ARTICLE IN PRESS
Nuclear Instruments and Methods in Physics Research A 525 (2004) 263–267
A PET scanner employing CsI films as photocathode F. Garibaldia,*, E. Cisbania, F. Cusannob, S. Colillia, R. Fratonia, F. Giuliania, M. Griciaa, M. Iodicec, M. Lucentinia, F. Santavenerea, G.M. Urciuolib, G. De Cataldod, R. De Leod, L. Lagambad, S. Marroned, E. Nappid, C. Coluzzab, V. Peskove, R. Panif, R. Pellegrinif, M.N. Cintif a
b
Laboratory of Physics, ISS, Rome, Italy INFN/Roma1 gr. Sanita, Inst. Superiore de Sanita, Viale Regina Elena 299, I-00161 Roma, Italy c INFN/RomaTre, I-00161 Roma, Italy d INFN/Bari, I-00161 Roma, Italy e KHT Physics Deptartment, Physics Center, Stockholm, Sweden f Department of Experimental Medicine, University of Rome La Sapienza, Rome, Italy
Abstract Medical imaging is a fundamental tool in the diagnosis, treatments, and monitoring of disease processes as cancer. Detectors of large area and high Field Of View are necessary to scan the whole body in a reasonable time. Relatively large area photodetectors are necessary even for imaging of small mice and rats with high sensitivities and spatial resolutions, generally obtained by using pinhole or multipinhole collimators. Standard PET scanners, with scintillators coupled to photomultipliers, have generally a limited detector area due to the high costs of both scintillators and photomultipliers. In this respect, the replacement of photomultipliers with gaseous photodetectors represents a possible solution of the problem and brings the additional advantage to provide devices with sensitive areas free from dead regions. In this paper we report on a PET scanner equipped with a multiwire proportional chamber with a CsI thin film as photoconverter. A similar approach has already been successfully pursued in nuclear and particle physics experiments. A prototype of such a PET detector has been designed and built, and will be tested soon. Possible solutions for increasing the photoelectron number, and thus the detector performance, are presented. r 2004 Elsevier B.V. All rights reserved. PACS: 07.05.Pj; 29.40.Mc; 29.40.Wk Keywords: PET; Photodetectors; Cancer
1. Introduction
*Corresponding author. INFN/Roma I, gr. Sanita, Inst. Superiore de Sanita, Viale Regina Elena 299, I-00161 Roma, Italy. Tel.: +39-6-4990-2243; fax: +39-6-4938-7075. E-mail address:
[email protected] (F. Garibaldi).
Medical imaging plays a fundamental role not only in the diagnosis and treatment of cancer but also in monitoring residual or recurring diseases occurring after the therapy. In this respect, detectors of large Field Of View (FOV) are very useful since they allow to scan the whole patient
0168-9002/$ - see front matter r 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2004.03.071
ARTICLE IN PRESS 264
F. Garibaldi et al. / Nuclear Instruments and Methods in Physics Research A 525 (2004) 263–267
body in a reasonable time. As a consequence of their high cost, conventional PET scanners, consisting of matrices of scintillators coupled to photomultipliers, have a limited surface detector. In the past, an important cost reduction was achieved by employing photosensitive gaseous multiwire proportional chambers (MWPC) to detect the light emitted from BaF2 scintillator arrays. PET detectors, proposed and built either for medical imaging [1] or small animal imaging [2], were based on the use of a photosensitive vapour (TMAE). Owing to the mismatch between the BaF2 emission spectrum and the TMAE spectral response, the sensitivity of the TMAE based detectors is lower than that of commercial scanners, although higher than that of dual head gamma camera having the same FOV (see Table 1) [3]. Spatial resolutions are instead comparable. As alternative to the nasty properties of TMAE, this paper reports on the possibilities to replace TMAE with a photosensitive CsI thin film deposited on a large area pad plane. This technology has been developed and successfully employed in RICH detectors [4,5] used in high-energy physics to detect high-energy elementary particles. To this end, a dedicated facility has been built for the vacuum deposition of CsI thin film on large area photocathodes [6]. A small-scale CsI MWPC prototype for bio-medical applications has been designed and is here presented. The technique can be usefully applied in the imaging of small animals. Here the energy resolution does not play a significant role, but photodetectors with a large
area, large FOV, and good sensitivity are needed to fully screen mice and rats by using pinhole collimators. The technique can also be used in the planar projection imaging [7], where a good spatial resolution is achieved not with collimators but with positron emitters and two photodetectors in coincidence. The possible application of the technique in the PEM imaging of breast is also under investigation [7].
2. Detector description We have designed and built a small prototype of CsI MWPC photodetector with a 10 10 cm2 active area, that can be used in the imaging of breasts and/or small animals. Fig. 1 shows a schematic drawing of the prototype. Some pictures of the prototype are reported in Fig. 2. 2.1. The scintillator The BaF2 scintillator has been chosen for its emission at short wavelength where both TMAE and CsI have a significant quantum efficiency (QE) extending up to 220 nm. This scintillator is known since long time, and its mass production technology is sufficiently developed because this scintillator has been heavily used in experiments at the SSC Collider. BaF2 is chemically stable and not hygroscopic, but rather brittle. Its scintillation light has two components, one peaking at 310 nm with a decay time of 630 ns, the other at 220 nm
Table 1 Comparison between a TMAE based PET (PETRRA) and other PET devices PETRRA
ECAT ART
ECAT EXACT
ADAC MCD
PRISM 2000
6.0 7.7 6.0 8.8 5.7 7.8 11400
6
5.0
6–10
5.0
6–8
Axial FoV (cm) Max. (kcps)
40 80
16.2 60–120
5.9 9.6 5.0 8.1 5.8 7.5 4050(2D) 5800(3D) 16.2 120–160
5.0
Sensitivity (cps/kBq/ml)
6.5 6.5 6.5 6.5 6.5 6.5 3000–4000
2100(PP) 3800(PC) 38.1 7–10
800 1900 36.8 5–10
Radial resolution (mm) Axial resolution (mm) Tangential resolution (mm)
ARTICLE IN PRESS F. Garibaldi et al. / Nuclear Instruments and Methods in Physics Research A 525 (2004) 263–267
Fig. 1. Schematic drawing of the detector prototype.
