Cerenkov emission has been shown to be suitable for radiotherapy dosimetry, 11 in this specific application of plastic scintillation detectors, the only desired component to measure the dose is the light produced by the plastic scintillator, as it is related to the radiation absorbed in the scintillator material. The Cerenkov and fluorescence radiation effects must be subtracted. Different methods to eliminate these undesired effects have been proposed such as: spectral method, 12 a discrimination method, 13 or the use of a second background fiber. 8,14 Studies on energy dependence of various organic scintillators such as anthracene or stilbene, and other liquid and plastic scintillators were performed previously, 15 as it was already known that these materials scintillate in response to nuclear radiation. Since then, several studies have investigated its application in radiotherapy, one of the first being the study of a new scintillator detector system for the quality assurance of Co-60 and high-energy therapy machines by Beddar. 16 More recently, Carrasco et al. 17 performed characterization studies on the first commercial plastic scintillation system available, the Exradin W1 scintillator (Standard Imaging Inc., Middleton, WI) for use in radiotherapy. In addition, Simiele et al. have several studies on the characterization of the response changes of various light guides used in megavoltage (MV) photon beam scintillation dosimetry as a function of irradiation conditions. 18 In addition, Simiele et al. studied the characterization of the intensity and spectral response changes in a plastic scintillation detector (PSD) as a function of magnetic field strength 19,20 and its impact on the stem-effect in plastic scintillation detectors. 21 An ideal detector should have good reproducibility, repeatability, linearity, high sensitivity and detection efficiency with no angular dependence. In addition, it should be easy to position with a stable signal over time with no dependency on radiation energy for small size fields. Furthermore, it should be capable of offering instantaneous readings within the range of the electrometer and present no dependency to temperature changes. 22 The purpose of this paper is to characterize a novel MRI compatible commercial plastic scintillation detector in a MR-Linac, to be subsequently used in various dosimetry applications in the presence of a magnetic field.

METHODS
For this study a novel commercial scintillator detector and acquisition unit were used, the Model 10 system (Blue Physics LLC, Lutz, FL, USA). All measures were performed on an Elekta Unity MR-Linac with a nominal photon energy of 7 MV and a 1.5 T magnetic field. The design of the Unity system can be found in several works 3,6,23 and requires a change in the way quality control (QC) is performed on other types of linac. Short-term repeatability, dose-response linearity, dose-rate, angular, and temperature dependencies have been measured. Furthermore, its suitability as a reference dose detector and its use for relative dosimetry have also been studied by measuring percentage depth doses (PDD) and output factors (OF). Finally, the viability of this scintillation detector as a dosimetry system while MR images are being taken for future in vivo applications was checked as well. All measurements were obtained with the long axis of the chamber/scintillator cladding material oriented perpendicular to the beam and parallel with the magnetic field. Manufacturer's recommendations were followed when specified to validate them.

