Characterization of a beam-tagging hodoscope for hadrontherapy monitoring
O. Allegrini, J.-P. Cachemiche, C.P.C. Caplan, B. Carlus, X. Chen, S. Curtoni, D. Dauvergne, R. Della Negra, M.-L. Gallin-Martel, J. Hérault, J.M. Létang, C. Morel, ?. Testa, Y. Zoccarato
PPrepared for submission to JINST
Characterization of a beam-tagging hodoscope forhadrontherapy monitoring
O. Allegrini, 𝑎, J.-P. Cachemiche, 𝑏 C.P.C. Caplan, 𝑏 B. Carlus, 𝑎 X. Chen, 𝑎 S. Curtoni, 𝑐 D.Dauvergne, 𝑐 R. Della Negra, 𝑎 M.-L. Gallin-Martel, 𝑐 J. Hérault, 𝑒 J.M. Létang, 𝑑 C. Morel, 𝑏 É.Testa, 𝑎 and Y. Zoccarato. 𝑎 𝑎 Univ. Lyon, Univ. Claude Bernard Lyon 1, CNRS/IN2P3, IP2I Lyon, F-69622, Villeurbanne, France. 𝑏 Aix-Marseille Univ, CNRS/IN2P3, CPPM, Marseille, France. 𝑐 Université Grenoble Alpes, CNRS, Grenoble INP, LPSC-IN2P3, UMR 5821, 38000 Grenoble, France. 𝑑 Univ. Lyon, INSA-Lyon, Univ. Claude Bernard Lyon 1, UJM-Saint Étienne, CNRS, Inserm, CREATISUMR 5220, U1206, F-69373, LYON, France. 𝑒 Department of Radiation Oncology, Antoine-Lacassagne Cancer Center, Nice, France.
E-mail: [email protected]
Abstract: A beam tagging hodoscope prototype made of squared 1 mm fibers arranged in twoperpendicular planes and coupled to multi-anode photomultipliers has been studied using 65 MeVproton as well as 95 MeV/u C beams at various intensities. This hodoscope successfully provided2D images of proton beams with a detection efficiency larger than 98% with logical OR conditionbetween the two fiber planes. The detection efficiency with a coincidence between the two planes isclose to 75% for beam intensities up to ∼ Corresponding author. a r X i v : . [ phy s i c s . i n s - d e t ] J a n ontents Ion beam therapy is a rapidly expanding radiotherapy modality, with more than 250,000 patientsbeing treated worldwide . The main advantage of this technique lies in the dose conformity with thetarget volume resulting from the sharp Bragg peak observed in the dose profile at the end of the ionrange. Moreover, therapies with ions heavier than protons benefit from increased biological effectin the tumor region, hence enhancing the treatment effectiveness [1–5]. However, this therapytechnique faces uncertainties concerning the Bragg peak position mainly due to X-ray imagingmodalities, anatomical changes of the patient during the treatment, organ motion and approximationsused in dose calculation [6]. As a consequence, the most widespread treatment planning techniquesare performed with several beam incidences accommodating treatment robustness at the expense ofhigher doses in healthy tissues. Moreover, additional safety margins are applied around the tumorvolume to ensure the full irradiation of tumor cells [2, 7].In this context, clinically applicable methods and instruments are under development to monitor in vivo ion ranges and dose profiles with millimeter accuracy using secondary radiation detection.Some detection systems exploit the production of 𝛽 + emitters during nuclear reactions undergoneby a fraction of incident ions. At present, ion-range verification is only performed after treatment in – 1 –ome hadrontherapy centers employing commercial PET/CT scanners. But in-beam PET scannersare currently developed and tested in clinical conditions to provide online ion-range monitoring[8, 9]. Besides the production of 𝛽 + emitters, nuclear reactions also lead to the emission of promptgammas (PG) that can be also considered for ion-range verification [10]. Several PG modalitiesare under development or in their clinical setting evaluation worldwide and a few of them make useof time-of-flight (TOF) measurement either to derive indirect information on ion-ranges (PromptGamma Timing, PGT [11, 12]) or to reduce neutron-induced background in PG imaging systems(collimated and Compton cameras) [13–15].TOF measurement requires a time reference that cannot be provided by beam monitoringsystems implemented in clinical facilities (based on ionization chambers [16]) since they are notdesigned to provide such a timestamp. The solution consisting in using the accelerator radio-frequency (RF) as time reference would have the advantage of simplicity, when there are a perfectperiodicity and short ion bunches (cyclotron accelerators). However, the precise correlation betweenthe RF phase and the ion/bunch arrival can be obtained in mono-energetic beam conditions only.Indeed the use of degraders to change beam energies in cyclotrons introduces time dispersion andshifts of bunches. Moreover, small variations of the cyclotron’s magnetic field slightly affect theorbital frequency of the ion trajectories. Hence a small phase shift can occur at each turn betweenthe ion trajectory and the RF signal, which results in a time-varying measurable mismatch [17].Several alternative solutions are currently being studied to provide a more accurate time referencecorresponding to the arrival of incident ions or ion bunches.Among them, some devices under development based on scintillating fibers provide particlestracking with integration [18] or particle-per-particle [19–21] acquisition mode for various fieldsof application, including proton radiography [22] and ion-range verification during hadrontherapy[23]. Recently, the performance of a time-tracker for a prompt-gamma spectroscopy system allowingfor a background TOF rejection with a sub-nanosecond time resolution has been demonstrated [24].Among the scintillating fibers hodoscope currently developed, the one of the ClaRyS collaborationis dedicated to be coupled with a PG imaging system (collimated or Compton camera). The desireddetection efficiency of the beam tagging hodoscope should be around 90% for coincidence eventsin the X and Y planes to ensure a spatio-temporal tagging of a maximum number of incident ionsin order to minimize the loss of statistics. The time resolution must remain below 2 ns FWHM,which corresponds to the bunch width of a clinical cyclotron. Furthermore, the device should becompatible with clinical conditions both in terms of beam intensities leading to counting rates upto 100 MHz, which is almost the frequency at which the ion bunches are delivered in a cyclotronfacility, and radiation hardness (device operational for at least 1000 patient treatments, consideringan average dose of 60 Gy per treatment).The present paper reports on in-beam tests of the hodoscope developed by the CLaRyS col-laboration. At first, performance tests were performed at GANIL with 95 MeV/u carbon ions tomeasure the detection efficiency as well as the variation of the response of the scintillating fibersas a function of the beam fluence. The second part of the paper presents the setting method of thedata acquisition system (thresholds and gains) and the results of the in-beam characterization per-formed during two experiments at the Mediterranean Protontherapy Institute, in Nice. Multiplicity,detection efficiency and time resolution were assessed.– 2 – Material and methods
The beam tagging hodoscope of the CLaRyS collaboration is designed to provide a spatio-temporaltagging of ion (synchrotrons) or ion bunches (cyclotrons) passages for counting rates up to 100 MHz,which corresponds to a period of 10 ns, the typical period of cyclotron beam microstructures inhadrontherapy centers.The final version of the beam tagging hodoscope is composed of 1 mm square-sectionpolystyrene scintillating fibers BCF-12 manufactured by Saint-Gobain [25]. The beam tagginghodoscope is constituted of two perpendicular planes of fibers. Each plane contains 128 fibers,which gives an active area of 128 ×
128 mm (Figure 1a). The scintillating fibers are kept togetherwith a transparent glue and accurate positioning of the fibers in the layers is ensured by the frameof the hodoscope (Figure 1b) and the eyelets pierced through it where they are coupled with opticalfibers. During operation, the sensitive part is isolated against external light. Fibers are readout onboth sides by 8 Hamamatsu multi-anode photomultiplier tubes (PMTs) H8500C. The number ofread-out channels is then 512. Each PMT is segmented in 64 pixels of 5.8 × and every fibersides are fixed to the PMTs anode surfaces through a plastic custom mask. Connections with fibersare made in such a way that every fiber is read-out by two neighboring channels of a PMT and twoadjacent fibers are handled with different PMTs. This configuration allows for optimizing detectorefficiency and time resolution. (a) Vue isométriqueEchelle : 1:2 Vue de faceEchelle : 1:1 - Hodoscope 2x128 fibres - A Vue isométriqueEchelle : 1:2 Vue de faceEchelle : 1:1 - Hodoscope 2x128 fibres - A (b) Figure 1 : (a) Large version of the beam tagging hodoscope. (b) Scheme of the frame of the hodoscope andeyelets arrangement. Each PMT is linked to a front-end (FE) card via a 64-channel connector. The main componentsof this card are two 32-channel readout ASICs “HODOPIC”, a signal-processing FPGA, a single-channel optical transceiver and an RJ45 connector [26]. As detailed in Figure 2, the ASICs providelogic signals associated to each channel and the logical OR of these signals, that triggers the storageof the channel states in an ASIC register. A specific gain is assigned to each channel while acommon threshold is applied on all channels of one ASIC. When a logical OR is generated, theduration of the logic signals corresponds to the time for which the signal amplitude is larger than– 3 –he threshold. It is worth noting that it has to be larger than the time required to generate the logicalOR which is about 1 . Hodoscope DAQDiscriminators
ASICs (×2)Front-End board
Logical OR
Time required: 1.5 ns
THR (×1)Discriminators TDCTDCExternal triggerAnalogsignals(×32) Logic signals (×32)Gains (×32) Coincidence B u ff e r Send
FPGA window R eg i s t e r Figure 2 : Simplified flowchart of the signal fibers processing in the FE card. The flow of signals and datais represented with black arrows while parameters are in green and writing order in blue.