265
cathode consists of a thin layer (300 nm) of CsI evaporated onto a plane segmented into pads (B8 8 mm2). The anode plane is made of gold plated tungsten wires, 20 mm of diameter, 4 mm pitch. The wires are soldered on a G-10 printed board with a precision of 0.1 mm and stretching tension of 50 g. A positive voltage around 2100 V is applied to the anode plane with cathodes grounded; the gas gain is B105. The cathode plane, opposite to the CsI photocathode, consists of 50 mm diameter wires, 2 mm pitch. Aluminum MWPC frames have been used in order to minimize outgassing into the chamber active volume. The gas tightness is ensured by soft O-rings placed in grooves in the chamber frames. 2.3. The readout system
Fig. 2. Detector prototype under testing.
with 0.8 ns. The fast component is due to the ‘‘cross over mechanism’’ [8] and produces about 1700 photons/MeV. Segmented BaF2 crystals with pixel sizes 3 3 20 mm2 have been built and successfully used. The average QE of TMAE for the scintillation light of BaF2 has been calculated to be /QES= 9.7%, and measured is 14%. The number of photoelectrons (p.e.) is dependent on crystal size and geometry and on other parameters of the readout system. For a realistic geometry this number is between 10 and 20 p.e./MeV [9,10]. In the CsI case, a lower /QES is obtained, and consequently fewer photoelectrons. Different numbers have been quoted, either 3.7 p.e./MeV [9] or 5–10 p.e./MeV [10]. 2.2. The photodetector A conventional 2 mm MWPC equipped with a CsI photocathode [4,5] will be employed to detect the photoelectrons created by the photoconversion of the UV photons generated in the BaF2 by gamma rays. The MWPC volume, enclosed between the CsI pad cathode and the scintillator, used as window, is flushed with pure methane at room temperature and pressure. The photo-
The readout electronics consists of 192 channels read out by daisy-chained GASSIPLEX chips [11]. The readout is based on the V551 sequencer and V550 CRAMS 10 bit FADC. When a trigger is issued, the sequencer provides at the same time the clock pulse to all the FEE rows and the related convert phase shifted pulses to the ADC modules. The synchronized clock-convert pulses allow each analogue channel to be correctly converted and stored. In our prototype, the trigger is given by the anode wire signal. We are also exploring the possibility of using a readout system based on the Vicking chips, commercially available from IDEAS [12]. 2.4. The evaporation system The dedicated evaporation system built and previously used for RICH detector photocathodes, extensively described elsewhere [6], can be used. It consists of a cylindrical stainless-steel vessel (approximately 110 cm high, 120 cm in diameter) equipped with 4 crucibles containing an amount of CsI powder sufficient to create a B300 nm of deposition, whose thickness is measured by a quartz oscillator. The prepolished pad plane (a printed circuit with 3 layers of metals, copper, nichel, and gold), glued on the vetronite substrate, is housed in a vacuum chamber (10 7 Torr) and heated at 50 C. The locations of the crucibles with
ARTICLE IN PRESS 266
F. Garibaldi et al. / Nuclear Instruments and Methods in Physics Research A 525 (2004) 263–267
respect to the photocathode are optimized to ensure a maximum variation in the deposited layer of 10%. The CsI powder evaporates at a temperature of B500 C and a layer of B300 nm is achieved after 150 s. Since H2O vapour severely affects the performance of the CsI layer, the assembling of the pad planes in the detector structure is always performed in argon atmosphere. An on-line QE measurement has been built and successfully used for the construction of RICH photodetectors [6].
3. Expected performance and possible improvements The intrinsic spatial resolution of our CsI MWPC detector is very good (B500 mm). The limit is given essentially by the charge spread in MWPC. It can be optimized by tuning the dimension of the pads with respect to the distance between anode wire and photocathode. For PET application, other well-known terms affect the final total resolution. For planar imaging the intrinsic resolution has to be convoluted with that of the collimator. A serious limitation to the total resolution is imposed by the use of parallel hole collimators. Nevertheless it has been shown that submillimeter spatial resolution can be obtained by using pinhole collimators [13]. Good efficiency, while keeping very good spatial resolution, can be attained by using multipinhole or coded aperture collimators [14,15]. As previously pointed out, the photoelectron yield of the examined photodetector is poor. For this reason, our group is investigating in the following directions.
has still to be proven. Significant resources are needed to start researches in this sector. 2. Rb2Te layers could be used as photoconverter. Rb2Te has high QE in the wavelength region of interest [16]. The technique exists and has shown to be reliable. 3. Search for new materials sensitive to visible light. It is well known that the absorption process is controlled by the band gap value. A large band gap material as diamond (5.4 eV) or CsI (6.2 eV) is transparent to visible light. But if an added substance forms an impurity level within the gap, visible light has enough energy to excite electrons to the conduction band or up to vacuum level. The structure of the host material is not significantly perturbed, but extra electrons enter a ