ARRONAX International Scientific Committee

Nuclear medical imaging using β+γ coincidences from 44Sc radio-nuclide with liquid xenon as detection medium

J.P.Cussonneau∗1, M. Fallot1, M.Fattahi-Vanani1, F. Gauché1, S.Girault1, C.Grignon1, A. Guertin1, F.Haddad∗1, P.Leray1, S.Lupone1, L.Luquin1, V.Métivier1, N.Michel1, E.Morteau1, N.Servagent1,

D.Thers∗1 J.Barbet2, M.Bardiès2, T.Carlier2, J.F.Chatal2, O.Couturier2, A.Faivre∗2, L.Ferrer2

A.Breskin∗3, R.Chechik3 T. Haruyama4

1 SUBATECH, EMN-IN2P3/CNRS-Université, Nantes, F-44307, France 2 INSERM/CHU Nantes, F-44093, France

3 Weizmann Institute, Herl St., 76100 Rehovot, Israel 4 KEK, High Energy Accelerator Research Organization, Oho, Tsukuba, Ibaraki 305-0801, Japan

Abstract

Gamma detection with liquid xenon has been studied at Subatech for two years, with the goal of improving PET (Positron Emission Tomography) imaging and diagnostic. The key idea is to integrate in a liquid xenon proportional chamber recent progresses made with gaseous devices, for detecting both ionization and scintillation. The ionization contribution will be detected by a PIM/Micromegas device, developed at Subatech, immersed in liquid xenon for the first time and the scintillation contribution will be detected by a new type of gaseous photomultiplier, particularly suited for covering large areas, and developed at the Weizmann Institute of Sciences.

Simulations of image characteristics obtained with such a detector have been performed and confirm the interest of a research effort around liquid xenon for PET. They demonstrate the potentiality to reduce injected activity and exposure time in human-body PET diagnostic when considering a large camera containing ~100 litres of liquid xenon. But, the technical difficulties associated to the xenon liquefaction (T ~ -110° C) and purification on a large volume put strong constrains, for which a dedicated development has to be pursued.

We report, in this document, on the construction of a first small prototype adapted for nuclear medical imaging associated to β immunotherapy. The proposed imaging technique is based on the measurement of the emitter location in the 3 dimensions with a few mm spatial precision using β+γ coincidences. It will also allow an accurate dosimetry along the therapy process.

A new generation of camera and an ad hoc radio-nuclide are then the foundations of this proposition: the incident direction of the third γ-ray emitted quasi simultaneously with the β+ decay of the 44Sc radio-nuclide is measured by a liquid xenon Compton telescope (LXeComp).

The innovative LXeComp/44Sc couple should be particularly interesting when associated to research in β immunotherapy/dosimetry with 47Sc/44mSc. We review in this report the major ideas initiating the experimental feasibility of this novel medical imaging technique. After this demonstration step, we foresee to submit in two years a new research program for the construction of a second prototype based on a larger liquid xenon Compton camera dedicated to the small animal imaging. ∗ Simulation corresponding author, Phone: +33 2 51 85 84 31, E-mail: jean-pierre.cussonneau@subatech.in2p3.fr ∗ Scandium production corresponding author, Phone: +33 2 51 85 84 67, E-mail: ferid.haddad@subatech.in2p3.fr ∗ Xenon corresponding author, Phone: +33 2 51 85 84 03, E-mail: dominique.thers@subatech.in2p3.fr ∗ RadioImmunotherapy corresponding author, Phone: +33 2 40 08 47 18, E-Mail: afaivre@nantes.inserm.fr ∗ Gaseous photomultiplier corresponding author, Phone :00972-8-9342645, E-mail: Fnbresk@wisemail.weizmann.ac.il

- 1 -

1. Interest of the Scandium radio-nuclides

1.1. Therapy with 47Sc

The reasons for developing scandium 47 labeled monoclonal antibodies lie in the fact that its decay half-life is close to that of antibody clearance in blood in humans. Moreover the emitted β radiations have an energy close to that of lutetium 177 [ref i], already used in targeted radio-nuclide therapy with somatostatin analogues. Despite these attractive dosimetry characteristics [ref ii] (table 1), scandium 47 has not yet been used in clinical practice. This may be explained, at least in part, by the limited availability of that radio-nuclide, since, to our knowledge, the Brookaven National Laboratory is the only source of production [ref iii].

Several publications from this laboratory describe the potential of scandium 47 in targeted radio-nuclide therapy, mentioning that its reactivity is close to that of yttrium 90. Several bifunctional chelating agents (BCA) have been described and tested, allowing for monoclonal antibodies labeling with scandium: MX-DTPA, DTPA, NHS-DOTA and 4-ICE. All are commercially available or can be easily synthesized [ref iv,v,vi,vii].

The tumor targeting group of Inserm U601 has a broad experience in antibody labeling procedures, including the production and use of BCAs, the evaluation of radiolabeling yields and experimental targeted radio-nuclide therapy.

Regarding potential clinical investigations, Inserm U601 works closely with the Nantes Cancer Center and the Nantes University Hospital, where clinical trials are being carried out.