Blue Physics Model 10 system
The Blue Physics Model 10 scintillator detector system consists of a plastic scintillation detector (PSD), two transport optical fibers, an acquisition unit and a minicomputer with the software. The PSD is a cylindrical shaped scintillator plastic of 1 mm diameter and 1 mm long, with a sensitivity volume of 0.785 mm 3 . The scintillator has a core material of polystyrene and a cladding material of acrylic. The scintillating core contains a combination of selected fluorescent dopants to produce the desired scintillation. When a pulse of radiation passes through the scintillator, it produces visible light.The amount of light generated is proportional to the amount of dose deposited in the sensor by that pulse. The sensor is encapsulated in a water equivalent plastic replica of the Semiflex ion chamber in order to facilitate its positioning within the water tank and solid water phantoms. Each scintillator is coupled to one of the transport plastic optical fibers, which is 0.25 mm in diameter and 20 m long. This fiber has a core of polymethylmethacrylate and a cladding of fluorinated polymer, and transports the light generated by the scintillator to the acquisition unit. However, when the high intensity radiation from the linac passes through this fiber it produces additional light caused by the Cerenkov radiation. Cerenkov radiation is a well-known effect that consists of a blue light produced in the optical fiber medium when the radiation from the linac generates electrons that move in the medium at a faster speed than the speed of light in that medium. 24 Cerenkov Radiation light will be added to the light generated by the scintillator and will be proportional to the length of the irradiated fiber, the intensity of the radiation and its angle of incidence on the fiber. In order to measure only the light from the sensor, the light produced by Cerenkov effect must be measured and subtracted.
To subtract the Cerenkov effect, Model 10 has a second identical fiber placed in an adjacent position along the first fiber but not connected to the PSD. The light produced by the Cerenkov effect in this second fiber will be almost identical to the light produced by the Cerenkov effect in the first fiber connected to the PSD. Measuring the amount of light in this second fiber allows subtracting the Cerenkov effect from the first fiber and measuring only the light from the PSD. This technique was previously described by Beddar et al. 8 The light signal from both fibers is transported to the acquisition unit ( Figure 1). The acquisition unit has a removable cartridge which contains two transducers that convert the light signals from the fibers into electric currents. The electric currents are accumulated in two identical capacitors during a period of time called integration time. The charge accumulated during the integration time in each capacitor is digitized and the data is sent to a minicomputer which has software that will plot and store the dose in real time. For each channel, there is a backup capacitor that will collect the charge if a Linac pulse occurs during the time of the readout. The readout time is 10 µs. The software interface allows the end user to choose between four different measuring ranges, with capacitors of 10 pF, 30 pF, 60 pF and 1800 pF and also to change the integration time between 300 and 700 µs. This way, the user can choose the ideal measuring range and integration time depending on the application and the intensity of the radiation. All the measurements performed in this work were done with the 10 µF capacitor and 700 µs integration time, as this was considered the most suitable for the signal to be measured in the Unity MR-Linac.
For this study, the set composed by the PSD, the fiber attached to it, the transducer attached to the other end of the fiber and the electronics where the charge is accumulated will be called the "Sensor Channel". The second parallel fiber with its transducer and its capacitor will be called "Cerenkov Channel". The crosstalk between channels was evaluated, concluding that there is not cross-talking between them.

Calibration
In order to subtract the Cerenkov radiation correctly the ratio between the two channels, Sensor Channel and Cerenkov Channel, must be known. This ratio is called Adjacent Channel Ratio (ACR).Even if both channels were identical without the PSD and both channels receiving exactly the same amount of light via the fibers, the reading of Sensor Channel (sch) and the reading of the Cerenkov Channel (cch) would not be the same due to the small differences in the optical couplings and in the electronic components. In this hypothetical situation, the ACR could be measured with the following equation: However, both channels are not identical because the Sensor Channel is connected to the PSD and the Cerenkov Channel is not. So, in this real situation, the manufacturer recommends measuring the ACR following these simple steps ( Figure 2): • Prepare first set up "A" placing the sensor under the beam field where a short length of fiber is exposed to the beam. Irradiate the sensor and the fiber with a known amount of radiation and take note of the readings in both channels. Let's call these readings schA and cchA for the reading in the Sensor Channel (sch) and the reading in the Cerenkov Channel (cch), respectively. • Prepare a second set up "B" making sure that more length of fiber is inside the radiation area but the same amount of radiation is delivered to the sensor as the dose delivered in the set up "A". The new readings will be called schB and cchB.
It can be deduced that since the only thing that changed between the two assemblies is the Cerenkov F I G U R E 2 Scintillator calibration steps. radiation in both channels, the ACR can be estimated as: The manufacturer even recommends repeating this exercise with several different setups in which the only thing that changes between them is the amount of fiber exposed, with the dose to the sensor always being the same. A scatter graph can be drawn with all the readings in the Cerenkov Channel represented in the x axis and all the readings in the sensor channel represented in the y axis. Then, ACR will be the slope of the best fit line between all the scatter points on the graph.
These calibration setups can be arranged in many ways. The manufacturer recommends (for conventional linacs) to irradiate the sensor with an asymmetric field with the sensor in the CAX and then rotating the collimator. This will create several set ups where the dose at the sensor would be always the same but the length of fiber under the radiation field will change. As it is not possible to rotate the collimator in the Unity MR-Linac, the sensor and fibers must be irradiated with different asymmetric fields, always leaving the sensor in the central axis of the beam.
Measurements have been carried out irradiating the scintillator with eight field sizes, 4, 5, 7, 9, 10, 12, 13 and 15 cm 2 , defined at the isocenter. All fields were axially symmetric. A pre-irradiation of 1000 MU (Monitor Unit) was performed.The scintillator was placed at the isocenter at 5 cm depth (SSD = 138.5 mm) with the PTW BEAMSCAN MR 3D water phantom (PTW, Freiburg, Germany). The required amount of MU to always deliver the same 200 cGy dose at the scintillator was calculated previously measuring the OF for the same field sizes with a PTV Semiflex 3D ionization chamber using the same setup. ACR was measured again after 6 months and more than 10000 MUs delivered to study its possible variation.