Apart from signal processing, the FPGA has to determine the time difference between thetrigger signal sent by the gamma camera and the time stamp associated to the hodoscope. Thistime difference is indeed required for the TOF measurement. The time stamps associated to thetwo signals (trigger and hodoscope) is determined by three time-to-digital converters (TDC), onefor the trigger and one for each ASIC. The time stamp value of the hodoscope corresponds to thelowest value measured by the two TDCs associated to the ASICs. Their time resolution (LSB, LeastSignificant Bit) is of 0.3125 ns (hodoscope) and 0.625 ns (trigger). A single time difference valueis therefore provided per ASIC for which at least one channel has been hit.If it falls within a given time window, the data are sent from the FPGA to the back-end card(AMC40) [27] through a 3 Gbit · s − optical fiber with a specific protocol [26, 28–30] and then to thePersonal Computer (PC) through 1 Gbit · s − ethernet link where data are processed and stored bythe data acquisition software. In the case of the nominal acquisition mode, the data consists of thelist of hit fibers with the fiber number and the associated time-of-arrival (TOA). Both optical andethernet links also transmit slow-control packages sent by a LabVIEW-based program. Figure 3shows schematically the experimental setup in which the whole acquisition chain is illustratedwith front-end (HODOPIC) and back-end (AMC40) cards, the slow control system and the dataacquisition software.In order to monitor the ageing of the fibers, another data acquisition mode has been foreseenin which the signal amplitude of a given channel is measured using an ADC implemented in theASIC.For this study, a smaller hodoscope with 32 fibers per plane has also been developed in orderto use a single acquisition board to collect all the data of the two planes. For this small prototype,each fiber plane is readout from a single side and by a single ASIC.Overall, the data acquisition parameters of the hodoscope are the following: the PMT high-voltage and the ASIC gains and thresholds. As detailed in section 2.5, these parameters can beoptimized to improve the data collection and the signal-to-background ratio.– 4 – SS P r o t o n s : M e V CAEN module (discrimination + coinc.)
H: small hodoscope prototypeS: plastic scintillator
JTAG
Systemconsole Trigger
JTAG: slow control & monitoring
Inputchannels ASICs FPGA
HODOPIC μ TCA A M C PC(DAQ software)slow control &monitoring(LabVIEW)
Figure 3 : Scheme of the data acquisition chain of the experimental setup.
In the final configuration of the CLaRyS camera, the beam tagging hodoscope is coupled to agamma detector in which the PG are absorbed. A trigger signal is generated by the absorber uponthe detection of a PG. This trigger signal is then sent to the FPGA of the hodoscope FE card. Thecharacterization of the hodoscope efficiency requires a suitable device to detect the protons passingthrough the hodoscope. For this reason, in this experimental study, the trigger signal provided bythe absorber of the CLaRyS prototype is replaced by an external trigger obtained by the coincidencesignal of two plastic scintillators (PSs) surrounding the hodoscope. Figure 4 illustrates the setupused to assess the performances of the hodoscope during in-beam tests.The fact that these detectors are slightly larger than the beam hodoscope does not matter sincewe have verified that the whole beam crosses the hodoscope (Figure 8 in the section 3). PSs arealigned with the beam and located about 5 cm upstream and downstream of the hodoscope. Theirsignals are read out by PMT which are connected to a Nuclear Instrumentation Module (NIM). Thismodule ensures the analog to logic signal conversion thanks to a preset threshold. This configurationallowed us to generate double (two PSs) and triple coincidences (two PSs AND the beam tagginghodoscope). A logic signal is generated by the hodoscope when one (logical OR) or two fiberplanes (logical AND) detect a particle. The ratio of the number of triple and double coincidencesis a direct measurement of the detector efficiency.
The setup described in the previous section was used to measure the hodoscope efficiency with95 MeV/u carbon beams at low intensities (20–30 pA) and to assess radiation damage. The singleHamamatsu multi-anode PMT connected to the small hodoscope prototype was biased at 800 V. Itsreadout was performed with various NIM modules to provide trigger signals (discriminator module),charge integration (QDC module) and timing (TAC module) . The discriminator threshold was setat 30 mV, which corresponds to about one sixth of the maximum of the amplitude distribution. QDC: charge-to-digital converter : TAC: time-to-amplitude converter. – 5 – igure 4 : Typical experimental setup used to assess the performance of the hodoscope during in-beam tests.The two plastic scintillators are used in order to provide an external trigger signal when a proton impingesthe hodoscope.
The effect of radiation damage was estimated by comparing the detection efficiency measured fordifferent fluences with high beam intensities (2–3 nA).
Since the counting rate capability of PS is around 10 Hz, a beam current monitor (BCM) [24, 31]was used to monitor larger beam intensities. It consists of a scintillator placed out of the beamirradiation field and calibrated with the in-beam PS at low beam intensity and with the intensitymeasured in the cyclotron stripper at high intensities.Moreover, the three single independent signals and the coincidence module output signal weresent to the LabVIEW-based program to visualize the various counting rates and to provide onlinemonitoring of the beam intensity. Finally, the time selection window applied in the FPGA of theFE card, before data transmission, was tuned from an additional PC through a JTAG link.The MEDICYC low energy treatment line of the CAL is intended for the treatment of oculartumors. The research area of this beam line is located a few meters upstream of the treatmentroom. The maximum beam energy is 64.5 MeV and the high frequency of the accelerator is24.85 MHz so that the beam consists of proton bunches arriving every 40.24 ns. Throughoutthe rest of the paper, beam intensities are expressed in Hz because the mean number of protonsper bunch is always below 1 for the range of intensities (much below typical clinical ones) usedduring these experiments. The counting rates recorded in the PSs are approximated to units of Hz.The performance of the hodoscope is mainly assessed in terms of detection efficiency and spatialresolution. The number of hit fibers per plane when a trigger is generated (multiplicity 𝑀 ) is alsoconsidered in this characterization study as well as in the detection profiles. JTAG: Joint Test Action Group. – 6 – .5 Criteria and settings
As mentioned in section 2.1, the data acquisition parameters of the hodoscope are the PM high-voltage and the ASIC gains and thresholds. Their settings aim at finding the best compromisebetween noise rejection and detection efficiency.The standard method consists in measuring the so-called S-curves obtained using a logicperiodic signal sent to each input channel, one by one. The S-curve corresponds to the number ofpulses detected by the FE card over a given acquisition time as a function of threshold. Actually, twotypes of events can be distinguished according to the various signals provided by the ASIC, namelythe logic signals associated to each channel and the request signal (logical OR of all channels): • good events (GEs): events for which at least one logic signal state is high, • events with short pulse (ESPs): undesired events for which the logic signal is not long enoughto have a high state when the reading of the channel states by the FPGA is requested by theASIC (see section 2.1).For illustration purpose, a typical S-curves is shown in Figure 5. As expected the GE curvepresents a plateau (the expected number of pulses during the acquisition time) up to the sharp fall-offfor threshold values close to the signal amplitude (point B). In this fall-off region, the larger thethreshold the smaller the probability to have long enough logic signals to be readout by the FPGA.This is the reason why the ESP curve presents a maximum at the bottom of the GE curve fall-off.Finally the peak observed in the ESP curve for very low threshold values (below point A) is due tothe electronic noise. The interval defined by points A and B corresponds to the suitable thresholdrange (STR) for a given channel gain. C o un t s GEESP
A B
Figure 5 : Number of GEs and ESPs events as a function of the threshold (THR) for a given channel at again value of 1.50. The range between points A and B corresponds to a suitable threshold range for thischannel.
The method used to determine a set of gains and threshold is described in Figure 6. It is basedon three main steps: – 7 – a first scan of the input channels consists in determining the noisiest channel for the minimalgain (0.25), i.e. the channel for which point A (in figure 5) corresponds to the largest thresholdvalue. Points A and B of this channel are then used as a reference STR; • the STR of the other channels is then measured for the minimal gain (0.25). The cumulativefunction of the overlap of this STR with the reference STR is then computed. The maximumof this distribution is defined as the optimal threshold value; • for channels whose STR does not include the optimal threshold value, the gain is increasedstep by step until the optimal threshold value shows up in the STR. optimal threshold value (OTV) Determination of the noisiest channel Cumulative function of the STR with minimal gain Is OTV withinall STR?Reference STR YesSelection of channels whose STR does not include the OTV optimal threshold value (OTV)
NoGain increasing step by step Recording
Figure 6 : Block diagram of the method used to set the threshold value of the ASICs and the gain ofthe channels: the optimal threshold value (OTV) is obtained from the cumulative function of the suitablethreshold range (STR). Gains are adjusted next to the OTV determination.
Finally, the cross-talk between neighboring pixels has been assessed by scanning the PMTswith a blue LED. This LED is mounted on a motorized double-axis table having a step resolutionof 20 µ m. It produces light pulses synchronized with a pulse generator, which are split in twopulses with a mirror. One is sent to the H8500C PMT via an optical fiber in order to obtain alight spot perpendicular to the cathode surface and the other to an Hamamatsu R5600 PMT used asreference for the correction of temperature fluctuations. Although slightly larger than the providerspecifications, the cross-talk measured with our setup always remains below 3%. This fraction ofcross-talk events can therefore be considered as relatively low. Moreover, since the associated signalamplitude is relatively low, one can expect that this source of background is cut with thresholdsdetermined by the setting of the ASIC parameters (gains and threshold) [32]. A total detection efficiency of 94% was obtained with a single plane, while the fraction of eventswith multiplicity
𝑀 > × ions · cm − , which represents more than 1000 patienttreatments, but it was almost recovered by lowering the threshold. Close to 90% detection efficiencywas kept after 3.6 × ions · cm − . Concerning the irradiation damages, it should be noted thatafter few days, a partial restoration of the response at laboratory test benches has been observed,but it was not possible to repeat the detection efficiency measurement with carbon ions. The timeresolution was measured between the logical OR of the Y plane and the cyclotron high-frequencysignal and was assessed to 550 ps RMS. Table 1 : Evolution of the hodoscope response as a function of 95 MeV/u carbon ion fluence. DT: Discrimi-nator threshold. The detection efficiency is measured on a single fiber plane with a multi-anode PMT H8500operated at 800 V. A threshold of 30 mV corresponds to 4.5 pC measured on QDC. The first column with anull fluence corresponds to a negligible fluence after a few short low-intensity irradiations at the beginningof the experiments.
Fluence (cm − ) ∼ ± × (2.2 ± × (2.2 ± × (3.6 ± × DT (mV) 30 15Mean QDCvalue (pC) 35 34 27 21 21Efficiency (%) 94 94 63 92 86
Figure 7 presents the distributions of multiplicity 𝑃 ( 𝑀 ) of both planes for three beam intensities:17 kHz, ∼ ∼
20 MHz. The mean number of protons per bunch is 6.8 × − , 5.2 × − and 8.0 × − respectively. The Poisson distribution of the number of protons per bunch for bunchesconsisting of at least 1 proton is superimposed for each beam intensity since it corresponds to theexpected multiplicity if we neglect the multiple proton arrivals in a single fiber. 𝑃 ( 𝑀 = ) = ∼ 𝑃 ( 𝑀 = ) values of 80% and70% for X and Y planes respectively, to be compared to the expected value of 100%. Experimentalevents with 𝑀 = 𝑀 ≥ 𝑃 ( 𝑀 ≥ ) remains almost constant, the experimental distribution of multiplicity signif-icantly deviates from the expected distributions. Further improvements of the ASIC are thereforerequired to be compliant with beam intensities used in clinical centers.– 9 – - - -
10 1 P ( M ) Poisson distributionX planeY plane (a) - - -
10 1 P ( M ) Poisson distributionX planeY plane (b) - - -
10 1 P ( M ) Poisson distributionX planeY plane (c)
Figure 7 : Distribution of event multiplicity 𝑃 ( 𝑀 ) for beam intensities of (a) 17 kHz, (b) ∼ ∼
20 MHz.
The monitoring software reconstructs two-dimensional maps for each acquisition. Figure 8represents maps obtained for two acquisitions at ∼ ∼
20 MHz. Coincidence events(logical AND between both fiber planes) are reconstructed using the average position in X and Y forevents with
𝑀 >
1. Although the shape of the beam varies between the two intensities, the centerof the beam is clearly defined. The shape modification is attributed to the change of the focusinglens of the beam when the intensity increases. P o s i t i o n Y · (a) P o s i t i o n Y · (b) Figure 8 :
2D map reconstructed from data collection of coincidence events (logical AND between X and Yplanes) for beam intensities of (a) ∼ ∼
20 MHz.
Figure 9 represents the X and Y profiles obtained with two events selection methods for theirradiation with a beam intensity of ∼ 𝑀 >
1) while the red curve has beenobtained with the selection of 𝑀 = ∼
20 MHz, the overlap of the two curves is lost due to the increase of “wrong hits” (notshown).Considering that no structure is expected in the beam shape, the structures observed in the1D profiles are attributed to slightly different efficiencies over the X and Y planes. The order of– 10 – · C o un t s · ‡ M M = 1 (a) · C o un t s · ‡ M M = 1 (b)
Figure 9 :
1D profiles of (a) X and (b) Y fiber planes with multiplicities M=1 and M ≥ ∼ magnitude of these differences has been assessed focusing on the fibers obviously under-responding,namely fiber 23 in X plane and fibers 15, 16 and 20 in Y plane. Applying a linear interpolationover these fibers, we derived an estimate of the efficiency loss due to these fibers. Overall, theloss increases with the beam intensity from 0.75% to 4% and 2.7% to 4% in the X and Y planes,respectively. Figure 10 represents the detection efficiency of X and Y planes separately and with logical OR andAND conditions on the data of both the planes for various intensities from 2 kHz to ∼
20 MHz.The data are corrected for under-responding channels (cf. 3.2.1). The detection efficiencies in Xand Y planes keep almost constant values around 84% and 90% respectively for intensities under ∼ ∼
20 MHz due to current ASIC limitations, especially ground fluctuations leading to data acquisitionretriggering and consequently dead time.