The tumor targeting group of Inserm U601 has developed a software (called OEDIPE) that allows for patient-specific dosimetry. Based on Monte Carlo (MCNPX) simulation of radiation transport and deposition, it provides a way to evaluate the delivered absorbed doses during targeted radio-nuclide therapy clinical trials [ref viii]. Patient-specific data are obtained using both anatomical (CT) and functional (SPECT or PET) imaging procedures.

1.2. 47Sc imaging and dosimetry

Scandium 47 emits a gamma radiation (159 keV, 69%) that makes it a suitable isotope for SPECT imaging. However, current gamma camera spatial resolution and quantitative imaging procedures in SPECT do not allow for the level of accuracy desirable for dosimetry. This is why PET quantitative imaging is now preferred whenever possible. Some examples associating beta minus (therapy) and plus (quantitative imaging) isotopes have been proposed in the literature, thus defining ‘isotopes pairs’: 131I-124I, 90Y-86Y, 67Cu-64Cu, … Even though dosimetry can be performed based on 47Sc SPECT quantitative imaging, it is therefore appealing to use a positron-emitting isotope of that nuclide. A survey of isotope charts indicates that scandium 44 could be used for quantitative PET imaging as a surrogate for scandium 47. 1.3. 44Sc for PET quantitative imaging

The 44Sc β+emitter has a short period close to 4 hours, too short for imaging processes when compared to biological period considered in targeted radio-nuclide therapy. But the suitable period (2.44 days) of the 44mSc isomeric state of 44Sc will overcome this drawback. Indeed, the γ decay of 44mSc produces the 44Sc β+emitter used for imaging, thus 44mSc can be considered as an in vivo 44Sc generator. It should be noticed that the recoil energy (0.89 eV) transferred to the 44Sc nucleus during the γ decay (E=270 keV) of 44mSc is too low to break any chemical bond.

- 2 -

Quantitative imaging of 44Sc would allow for 47Sc-labeled radiopharmaceutical kinetics to be estimated within the duration of therapy. 44Sc is thus a β+γ emitter which offers unique capability for quantitative imaging and dosimetry when associated to LXe-based detector. This novel medical imaging technique will be extensively described in section 3 and constitutes the main idea of the present report.

- 3 -

2. Scandium production

The production of 47Sc as well as 44Sc/44mSc at the Nantes Cyclotron has been investigated. The main characteristics of the radio-nuclides are summarized in table 1.

44Sc 44mSc 47Sc T1/2 3.97 h 2.44 days 3.35 days

End point Eβ

(MeV) 1.4743 : 94.27 % - 0.44 : 68.4 %

0.60 : 31.6 % γ-ray Energy (MeV) 1.157 : 99.9 %

τ = 2.6 ps 0.270 : 100% 0.159 : 69%

Production reaction 44Ca(p,n) 44Ca(p,n) 48Ti(p,2p)

Table 1: Properties of 47Sc/44mSc/44Sc and production reaction with proton beam It should be notice that the characteristics of the γ-ray associated to the 44Sc and to the 47Sc are

the one of the respective son nucleus : 44Ca and 47Ti. The key parameter for the present report is the very short life time of the 44Ca excited state, which allows a quite perfect time coincidence between the β+ decay and the γ-ray emission. In the following sections, the presented data are extracted from http://www.nndc.bnl.gov/exfor/exfor.html

2.1. 47Sc production Two different reaction channels can be used to produce the 47Sc radio-nuclide:

- A first possibility is to use 47Ti target to produce 47Sc near a nuclear reactor thanks to the high flux of fast neutrons inducing the following 47Ti(n,p)47Sc reaction.

- An alternative solution is to produce 47Sc with a 48Ti target via the 48Ti(p,2p)47Sc reaction using a high intensity proton beam delivered by a cyclotron.

Figure 1: Production cross sections on 48Ti target: Triangles (p,2p)47Sc, Crosses and Squares (p,2n) 47V, Circles (p,n)48V.

- 4 -

The characteristics of the Nantes cyclotron proton beam in term of intensity (350 µA) and available energy (up to 70 MeV), make this solution competitive with the fast neutron induced production near reactor. In order to be more selective in the 47Sc production and to avoid contamination from other (p,xn) channels at low energy, a proton beam energy greater than 25 MeV seems to be a good choice. Within this energy domain, the 48Ti(p,2p)47Sc reaction cross section can be considered as constant with a value equal to 20 mb. The cross sections of the different reactions are presented on figure 1.

2.2. 44Sc/44mSc production

Figure 2: Production Cross section on 44Ca target: blue (p,n)44Sc, green (p,2n)43Sc, red (p,2p)43Kr.

The 44Sc production could be done either in the 44Ca(p,n) or in the 47Ti(p,α) channel. The

cross section of the first one (figure 2) is more favourable and allows to limit the contamination from other elements if the beam energy is chosen in the 4.54 MeV-14 MeV range. From the isotopic distribution of natural Calcium it appears that a 44Ca enriched target should be used for 44Sc production.