Dose calibration
Once the value of ACR is obtained, it is possible to calculate the amount of charge in nC produced by the light coming from the sensor only (CS) from the readings of both channels (sch and cch) using this equation: To calculate the amount of dose measured by the sensor, the calibration factor to convert charge measured in nC to cGy (CalF (cGy/nC)) has to be obtained. MR-Linac dose calibration was carried out previously in water following the O'Brien et al. formalism 25 for reference dosimetry in magnetic fields and an adaptation of the formalism by van Asselen et al 26 consistent with the Elekta advice. The MR-Linac was calibrated to give 100 cGy per 100 MU to the isocenter, at 5 cm depth in water, source to water surface distance (SSD) 138.5 cm, for a 10 × 10 cm 2 field and 0 degrees gantry angle. The PTW BEAMSCAN MR 3D water phantom was used and a PTW Farmer TW30013 ionization chamber was placed at the isocenter. Once the calibration is done, the Blue Physics Model 10 scintillator was cross-calibrated against the Farmer chamber. The scintillator was placed under the same conditions at the isocenter and 200 MU were delivered, controlled by the MR Linac monitor chambers. This measure was repeated three times. Then, CalF (cGy/nC) will be: Knowing ACR and the CalF calculate the dose measured from the sensor reading in nanocoulombs (CS) can be calculated: Alternatively, to ensure the calibration factor value, it was also obtained by performing multiple measurements with the scintillator at various known doses previously measured in water with a Farmer ionization chamber with the same setup already described. MU values delivered were: 1, 5, 10, 20, 40, 60, 80, 100, 120, 160, 200, 220, 260, 300, 350 and 400. CalF was determined from the slope of the dependence between both detectors.

Short-term repeatability
Short-term repeatability was carried out using the same setup in water as previously described. The scintillator was placed at the isocenter at 5 cm depth in water (SSD = 138.5 mm). To study the behavior of the system under different operating modes and to check its dependency, the scintillator was irradiated 10 times. All measurements were performed with 430 MU/min dose rate and 100 MU. To study the performance under low-dose conditions, the study was repeated delivering 20 MU.

Dose-response linearity and dose-rate dependence
Measurements were performed using the same setup in water as previously described, with the scintillator again placed at the isocenter at 5 cm depth in water (SSD = 138.5 mm). The linearity of the response with the dose was performed for MU values: 1, 2, 3, 4, 5, 10, 20, 50, 75, 100, 150, 200, 300, 500, 600, and 1000 MUs. Three measurements were done for each MU value. Dose value was calculated with Equation (5).
To measure the dose-rate dependence, with the same setup in water, the dose rate was changed from 43 to 430 MU/min, in 43 MU/min increments. Three measurements were done for each MU value. Only the time between pulses was varied, as the Elekta MR-Linac delivers always the dose at maximum dose rate with step and shoot IMRT technique.

Angular dependence
The angular dependence of the scintillator as a function of the gantry angle was studied. Measurements were carried out with the same setup in water as previously described. The scintillator was placed at the isocenter and irradiated every 45. 100 MU were delivered with a 10 × 10 cm 2 field size. Because the cryostat is filled with approximately 80% helium, the gantry angle was kept at 0 • and the scintillator was rotated instead, to avoid dependence on differences in the thickness of helium traversed, couch attenuation and magnet inhomogeneities. Since the scintillator does not have F I G U R E 3 In-house water phantom with a thermometer.
any markings indicating the zero angle, it was chosen arbitrarily.