Figure 11 shows the distributions of the time difference between the trigger signal and the firstlogic signal associated to each fiber plane (X and Y planes) for beam intensities of 43 kHz (11aand 11b) and ∼
10 MHz (11c). The distributions obtained with a beam intensity of 43 kHz presentwell-defined peaks whose widths have been assessed employing a Gaussian fit. The measuredfull widths at half maximum (FWHM) are close to 1.8 ns which fulfills the specifications. Thedistributions at ∼
10 MHz reveal an additional component before the main peak that is due to ASICoscillations. Note that the time distributions are governed not only by the time resolution of thehodoscope but also by the ones of the PS detectors, which have not been measured.– 11 – - -
10 1 10Beam Intensity (MHz)50556065707580859095100 D e t ec t i o n e ff i c i e n c y ( % ) X planeY planeCoinc. XY (logical AND)Coinc. XY (logical OR)
Figure 10 : Hodoscope detection efficiency as a function of the proton beam intensity. A correction factor hasbeen applied to take into account the loss of efficiency due to under-responding channels (see section 3.2.1).
50 52 54 56 58 60 62 64Time difference (ns)0100200300400500600700800 · C o un t s (a)
50 52 54 56 58 60 62 64Time difference (ns)0100200300400500600700800900 · C o un t s (b)
50 52 54 56 58 60 62 64Time difference (ns)02004006008001000120014001600 · C o un t s (c) Figure 11 : Distribution of time difference between X and Y fibers (red and blue curves, respectively) andthe trigger signal for beam intensities of 43 kHz ((a)–(b)) and ∼
10 MHz (c).
The beam tagging hodoscope under development intends to provide incident ion detection withan efficiency larger than 90% with a time resolution below 2 ns FWHM. The life duration of thescintillating fibers should be typically larger than 1000 patient treatments and the device should becapable to cope with 100 MHz counting rates expected with beam microstructures encountered intreatment centers.Regarding radiation hardness, the experiment performed at GANIL represents the worstcase for radiation damage in hadrontherapy (carbon ions with less than 3 cm range in water).During a single irradiation experiment, less than 10% efficiency decrease was observed with– 12 –3.6 ± × ions · cm − , which represents more than 1000 clinical irradiations for the central ho-doscope area, without accounting for the longer time scale recovery. Comprehensive studies havebeen reported in [34] and [35]. For instance Joram et al. mention a dose of 50 kGy beyond whichLHCb scintillating fibre suffer from scintillation light yield degradation. Although our criterion toestimate fiber degradation is slightly different it is worth noting that the aforementioned fluence of95 MeV/u carbon ions corresponds to a dose of ∼
150 kGy which is of the same order of magnitudeof the one reported by Joram et al.In the meantime, significant progress has been made in the development of FE electronicsand data acquisition system in order to fulfill the specifications in terms of detection efficiency,time resolution and counting rate capabilities. Moreover, specific configuration methods of theacquisition boards have been defined. Although the detection efficiency with coincidence betweenthe X and Y planes is ∼ ∼
10 MHz (figure 11c).Although the time resolution of the CLaRyS hodoscope is slightly larger than the one of otherprototypes developed worldwide, it is close to the expectations and further improvements can beforeseen with an upgrade of the ASIC. Moreover, the detection efficiency of a single fiber plane iscomparable to the one of other prototypes for which the detection efficiency with a logical ANDbetween two planes has not been reported yet.
The performances of the CLaRyS beam tagging hodoscope have been assessed in terms of detectionefficiency, time resolution, event multiplicity and radiation hardness during several in-beam tests.A methodology for the configuration of the ASIC thresholds and channel gains has been tested andit allowed us to obtain a detection efficiency larger than 98% with a logical OR and 72% with alogical AND between X and Y fiber planes as well as a time resolution lower than 2 ns FWHMfor intensities under 1 MHz. In conclusion, these performances are in accordance with the state ofthe art. Further improvements of the “HODOPIC” ASIC are required to reach the counting ratecapability defined in the specifications (100 MHz).