E (MeV) 6. 9. 12. 15. 18 21 24 30 40 50 60 72 85 44mSc/44Sc 0.02 0.06 0.09 0.14 0.20 0.24 0.23 0.19 0.16 0.14 0.15 0.16 0.17

Table 2: Fraction of 44mSc produced in 44Ca(p,n) reactions as a function of the beam energy It should be stressed that in parallel to the 44Sc production the interesting isomeric state 44mSc is also produced at the level of 10 % for typical beam energy of 14 MeV (table 2). The 44mSc/44Sc fraction is increasing for higher beam energy but the production cross section of 44Sc is dropping while the one of other undesirable radio-nuclides as 43Sc and 43Kr is increasing. Hence, the beam energy could not be too high even if the increase of 44mSc production is of particular interest for our project.

- 5 -

Table 3: Fraction of 44mSc produced in 41K(α,n) reactions as a function of the beam energy

E (MeV) 6. 11. 16. 19. 22 24.5 25 26.5 28 32 34.5 40 44mSc/44Sc 0.3 0.34 0.75 1.1 1.55 1.6 2 1.9 1.3 1.12 1 0.9

Moreover, with α particles as projectile and with 41K target, the fraction of 44mSc can be drastically increased due to the high angular momentum transferred (table 3 and figure 3).

Figure 3: Cross sections for the 41K(α,n)44Sc (green) production and for the 41K(α,n)44mSc

(blue) production.

- 6 -

3. β+γ imaging with liquid xenon as Compton telescope

3.1. Principles and concepts for β+γ coincidences detection and reconstruction The principle of this nuclear medical imaging method has been already proposed by J.D.

Kurfess et al. [Ref ix]. The basic idea is to use a radio-nuclide which emits a positron accompanied by a quasi simultaneous γ-ray to measure event by event the three coordinates of the radio-nuclide decay position. The desirable properties of the new radio-nuclide are: a low energy β+, and a third γ-ray with around 1 MeV energy emitted in coincidence in most of the β+ decay (eg 44Sc). The 3-dimensionnal localization of the emission point is realized by the intersection of the direction cone of the γ-ray with the reconstructed line of response of the positron annihilation. This technique has the advantage of a direct 3-dimentionnal imaging with a lower activity and a comparable resolution in view of standard PET technique. Moreover the image quality is less sensitive to low statistic data samples since complex reconstruction algorithms as in PET are no more needed. This means that this imaging technique is particularly adapted to low activity applications and to dynamical imaging. Figure 4: Principle of β+ γ coincidence measurement. The localization of the emission point is determined as the intersection of the Line Of Response (LOR) with the reconstructed direction cone from multiple Coulomb interaction. It assumes that for each interaction of the coincident γ-ray in the detection medium the 3 coordinates and the deposited energy are measured.

E0

Ee 3rd γ

LOR β+

Lθ

∆L

The cone direction of the emitted γ-ray is obtained from the measurement of the three coordinates of the first Compton interaction and the three coordinates of the second Compton or photoelectric interaction in the position-sensitive detector. The cone aperture θ is derived from the measured energy E e of the first Compton interaction and the known energy E0 of the incident γ-ray:

)(1cos00

2

e

e

EEEEcm −−=θ

Where mc2 is the rest mass energy of an electron,

The achievable spatial resolution ∆L (Figure. 1) on the position of the radio-nuclide decay will be limited by both the precision on the coordinate’s measurements and by the energy resolution of

- 7 -

the detector. Figure 5 shows the angular uncertainties than can be achieved for the 1.157 MeV γ-ray from the 44Sc emitter. The spatial resolution ∆L is then proportional to the distance of the cone apex to the emission point. Typically, if only small scatter angle are considered, below 90 degrees, a spatial resolution of ~2 mm can be reached for a detector located at a distance of 10 cm from the source.

Figure 5: Angular resolution with respect to the Compton scatter angle for 1.157 MeV γ-rays. The detector has 4% FWHM energy resolution @ 1 MeV and 100 µm spatial resolution. The distance between the two interaction points is taken to be 1 cm as a typical value. The effect of the Doppler broadening is not represented here, but its contribution is getting bigger at large scatter angle where the angular resolution is dominated by the energy resolution.

3.2. Liquid xenon prototype for β+ γ coincidences measurement

Our proposed approach consists of a Liquid Xenon (LXe) based detector coupled to large-area fast gas-avalanche imaging photomultipliers (GPM). The properties of LXe, relevant to PET, are compared in table 4 with those of common crystals. The relative light outputs are expressed as a fraction of the NaI(Tl) yield (for LXe, the relative light output and the ionization yield are given in presence of an electrical field of 2 kV/cm).

Scitillation materials NaI BGO LSO GSO LXe

Effective atomic number 50 73 65 58 54Density (kg/l) 3.7 7.1 7.4 6.7 3.0Relative light output (%) 100 15 45-7 20-40 25Decay time (ns) 230 300 40 60 2.2Ionisation Yield (per MeV) / / / / ~ 60000

Table 4. Comparison between the properties of common crystals and liquid xenon

Lavoie [ref x] was the first to point out the potential applicability of liquid xenon for PET; he showed that measuring both, scintillation photons and ionization, it is possible to accurately determine the three coordinates of the incident photons. A team at Coimbra [ref xi] designed a first prototype of LXe detector for PET, combining ionization and scintillation recording. They used the scintillation signal, measured with a vacuum photomultiplier tube, to trigger the data acquisition system and the ionization signal to measure the position and the energy of the γ. As compared to

- 8 -

scintillation-crystal systems (BGO and LSO blocks), their LXe detector yielded better time (1.3 ns), spatial (0.8 mm) and energy (80 keV) resolutions [ref xii]. Developments taking benefit of the very fast response of the LXe scintillation are under consideration for TOF-PET [ref xiii]; time resolutions of 650 ps are reported with PMTs.

The possibility of detecting ionization and scintillation light in liquid noble gases with gas-avalanche photomultipliers has been investigated by Peskov et al. [Refxiv] for nTOF and ICARUS experiments and Bondar et al. [Refxv] for dark matter and solar neutrino detectors. For that purpose they studied different solutions of gaseous photon detectors like wire chambers, capillary plates and GEMs. Bondar et al. proved that GEM photon detectors could operate at cryogenic temperatures, as required by LXe cameras.

The first prototype :

The concept of the proposed LXe γ camera prototype, combining a LXe converter with a gas-avalanche imaging photomultiplier (GPM) is shown in figure 6 and figure 7.

Figure 6: Active volume of the liquid xenon Compton prototype

M

3rd γ

Internal cryostat

Following a γ conversion, both ionization and scintillation sig

ionization signal of drifting electrons is recorded by the anode, after microthe UV photons resulting from LXe scintillation are detected in the GPassembled in a stainless-steel vessel with a dedicated pulse tube cryoc[refxvi], filled with 100 cl of high purity xenon; it will be pumped to high vparts will be made of ultra clean materials, and will be commissioned winstitute. Therefore, the prototype design will fully benefit from the estknow-how in LXe purification, liquefaction and handling. There has beenthis field, mostly directed towards the search for Dark matter or neutrin[refxvii], XMASS [refxviii], MEG [refxix, refxx], ZEPPLIN [refxxi], XENON [refxxiv].

- 9 -

GPM

icromesh frame

s

Field rings

Entrance window

Active liquid xenon volume

in construction

nals are collected. The mesh “Frish-Grid”, while M. The camera will be

ooler developed at KEK acuum prior to filling; all ith the help of the KEK ablished knowledge and long ongoing activity in os physics like in: EXO [refxxii, refxxiii], LXe-Grit

External cryostat

Pulse tube cryocooler

Internal cryostat

s

Figure 7: External view of the liquid xenon Compton prototype in construccryostat has not been specially designed for this application

When a γ-photon interacts with a liquefied rare gas, it gives rise to an excited species. At 511 keV, photoelectric absorption and Compton scatteringprocesses occurring (22% photoelectric, 78% Compton with xenon). Photoelecto the apparition of one photoelectron that ionizes the liquid along a short dis511 keV photoelectron). During its stopping process, secondary electron/ion paphotoelectron trajectory are dissociated, allowing for the localization of the phothe measurement of the energy loss. For each Compton scattering, an electraccompanied by a lower energy γ, leading to a sequence of Compton verphotoelectric interaction of the γ with the liquid.

The localization of the photoelectric and Compton vertices in the liquidby a “time projection chamber” (TPC) read-out: the conversion depth is determdrift time and the two other dimensions by the anode sampling.

Under an electrical field, a fraction of the ionization signal escapes reelectrons drift in LXe with velocities of ~2.2 mm/µs at 2 kV/cm; the ions remaAs the electron diffusion in liquid is low, the charge induced by their motion cposition resolution. Moreover, as liquid rare gases have high free electron factor, good energy resolutions are reported with the induced-charge meaaddition, LXe is a very good scintillator, emitting strongly VUV at 179 nmdecay-time. Hence, the scintillating light can provide the missing information inthe absolute time of the γ interaction. It will provide a very fast signal for coinfor the data acquisition system.

- 10 -

Entrance window

Isolating vacuum

tion (the external ).

electron-ion pair and are the predominant tric interaction leads tance (half a mm for irs created along the toelectric vertex and on is emitted and is tices, up to the last

volume is provided ined by the electron

combination and the in almost stationary. an provide excellent yield and low Fano surement [refxxv]. In , with a short 2.2ns the induced charge, cidences and trigger

Moreover, in a classical PET system, simultaneous detection of back-to-back γ rays sign the presence of the β+ annihilation somewhere on the line (LOR: Line Of Response) between the two fired detectors (scatter and random events unconsidered). Time-of-Flight (TOF) PET systems measure also the difference between the two arrival times in order to determine the position of the disintegration along the LOR. The time resolution of the detector is then the crucial point, which allows to reduce the length of the LOR, reducing then the PET exposure time. Nevertheless, such considerations have to be moderated by the very good time resolution required, 100 ps for 1.5 cm spatial resolution along the LOR. Facing this problem, TOF-PET systems achieve today time resolution of 350 ps with BaF2 [ref xxvi], 500 ps with LSO [ref xxvii] and 650 ps with LXe [ref16]; but these pure scintillation systems suffer from an additional difficulty coming from the multiple Compton interactions within the same volume element, and then to a non- negligible bias on the absolute time measurement. We propose in our prototype to show the feasibility of a LXe device using ionization signal to localize and identify Compton vertices and then to improve the time measurement from the bias met with pure scintillation detectors. This objective seems to be realistic regarding both the fast and large scintillation of LXe and the fast response reached with the GPM technology.