Temperature dependence
The variation of the dose with the temperature was determined with the scintillator within an in-house water phantom (Figure 3), with the scintillator long axis parallel to the magnetic field. The water phantom was positioned in the bore with the scintillator plastic located at the isocenter. AP-PA and lateral MVI images were used to confirm the scintillator position. This reservoir was filled with previously heated water to 40.6 • C (104.6 • F) and 58 measures were taken as the temperature dropped. Each reading was performed with a 10 × 10 cm 2 field size, 200 MU and AP field. The temperature value was obtained through a calibrated probe installed inside the reservoir.

PDDs and profiles
The PTW BEAMSCAN MR 3D water phantom was used to perform the measurements of two different small fields and one standard size field. PDDs were measured with the scintillator point by point for 10 × 10 cm 2 and 1 × 1 cm 2 field sizes, and profiles for 10 × 10 cm 2 and 2 × 2 cm 2 field sizes. The profiles, crossplane, and inplane, were measured at 10 cm depth (SSD = 133.5 cm). To measure the inplane profiles and keep constant the portion of optic fiber inside the radiation field, the scintillator was placed with its axis perpendicular to the magnetic field using an in-house holder ( Figure 4). PDDs normalized to maximum and profiles normalized to CAX were compared to those measured with three different PTW detectors: TW31021 Semiflex 3D, TW31022 Pinpoint ionization chambers and a TW 60019 diamond diode. Data measured with the scintil-F I G U R E 4 PTW BEAMSCAN MR 3D water phantom with the scintillator placed with its axis perpendicular to the magnetic field. lator and the PTW detectors were analyzed with an in house Phyton software.

2.9
Output factors OF  for field sizes ≤3 × 3 cm 2 were applied to the PTW detectors, and was assumed 1 for the scintillator, as described by Galavis et al. 28 .

Anthropomorphic phantom measurements
The feasibility of measuring the dose while obtaining MR images in real time during treatment was tested with an anthropomorphic SRS head phantom (Integrated Medical Technologies, Troy, NY, USA). The phantom was scanned in a Philips Brilliance CT Diamond Select (Philips Medical Systems, Bothell, WA, USA), with the scintillator placed inside one of the inserts for ion chambers that the phantom has ( Figure 5). The images were sent to Monaco treatment planning system (v. 5.40, Elekta AB, Stockholm, Sweden). In the TPS, the scintillator was delineated and considered as CTV. Then, the planning target volume (PTV) was created with 1 cm margin from the CTV. 10 Gy in five fractions were planned to the PTV.
The phantom was positioned on the Unity couch with the scintillator placed inside the insert ( Figure 5).

F I G U R E 6 Phantom treatment planning (a). Coronal and sagittal MR images taken during the irradiation (b).
The cavity to insert the scintillator capsule was previously filled with water to prevent the presence of air gaps and the electron return effect (ERE). 25 Then, the normal Unity treatment workflow was followed. The treatment was adapted and delivered while acquiring MR image in real time with the scintillator in place ( Figure 6). This same treatment was adapted once more without any changes in phantom position. A PTW Semiflex 3D ionization chamber inside the phantom cavity instead of the scintillator and without MR imaging in real time for comparison.

Calibration
The procedure previously described after pre-irradiation with 1000 MU produced the scatter graph shown in Figure 7.
The procedure was performed three times and a final ACR value obtained was 1.085 ± 0.006. The ACR value measured again after 6 months and more than 10000 MU delivered resulted 1.083 ± 0.005, following the same methodology.

Dose calibration
Equation (5) was applied after repeating the measure with the sensor three times and utilizing  Figure 7). The calculated factor was obtained without subtracting the observed charge between pulses, since the subtraction of this dose was found to be incorrect in comparison with other detectors, determined from the slope of the dependence of the measured dose at various known doses with a Farmer ionization chamber and then with the scintillator resulted 0.9682 cGy/nC, a 0.33% higher. The CalF value used in this work was 0.9650 cGy/nC, as it was intended to follow the manufacturer's advice.

Short-term repeatability
The results obtained are shown in Table 1 for 100 and 20 MU delivered. The ACR and calibration factors previously obtained were applied to calculate the measured dose with the scintillator. For 100 MU the standard deviation of the measured dose was 0.31% and when 20 MU were delivered the standard deviation was 0.37%. Both values were calculated with respect to the mean value.

Dose-response linearity
The deviation from linearity remained below 1% for all MU values, except when only 1 MU was delivered, in which case deviation resulted in 1.16%. Linearity with the number of pulses recorded was also analyzed and similar results were obtained. In both cases, MU and pulses, the fit to a line gave an R 2 value of 1 (Figure 8a). Calculated root-mean-square value (RMS) as described in Carrasco et al., 17 with the equa- where "x i " are percentage deviations from a reference value and i = 1 to n are all measurements within the corresponding test, was 0.45%. Figure 8(b) shows the percentage differences in linearity of the measured local dose between the expected and measured values, from 1 MU to 1000 MU. Because of uncertainties in measuring low doses, the fit to a line does not cross the origin.

Dose-rate dependence
Deviation of the measured dose at different dose rates was 0.38% considering all the dose rates analyzed.
The RMS value of all differences with respect to the average value resulted 0.35%. Measured dose percentage deviation values showed in Figure 8(b) resulted negative as the treatment dose rate, 450 MU/min, was selected as reference. The measured dose was observed to increase with dose rate. Specifically, an increase of 1.19% was measured between the minimum and maximum dose rates analyzed.

Angular dependence
If the value obtained at 0 • gantry angle is taken as reference, the mean angular dependence was 0.24% ± 0.14%. Maximum deviation from the reference was found at 90 • (Figure 8d), which indicates that the scintillator is not perfectly centered inside the cladding material. The obtained RMS value shown in Table 2 was slightly larger, 0.28%.

Temperature dependence
Measurements performed in a temperature range of 19.2 • C (40.6-21.4 • C) resulted in a random variation with respect to the mean value of the measured dose over the entire range of 0.39% (Figure 9). The Blue Physics Model 10 system proved to be almost independent of the temperature, with small random variations around a mean value. This result is very important, as there is great interest in studying whether the system can be used for in vivo dosimetry on the Elekta Unity, with the scintillator in contact with the patient.

PDDs and profiles
Main characteristics of PDD and profiles measured in an Elekta Unity Linac have been described by other authors. 7   size, the standard deviation between measurements performed with all detectors is larger in the buid-up zone, up to 5.07% at the surface, but decrease to lower than 1% at 6 mm depth. For the larger field size, 10 × 10 cm 2 , again the greater standard deviation between all measurements is found in the build-up zone, up to 12.98% at the surface, but decrease to lower than 1% at 5 mm depth, and less than 0.5% for depths greater than 9 mm. The measured profiles are represented point by point in Figure 11, and the 80%-20% penumbra distances are given in Table 2 measured with the PTW detectors, within experimental uncertainity.
As expected, the Semiflex 3D ionization chamber gives the larger penumbra for both field sizes, due to volume averaging and non-water equivalence of the air inside the ionization chamber, what is also applicable to the PinPoint ionization chamber. 29 The values of the penumbras given by the scintillator are between to the ones obtained with the diamond detector and the PinPoint ionization chamber.

Output factors
Measured OF corrected by TRS-483 publication for small fields are shown in Figure 12. All values are normalized to 1 for 10 × 10 cm 2 field size. The value measured with the scintillator was the highest between all detectors and field sizes lower than 10 cm 2 . For the field 0.5 × 0.5 cm 2 , the output factor value with the scintillator was a 1.26% higher than the value obtained with the microdiamond, a 6.86% higher than the value obtained with the pinpoint ionization chamber and 16.52% higher than that measured with the Semiflex 3D.For field sizes 3 10 × 10 cm 2 all detectors register similar OF values, with a maximum standard deviation between all detectors of 0.39%, which corresponds to the biggest field size available in Unity, 22 × 22 cm 2 .

Anthropomorphic phantom measurements
Calculated dose with the TPS and measured dose corrected by the beam angle with the scintillator are presented in Table 3, as well as the differences in percentage between the measured dose with the Semiflex 3D and the calculated dose with respect to the value measured with the scintillator. The measured total dose with the scintillator was a 0.51% higher than the calcu-lated dose with Monaco TPS, while the measured total dose with the Semiflex 3D resulted 1% higher. If the measured doses are calculated field by field, differences with the semiflex chamber were between −5.77% and 4.64%, while with Monaco TPS were between −13.74% and 8.07%.In any case,the differences in doses per field are compensated at the end of the treatment for both comparisons. Figure 13 shows how the differences observed between the dose measured with the scintillator and the 3D Semiflex, as well as the dose calculated by Monaco, follow almost the same trend with angle, except for 102 • gantry angle.

DISCUSSION
This work aims to characterize a novel plastic scintillation dosimeter on an Elekta Unity MRI-Linac and study its possible applications as a dosimeter in a magnetic field. The scintillator can measure pulses every 700 ms and less, so it is capable of individually measuring the linac pulses and displaying them on screen, which may lead to the development of new ways of machine quality control and dose measurement.

Scintillator calibration and dose calibration
The proposed method was applied to obtain the ACR value, which allows posteriorly to convert the measured charge to measured dose with Equation (3). It was observed that the ACR value obtained is dependent on a previous irradiation. The ACR value changed if a previous irradiation was applied on the scintillator. Specifically, the ACR value changed 1.3% between the measures performed before and after the irradiation with 1000 MU. Following this previous irradiation, the value kept constant for the rest of the measurements. So, it seems important with this scintillator to perform a previous irradiation, and also to calculate the ACR value each time it is going to be used, as this parameter can slightly change between different days. Specifically, the ACR value standard deviation between different days,after the initial irradiation,was 0.51%,very close to the observed value obtained by Carrasco et al. 17 and Beierholm et al. 30 when analyzing another plastic scintillator.

4.2
Short-term repeatability, dose-response linearity, dose rate dependence The Blue Physics Model 10 scintillation detector system performed well in all three tests. Results were simi-F I G U R E 1 0 PDDs of 1 × 1 cm 2 and 10 × 10 cm 2 field sizes.
F I G U R E 1 1 2 × 2 cm 2 and 10 × 10 cm 2 profiles measured point by point with the Blue Physics scintillator, diamond detector, and PinPoint and Semiflex 3D ionization chambers. lar to those obtained by Carrasco et al., 17 Schoepper et al., 31 and Beierholm et al., 30 with a slightly better RMS value for dose-response linearity. In the latter test, the deviation from linearity was less than 0.6% for all MU supplied, except when 5 MU (0.92%) and 1 MU (1.19%) were delivered. The same test performed with a Farmer chamber and a diamond detector did not show this deviation from linearity at 5 MU (0.14% and 0.33%, respectively), but deviation with both detectors were higher with respect to the scintillator for 1 MU (2.3% and 4.35%, respectively). In fact, deviation from linearity resulted always higher with the Farmer and diamond detector with respect to the scintillator in the range of 4-1 MU.
Values of dose-rate dependence presented less variation than the ones reported by Schoepper et al 31  showed in Figure 8(b) resulted negative as the treatment dose rate,. 450 MU/min, was selected as reference.
It should be noted that the Blue Physics scintillator is capable of measuring independent pulses that can be visualized ( Figure 14) in real time. In addition, since independent pulses are displayed while measurements are being taken, it was observed that the linac changes the pulse rate to deliver the correct dose. If the number of pulses for MU is analyzed, between 1 MU and 1000 MU, it can be seen that the linac deliver 36.9 ± 0.28 pulses per MU.

Angular and temperature dependence
Angular dependence showed an RMS value of 0.28%. The standard deviation from the average value was 0.24 ± 0.14%. These values are similar to the ones obtained by Carrasco et al., 17 but the results indicate that the Blue Physics scintillator is not perfectly centered inside the cladding material.
Temperature dependence results show that the scintillator is almost temperature independent, allowing it to F I G U R E 1 4 Different pulse frequencies at different dose rates be used for in vivo measurements without concern for changes with temperature. Several studies on scintillators, such as those performed by Carrasco et al., Galavis et al.,Schoepper et al.,or Buranurak et al. 17,28,31,32 show different dependencies with the temperature that make it necessary to apply a correction factor, which is not necessary with the Blue Physics scintillator.