Acknowledgments
This work was partially performed in the framework of Labex PRIMES (ANR-11-LABX-0063)and within the frame of the EU Horizon 2020 project RIA-ENSAR2/MediNet (654 002).– 13 – eferences [1] S. Braccini,
Astroparticle, Particle and Space Physics, Detectors and Medical Physics Applications ,World Scientific (2010), pg. 598–609[2] M. Durante and H. Paganetti,
Nuclear physics in particle therapy: a review. , Rep. Prog. Phys. (2016) pg. 096702[3] D. Schardt, T. Elsässer, and D. Schulz-Ertner, Heavy-ion tumor therapy: physical and radiobiologicalbenefits. , Rev. Mod. Phys. , (2010) pg. 383–425[4] H. Paganetti and P. Van Luijk, Biological considerations when comparing proton therapy with photontherapy. , Semin. Radiat. Oncol. , (2013) pg. 77–87[5] O. Jäkel, The relative biological effectiveness of proton and ion beams. , Z. Med. Phys. , Range uncertainties in proton therapy and the role of Monte-Carlo simulations. , Phys.Med. Biol. , (2012) pg. R99–R117[7] A.-C. Knopf and A. J. Lomax, In vivo proton range verification: a review. , Phys. Med. Biol. , (2013), pg. R131[8] Y. Shao, X. Sun, K. Lou, X. R Zhu, D. Mirkovic, F. Poenisch, and D. Grosshans, In-beam PETimaging for on-line adaptive proton therapy: an initial phantom study. , Phys. Med. Biol. , (2014)pg. 3373–3388[9] V. Ferrero, E. Fiorina, M. Morrocchi, F. Pennazio, G. Baroni, G. Battistoni, N. Belcari,N. Camarlinghi, M. Ciocca, A. Del Guerra, M. Donetti, S. Giordanengo, G. Giraudo, V. Patera,C. Peroni, A. Rivetti, M. Da Rocha Rolo, S. Rossi, V. Rosso, and M. Giuseppina, Online protontherapy monitoring: clinical test of a silicon-photodetector-based in-beam PET. , Sci. Rep.-UK , (2018) pg. 4100.[10] J. Krimmer, D. Dauvergne, J.M. Létang, and É. Testa, Prompt-gamma monitoring in hadrontherapy:A review. , Nucl. Instrum. Meth. A , (2018) pg. 58–73[11] C. Golnik, F. Hueso-González, A. Müller, P. Dendooven, W. Enghardt, F. Fiedler, T. Kormoll,K. Roemer, J. Petzoldt, A. Wagner, and G. Pausch Range assessment in particle therapy based onprompt γ -ray timing measurements. , Phys. Med. Biol. , (2014) pg. 5399–5422[12] S. Marcatili, J. Collot, S. Curtoni, D. Dauvergne, J-Y. Hostachy, C. Koumeir, J.M. Létang,J. Livingstone, V. Metivier, L. Gallin-Martel, M.L. Gallin-Martel, Muraz J.F., N. Servagent, É Testa,and M. Yamouni, Ultra-fast prompt gamma detection in single proton counting regime for rangemonitoring in particle therapy. , Phys. Med. Biol. , (2020) [physics.med-ph/2001.01470][13] M. Fontana, J.-L. Ley, D. Dauvergne, N. Freud, J. Krimmer, J. M. Létang, V. Maxim, M.-H. Richard,I. Rinaldi, and É. Testa,
Monitoring ion beam therapy with a Compton camera: simulation studies ofthe clinical feasibility. , IEEE Trans. Radiat. Plasma Med. Sci. , (2020) pg. 218–232[14] R. Dal Bello, P. Magalhaes Martins, S. Brons, G. Hermann, T. Kihm, M. Seimetz, and J. Seco, Prompt gamma spectroscopy for absolute range verification of C ions at synchrotron-basedfacilities. , Phys. Med. Biol. , (2020) pg. 095010[15] S. Aldawood, P.G. Thirolf, A. Miani, M. Böhmer, G. Dedes, R. Gernhäuser, C. Lang, S. Liprandi,L. Maier, T. Marinšek, M. Mayerhofer, D.R. Schaart, I. Valencia Lozano, and K. Parodi, Developmentof a Compton camera for prompt-gamma medical imaging. , Radiat. Phys. Chem. , (2017) pg. 190– 197 – 14 –
16] H. Stelzer and B. Voss, Ionization chamber for ion beams and method for monitoring the intensity ofan ion beam, US Patent 6,437,513. (2002)[17] J. Petzoldt, K-E. Roemer, W. Enghardt, F. Fiedler, C. Golnik, F. Hueso-González, S. Helmbrecht,T. Kormoll, H. Rohling, J. Smeets, T. Werner and G. Pausch,