In the proposed prototype, the distance between the cathode and the micromesh is 12 cm, defining the depth of the converter. The distance between the micromesh and the anode is 50 µm, providing an electromagnetic shielding efficiency better than 2% (evaluated from Bunemann’s formula [refxxviii]) in regards of the slow drift of ions; the grid is copper micromesh of 3 µm thickness and 50µm pitch.

Negative high voltage is applied to the cathode and the micromesh to set a uniform electric field across the drift region. Scintillation photons will be reflected from the walls of a PTFE-metal assembled drift column (with a resistive voltage divider), with about 95% diffuse reflectivity at 179 nm [refxxix].

The sensitive area of the GPM electrodes is 3x3 cm2. In our design, the primary UV photons are detected by the GPM placed above the liquid-gas interface. The GPM combines a CsI photocathode, sensitive to the LXe scintillation (QE = 30% @ 170 nm [Refxxx]), and a gas-avalanche electron multiplier. The gas filling will be chosen for high-gain and low-temperature operation conditions. We will investigate a few GPM concepts: a multi-GEM (Gas Electron Multiplier [Refxxxi]) [Refxxxii], the newly conceived Thick GEM-like multiplier (T-GEM [Refxxxiii]), both developed at the Weizmann Institute, and the PIM (Parallel Ionization Multiplier) [refxxxiv], developed at SUBATECH. We abandoned the solution of a windowless GPM operating in the Xe gas phase for a few reasons: it can operate only in a horizontal position above the LXe; The multiplication factor in pure LXe (not yet measured) should be limited to about 104 -105 even with a multi-GEM GPM [Refxxxv]; the effective QE of the CsI photocathode will be very low due to photoelectron backscattering in noble gases [Refxxxvi].

Gas photomultipliers The PIM electron multiplier has been developed at SUBATECH for high resolution β autoradiography [refxxxvii]. It is made of a thin sandwich of two metallic micromeshes separated by 100 microns with an appropriate insulating spacer (see figure.8). The accurate and robust thin gap and the very high electric fields permit reaching gas multiplication factors closed to 5.105 at atmospheric pressure with Ne/CO2 gas mixture and 5.9 keV X-rays [refxxxviii].

- 11 -

Figure 8. A microscope view of a copper micromesh produced by laser machining: holes are 14 µm diameterwith a pitch of 40 µm. The concept of a PIM photon detector with a reflective photocathode.

Reflective photocathode

Ampli 1

Ampli 2 Multi-GEM GPMs were proposed and extensively investigated by the WIS team and yielded very good performance both with CsI UV photocathode and with visible-range bialkali photocathode. The GEM (see figure.9) is a densely perforated metalized dielectric (KAPTON), about 50 microns thick, with typically 50 micron diameter holes, 200 micron spaced [Ref43]. Gains of the order of 103 to 107 are obtained in single GEM electrodes and in 3-4 cascaded GEMs, respectively [Refxxxix][Refxl]. Multi-GEM GPMs with a CsI photocathode deposited on the top GEM surface (~80% coverage) operate in a stable way and are sensitive to single photons. Time resolutions in the 300ps range were recorded with a ~ 200 photons and single photons were localized with 150 microns resolution.

Figure 9. A microscope view of a Gas Electron Multiplier (GEM), produced of 50 micron Cu-cladded Kapton, perforated with 80 micron diameter holes; equi-potentials, electric fields and avalanche multiplication scheme occurring when the GEM is polarized by a few hundred volts; The concept of a multi-GEM photon detector, with a reflective photocathode.

- 12 -

The TGEM (Thick-GEM) is a ten-fold dimensionally expanded version of the standard GEM (figure. 10). It is produced by standard printed-board technologies, where a dense array of small, mm-size holes is mechanically perforated in a millimetre-scale thick copper claded G-10 plate. Avalanche multiplication occurs upon the application of a few hundred volts across these millimetric gas volumes. The multiplication within the holes prevents secondary effects, which yields gains for singe electrons of the order of 105-107 [Refxli] in single and double-TGEM elements, respectively. The TGEM electron multiplier has a fast response, yielding current pulses of a few ns rise-time. The efficient extraction of photoelectrons from a CsI photocathode deposited directly on top of the front TGEM face was recently demonstrated.