PDDs and profiles
The results obtained show the agreement between the PDDs measured with the scintillator and the rest of the detectors employed. The largest differences are observed in the build-up region due to differences between the cladding material covering the scintillator and the material of the rest of the detectors. These differences are not seen in the work on a different plastic scintillator by Okamura et al. 33 For depths beyond the maximum, the measurements of all the detectors used agree,with a standard deviation from the mean of less than 0.5%. The PDD measured with the scintillator appear more stepper as the field size decrease. This effect was also reported by Wilcox et al. 34 when comparing EBT film PDD measurements with other detectors, and the similar behavior of the scintillator can be attributed to the its equivalence to water. The depth of maximum with the scintillator is displaced 1 mm respect to the rest of the detectors, towards the surface for the 10 × 10 cm 2 field size, and deeper for the 1 × 1 cm 2 field size, what implies that the scintillator is not perfectly centered inside the cladding material. In any case, values measured with the scintillator agree with other published for the Unity Linac. 35 Measured profiles with the PinPoint and the Semiflex 3D ionization chambers are affected by the volume aver-aging and the non-water equivalence of the air inside the chamber 29,36 and a broadening of the penumbra is expected, which has been previously reported. 7 The penumbra measurement with diodes can exhibit perturbations due to the intrinsic detector materials and diode design. The electron transport will be changed if the detector is not water equivalent. In this case the range of the electrons is reduced relative to the range in water and the effect of lateral electron travel is reduced on the width of penumbra, sharpening the profile. As competing effect, the finite size of the detector will broaden the measured penumbra relative to the true penumbra. 37 The scintillator and its cladding material are water equivalent and the measured penumbras should not be affected by these issues. The penumbra values obtained with the scintillator are slightly higher than the ones measured with the diamond diode, since the profile sharpening effect does not occur with scintillators. This result coincides with those obtained in the work by Beddar et al., 37 which reports profile broadening of 0.5 mm, and is slightly observed in the work by Okamura et al. 33 when measuring profiles with a scintillator for CyberKnife small-field dosimetry. Differences in the outer field region of the 10 × 10 cm inline profile are due to the high Cerenkov emission compared to the signal registered by the scintillator,which makes it difficult to obtain the correct value. When measuring the X-profile the Cerenkov emission effect is less pronounced as less fiber is radiated.

Output factors
For field sizes smaller than 4 × 4 cm 2 , the OF obtained with the plastic scintillators were similar to those obtained with the diamond diode. This result was also reported by Okamura et al. 33 with another scintillator. Also, the OF obtained with the scintillator are in agreement with all detectors for larger field sizes, indicating that the scintillator is capable of correctly measuring OF with all possible field sizes in an MR-Linac assuming a correction factor k f clin ,f msr Q clin ,Q msr = 1, and can be used in several future applications.

Anthropomorphic phantom measurements
To our knowledge this is the first work to attempt measurements in an Elekta Unity MR-linac while MR imaging is being acquired in real time. With the Blue Physics scintillation detector, it is possible to measure the dose to a point while acquiring MR images quite accurately. Comparison with Monaco TPS and semiflex in Table 2 show analyzed field by field, show that all dose differences with the semiflex chamber were between −5.8% and 4.6%, while with Monaco TPS were between −13.7% and 8.1%. The largest differences with Monaco are due to beams which pass through heterogeneities in the anthropomorphic phantom, which have their electronic density overridden.
In any case, the differences in doses per field are compensated at the end of the treatment for both comparisons and the total treatment dose is measured with only 0.5% deviation respect to the average between the measured and calculated doses. Schoepper et al. 31 report dose differences between −1.67% and 3.15% for clinical patients, while Wootton et al. 38 in their study on rectal wall dose measurement in patients treated for prostate cancer report mean dose differences between 5.6% and 7.1% for four of the five patients studied. Further studies on the angular dependence and centering of the scintillator within the coating material with the Blue Physics scintillator will probably help to improve the measurement and apply more accurate correction factors.

CONCLUSION
The Blue Physics Model 10 scintillator detector represents an exceptional alternative to other detectors, due to its excellent dosimetric characteristics. In addition, it has the advantage over diodes of being equivalent to water, and its response is independent of the temperature, what make it suitable for in vivo dosimetry without the application of correction factors. Further studies should be carried out to study an observed drift of the signal over time and to study the possible degradation of the scintillation detector due to radiation. Since its not affected by magnetic fields, the dosimetry can be performed in a MR-linac while acquiring MR images in real time. The time resolution of the Blue Physics Model 10 scintillator allows to visualize each single pulse the linac emits, and should be investigate further as this feature could lead to new ways of machine QA.