Characterization of the microbunch timestructure of proton pencil beams at a clinical treatment facility. , Phys. Med. Biol. , (2016) pg.2432–2456[18] B.D. Leverington, M. Dziewiecki, L. Renner, and R. Runze, A prototype scintillating fibre beamprofile monitor for ion therapy beams. , J. Instrum. , (2018) pg. P05030[19] S. Horikawa, I. Daito, A. Gorin, T. Hasegawa, N. Horikawa, T. Iwata, K. Kuroda, I. Manuilov,T. Matsuda, Y. Miyachi, A. Riazantsev, A. Sidorov, N. Takabayashi, and T. Toeda, Development of ascintillating-fibre detector with position-sensitive photomultipliers for high-rate experiments. , Nucl.Instrum. Meth. A , (2004) pg. 34 – 49[20] P. Achenbach, C. Ayerbe Gayoso, J.C. Bernauer, R. Böhm, M.O. Distler, L. Doria,M. Gómez Rodríguez de la Paz, H. Merkel, U. Müller, L. Nungesser, J. Pochodzalla, S. SánchezMajos, B.S. Schlimme, Th. Walcher, M. Weinriefer, L. Debenjak, M. Potokar, S. Širca, M. Kavatsyuk,O. Lepyoshkina, S. Minami, D. Nakajima, C. Rappold, T.R. Saito, D. Schardt, M. Träger, H. Iwase,S. Ajimura, A. Sakaguchi, and Y. Mizoi. In-beam tests of scintillating fibre detectors at MAMI and atGSI. , Nucl. Instrum. Meth. A , (2008) pg. 353–360[21] S. Braccini, A. Ereditato, F. Giacoppo, I. Kreslo, K. P. Nesteruk, M. Nirkko, M. Weber, P. Scampoli,M. Neff, S. Pilz, and V. Romano, A beam monitor detector based on doped silica and optical fibres. , J.Instrum. , (2012) pg. T02001[22] D. Lo Presti, D.L. Bonanno, F. Longhitano, D.G. Bongiovanni, G.V. Russo, E. Leonora, N. Randazzo,S. Reito, V. Sipala, and G. Gallo, Design and characterisation of a real time proton and carbon ionradiography system based on scintillating optical fibres. , Phys. Medica , (2016) pg. 1124–1134[23] A. Papa, P.-R. Kettle, E. Ripiccini, and G. Rutar, Scintillating fibres coupled to silicon photomultiplierprototypes for fast beam monitoring and thin timing detectors. , Nucl. Instrum. Meth. A , (2016) pg.128 – 130[24] P. Martins, R. Dal Bello, M. Seimetz, G. Hermann, T. Kihm, and J. Seco, A single-particle trigger forTime-of-Flight measurements in prompt-gamma imaging. , Frontiers in Physics , (2020) pg. 169[25] Saint Gobain Ceramics & Plastics, Inc, Plastic Scintillating Fibers. Product brochure (2017).[26] X. Chen, O. Allegrini, B. Carlus, C. Caplan, L. Caponetto, J. P. Cachemiche, S. Curtoni,D. Dauvergne, R. Della Negra, M. Fontana, L. Gallin-Martel, M.-L. Gallin-Martel, J. Hérault,D. Lambert, G.-N. Lu, M. Magne, S. Marcatili, H. Mathez, C. Morel, G. Montarou, E. Testa, andY. Zoccarato, A Time-of-Flight gamma camera data acquisition system for hadrontherapymonitoring. , in
IEEE MIC , Manchester, United Kingdom, Oct 2019.[27] J.-P. Cachemiche, P.-Y. Duval, F. Hachon, R. Le Gac, and F. Marin,
Study for the LHCb upgraderead-out board. , J. Instrum. , (2010) pg. C12036[28] S. Deng, D. Dauvergne, G.-N. Lu, H. Mathez, and Y. Zoccarato, Very fast front end ASIC associatedwith multi-anode PMTs for a scintillating-fibre beam hodoscope. , J. Instrum. , (2013) pg. C01047[29] X. Chen, B. Carlus, C. Caplan, L. Caponetto, J.-P. Cachemiche, D. Dauvergne, R. Della-Negra,M. Fontana, L. Gallin-Martel, D. Lambert, G.-N. Lu, M. Magne, H. Mathez, C. Morel, G. Montarou,M. Rodo, E. Testa, and Y. Zoccarato, A data acquisition system for a beam-tagging hodoscope used in – 15 – adrontherapy monitoring. , in
IEEE Nuclear Science Symposium and Medical Imaging Conference(NSS/MIC) , Atlanta, GA, USA, pg. 1–4, Oct 2017.[30] C. Caplan, O. Allegrini, J. . Cachemiche, B. Carlus, X. Chen, D. Dauvergne, R. Della-Negra,M. Fontana, L. Gallin-Martel, M. . Gallin-Martel, J. Hérault, D. Lambert, G. . Lu, M. Magne,H. Mathez, G. Montarou, C. Morel, M. Rodo Bordera, E. Testa, and Y. Zoccarato, A µ TCA back-endfirmware for data acquisition and slow control of the CLaRyS Compton camera. , in
IEEE NuclearScience Symposium and Medical Imaging Conference (NSS/MIC)
Manchester, United Kingdom, pg.1–4, Oct 2019.[31] L. Kelleter, A. Wrońska, J. Besuglow, A. Konefał, K. Laihem, J. Leidner, A. Magiera, K. Parodi,K. Rusiecka, A. Stahl, and T. Tessonnier,
Spectroscopic study of prompt-gamma emission for rangeverification in proton therapy. , Phys. Medica , (2017) pg. 7 – 17[32] M. Fontana. Tests and characterization of gamma cameras for medical applications . PhD thesis,Université de Lyon, 2018.[33] J.F. Ziegler, M.D. Ziegler, and J.P. Biersack,
SRIM – the stopping and range of ions in matter (2010). , Nucl. Instrum. Meth. B , (2010) pg. 1818–1823[34] C. Joram, U. Uwer, B.D. Leverington, T. Kirn, S. Bachmann, R. J. Ekelhof, and J. Müller, LHCbScintillating Fibre Tracker Engineering Design Review Report: Fibres, Mats and Modules.
TechnicalReport, CERN, 2015.[35] R.J. Ekelhof.
Studies for the LHCb SciFi Tracker - Development of Modules from Scintillating Fibresand Tests of their Radiation Hardness.