Figure 10. A scheme and a photograph of a TGEM

- 13 -

4. Outlooks for small animal imaging

4.1 Simulation framework A Monte Carlo simulation of a LXeComp module has been carried out to evaluate the

physical performances of the β+γ imaging technique. The Monte Carlo used is Geant3.21 which has been interfaced to ROOT (http://root.cern.ch) by the ALICE collaboration at CERN (http://AliSoft.cern.ch/offline). The detector response has been included in the simulation code and the main parameters of the liquid xenon medium are the following:

• The mean energy needed to create an electron/ion pair in LXe is taken to be Wi = 15.6 eV. 10% of the produced ionization electrons are subjected to the recombination for a 2 kV/cm drift electric field. The LXe purity is assumed to be perfect and then electrons are drifting along the 12 cm depth cell without any captures.

• An uncertainty on the drift time measurement of σt = 10 ns was considered and leads to a good spatial resolution of σz = 22 µm as the drift velocity is known to be 2.2 mm/µs for a 2 kV/cm electric field .

• The transverse diffusion of the electrons moving toward the anode readout plane is taken to be 170 µm per root of centimeter of drift length.

• The readout anode plane is segmented in 5.05.0 × mm2 pads multiplexed in strips using a “checkerboard” readout scheme.

• The intrinsic energy resolution is expected to be E(MeV)0.04 (FWHM). • An amplification of the ionization electrons within the micro gap of a factor 5 is assumed. • The electronic noise on each readout strip is assumed to be 300 e- (RMS). Such a noise level

can be reasonably reached with existing amplifiers. A zero suppression applied to the ionization signal on each readout channel leads to a 20 keV threshold on the energy deposited by electrons.

• The mean energy necessary to create a VUV photon is taken to be Wph = 21.4 eV. The light production is 35 % of the maximum for a 2 kV/cm electric field. The photons are reflected by the PTFE walls according to a Lambert law with a refraction coefficient of 0.95. The quantum efficiency of the Hamamatsu PMT photo cathode is 0.25.

• The detection inefficiency of the 1.157 MeV γ−rays due to the light collection has been estimated with the simulation and is less than 1% depending on the electron noise of the PMT considered.

Spatial and energy resolutions characterizing the measurement of created vertices inside the

liquid xenon obtained from the simulation with 1.157 MeV incident photons are shown on figures 11 and 12.

- 14 -

Figure 11: Spatial resolution as a function of the energy deposited by electrons in liquid xenon

obtained from simulation, in the anode plane (left) and along the drift direction (right).

Figure 12: Energy resolution as a function of the energy deposited by electrons in liquid xenon

obtained from simulation.

The reconstruction algorithm starts from all the possible pairs of measured vertices whatever the number of interaction points is. A candidate pair is accepted when the calculated cone intercepts the β+ LOR inside the volume defined by the imaged object. If several pairs are accepted in the same event then a choice is made by selecting the track associated to the pair which has the best χ2. This χ2 is calculated from the Compton interaction sequence by testing the compatibility of the spatial measurements with the theoretically known Compton kinematics.

Ei (γ)

Ei (e-)

Ei-1 (γ) (511 keV) θi

- 15 -

This is achieved by calculating the following χ2 :

∑−=

=

−=

1

1 2

2).cos.(cos2 Ni

ii

measi

calci

σ

θθχ

1

. 111cos−

+−=ii

calci WWθ

2)(

mciE

iWγ

=

2.cos

2.cos

2measi

calcii θθ

σσσ +=

Where cosθi

calc is calculated from the Compton kinematics which depends on the energy measurement and cosθi

meas is derived from the geometric information given by the position measurements. N is the number of interaction vertices.

At the end the best pair is selected and the cone apex is defined as the position of the first vertex, the cone direction is defined by the first two points of the track and the aperture angle is calculated according to the formula presented in the paragraph 3.1 using the measured Compton electron energy Ee . Then the intersection point coordinates of the β+ LOR with the cone surface are calculated. If there is only one valid intersection i.e. inside the volume defined by the imaged object then the event is accepted, it is rejected otherwise.

4.2 Compton telescope associated to a microPET

The first simulated geometry which corresponds to the proposition of adding a Compton telescope device to an existing microPET system is shown on figure 13. To evaluate some basic characteristics of this LXeComp device as for instance the angular resolution or the detection sensitivity, the emission of β+ from 44Sc in coincidence with 1.157 MeV γ−ray at the centre of the PET camera has been considered. The detector of size 24x24x12 cm3 is made of individual cells of 3x3x12 cm3 separated by 500µm thick Teflon (PTFE) walls. The depth of the liquid xenon module has been chosen to be 12 cm to efficiently detect 1.157 MeV γ−ray, of which the mean free path in liquid xenon is 5.8 cm. The detector segmentation is mandatory in order to limit the occupancy due to the combination of the high interaction rate and the maximum detection dead time of 54.5 µs corresponding to the drift of electrons along the 12 cm conversion area. Hence, the maximum acceptable injected activity will be mainly limited by this detection dead time; it has been estimated to 0.5 MBq in the field of view. To improve it, it might be possible to recover part of the pile up events with the help of an adapted geometry for the lXeComp module containing more than one anode read-out plane. Such segmentation should permit to support up to ~5 MBq in the field of view. Anyway, this possibility will be considered after validation of the present design containing only one read-out plane; it seems to be relevant in view of the injected activity used for mouse, rat or monkey to make images with available micro PET camera.

- 16 -

LXeComp

Figure 13: Liquid xenoncm3) is made of individurat phantom is a water c

The angular reso

expected the resolution degrees has been done toadded, as a minimum dithe simulation, the sensileading to an overall sen

Figure 14: Overall anCompton scatter angle o In this conditiontails in the residual distralong the LOR directiontwo other directions by precision along the LOR20 ps (FWHM), value m

LSO

module associated to LSO microPET camera (left). The module (24x24x12 al cells of 3x3x12 cm3 separated by 500µm thick PTFE walls (right). The ylinder of 6 cm diameter and 15 cm length.

lution obtained from this simulation is shown on figure 14. As it was is worsening for large Compton scattering angles. An angular cut at 90 keep the resolution below ~3 degrees for all events. Other cuts have been

stance of 4 mm between the two vertices defining the cone direction. From tivity to detect the third γ-ray of the proposed liquid xenon module is 3.4 %, sitivity of 0.14 % including the LSO camera to measure the 511 keV γ-rays.

gular residual distribution and angular resolution with respect to the btained from simulations.

s, the simulated overall angular resolution is ~ 1.25° when eliminating the ibution (corresponding to scattered events), leading to a spatial resolution of ~2.5 mm. This spatial resolution is close to the ones estimated in the

the LSO camera with the 511 keV LOR photons. Notice that to reach such direction, Time Of Flight PET cameras might reach a time resolution of ~ ore than 10 times better than the present time resolutions measured by the

- 17 -

best TOF PET cameras (~ 300 ps). This is the major benefit of the β+γ imaging technique associated to a liquid xenon Compton telescope.

4.3 Full liquid xenon for β+ γ imaging

In order to evaluate the characteristics of a full liquid xenon camera for the β+γ imaging technique, we proceeded to a simulation which should be the second step of the LXeComp project. In this approach, liquid xenon is used to detect both 511 keV and 1157 keV photons. The camera will contain 4 identical modules (24x12x12 cm3) disposed around the field of view as showed on figure 15, defining an active liquid xenon volume of 13.8 l.

Figure 15: Full liquid xenon microPET camera. Each module (24x12x12 cm3) is made of individual cells of 3x3x12 cm3 separated by 500µm thick PTFE wall. The rat phantom is the same as in previous simulation.

Each module is segmented in individual cells of 3x3x12 cm3 separated by 500µm thick

PTFE wall. The performances of the present device have been evaluated in the case of a rat phantom as in the previous setup with a 44Sc source localized at the center of the volume. The measured energy spectrum including the effect of the scatter photons in the phantom volume is shown on figure 16. The obtained energy resolution is 4.3 % FWHM at 1157 keV and 5.6 % FWHM at 511 keV.

Figure 16: Energy spectrum measured in the LXeComp modules. The energy peaks from the 511 keV annihilation γ-rays and the third 1157 keV γ-rays are clearly identified. The continuum is coming from partial calorimetrisation in the detector and from the scatter events in water.

- 18 -

The position of each β+γ correlated events has been reconstructed according to the algorithm presented on paragraph 4.1. The obtained image is shown on figure 17. The spatial resolution deduced from the projection of the transverse image slice on a single coordinate is 2.3 mm.

Figure 17: Transverse slice of the reconstructed image of a point source with a voxel size of 2x2x2 mm3 (left) and its projection on one x axis (right). Results have been obtained with 0.5 millions of generated 44Sc decays.

The overall reconstruction efficiency for the β+γ events is 1.3 %, quite 10 times higher than for the case where the liquid xenon is only used as a Compton telescope interfaced with a LSO camera. A full liquid xenon β+γ camera constitutes then an ideal solution for detecting precisely the location of the 44Sc decays with a good sensitivity. Hence compared to the present state of the art in functional imaging, an equivalent diagnostic should be done faster and with less activity. 5. Questions to the ARRONAX International Scientific Committee

Does the present project fit with the research topics of the future Nantes Cyclotron ? Will the productions of scandium radio nuclides 47Sc, 44mSc and 44Sc be encouraged by the Arronax Cyclotron ? Will the liquid xenon Compton telescope be an opportunity for future imaging/Radioimmunotherapy research near the Arronax Cyclotron ?

i D.A. Weber, K.F. Eckerman, L.T. Dillman and J.C. Ryman, MIRD: Radionuclide data and decay schemes, The Society of Nuclear Medicine: New York, 1989 ii M. Bardiès and J.F. Chatal, Absorbed doses for internal radiotherapy from 22 beta-emitting radionuclides: beta dosimetry of small spheres, Phys. Med. Biol., 39(1994) 961 iii L.F. Mausner, K.L. Kolsky, V. Joshi, and S.C. Srivastava, Radionuclide development at BNL for nuclear medecine therapy, Appl. Radiat. Isot., 49(1998) 285 iv K.L. Kolsky, V. Joshi, L.F. Mausner and S.C. Srivastava, Radiochemical purification of no-carrier-added scandium-47 for radioimmunotherapy, Appl. Radiat. Isot., 49(1998) 1541 v L.F. Mausner, K.L. Kolsky, V. Joshi, G.E. Meinken, R.C. Mease and S.C. Srivastava, Evaluation of chelating agents for radioimmunotherapy with scandium-47, Nucl. Med. Biol., 36(1995) 104

- 19 -

vi S.C. Srivastava and R C. Mease, Progress in research on ligands, nuclides and techniques for labeling monoclonal antibodies, Nucl. Med. Biol., 18(1991) 589 vii M.P. Sweet, R. C. Mease, S.C. Srivastava, J.F. Gestin, G.E. Meinken, V. Joshi, J.F. Chatal, T. Kieber-Emmons and Z. Steplewski, New synthesis of 4-amino-trans-1,2 diaminocyclohexane-N,N,N’,N’ tetraacetic acid (4-ICE), conversion ti bifunctional chelatinfg agents and evaluation of their 111In and 57Co labeled immunoconjugates, Bioconjugate chem., 35(1997) 1047 viii S. Chiavassa, M. Bardiès, F. Guiraud-Vitaux, D. Bruel, J.R. Jourdain, D. Franck and I. Aubineau-Lanièce, OEDIPE: a personalised dosimetric tool associating voxel-based models with MCNPX, Cancer Biother. Radiopharm. 20(2005) 325 ix J.D.Kurfess et al., NIM A 2003, 505, 256-264. x L. Lavoie, Med. Phys. 3 (1976) 283-293. xi V. Chepel et al., Development of liquid xenon detectors for medical imaging, submitted to World Scientific (28-11-02). xii V. Solovov et al., Nucl. Inst. Meth. A477 (2002) 184-190 xiii F. Nishikido et al., Jap. Jour. of Appl. Phys. 43 (2004) 779-784. xiv L. Periale et al., Nucl. Inst. Meth. A478 (2002) 377-383. xv A. Buzulutskov et al., IEEE Trans. On Nucl. Sci. 50(2003) 2491-2493 xvi T. Haruyama et al., Development of a high-power coaxial pulse tube refrigerator for a liquid xenon calorimeter. Adv. Cryog. Eng. 49 (2004) 1459-1466 xvii M. Danilov et al. Phys. Lett. B 480, 12 (2000) xviii Y. Takeuchi, ICHEP04 in Beijing xix S. Mihara, Nucl. Inst. Meth. A518(2004) 45-48 xx T. Haruyama et al., Cryogenic performance of a 120 L liquid xenon photon calorimeter. Proc. of ICEC 19 (2003) 613-616 xxi H.M Araujo et al., Nucl. Inst. Meth. A521 (2004) 407-415 xxii E. Aprile et al., IEEE Trans. Nucl. Sci.50 (2003) 1303-1308 xxiii T, Haruyama et al., High-power pulse tube cryocooler for liquid xenon particle detectors. Cryocooler 13 (2005) 689-694 xxiv E. Aprile et al., arXiv:astro-ph/0212005 v2 4Dec 2002. xxv E. Aprile et al., Nucl. Inst. Meth. A480 (2002) 636-650 xxvi T. Bäck et al., Nucl. Inst. Meth. A477 (2002) 82. xxvii W.W. Moses and S.E. Derenzo, IEEE Trans. Nucl. Sci. 46(1999) 474. xxviii O. Bunemann, T.E Cranshaw et al., Can. J. Res. 27 (1947) 191. xxix M. Yamashita et al., Astropart. Phys. 20 (2003) 79-84. xxx A. Breskin, CsI UV photocathodes: history and mystery. Nucl. Instrum. and Meth. A371 (1996) 116. xxxi F.Sauli, GEM: a new concept for electron amplification in gas detectors, Nucl. Instr. and Meth. A 386 (1997) 531. xxxii A. Breskin et Al., Recent advances in gaseous imaging photomultipliers. Nucl. Instr. Meth. A513 (2003) 250 and references therein. xxxiii R. Chechik et al., Thick GEM-like hole multipliers: properties and possible applications. Nucl. Instr. Meth. A, in press. xxxiv D.Thers et al., Nucl. Inst. Meth. A 504 (2003) 161-165. xxxv A. Buzulutskov et al., The GEM photomultiplier operated with noble gas mixtures. Nucl. Instrum. and Meth. A443 (2000)164. xxxvi T.H.V.T Dias et al., J. Phys. D: Appl.Phys. 37(2004) 540-549. xxxvii J. Samarati et al.,. Nucl. Instr. Meth. A534 (2004) 550-553. xxxviii D. Thers et al., Nucl. Instr. Meth. A534 (2004) 562-565. xxxix D. Mörmann et al.,GEM-based gaseous photomultipliers for UV and visible photon imaging. Nucl. Instr. Meth. A504 (2003) 93 xl D. Mörmann et al., Operation principles and properties of the multi-GEM gaseous photomultiplier with reflective photocathode. Nucl. Instr. and Meth. A530 (2004) 258 xli R.Chechik et al., Progress in the TGEM. to be published.

- 20 -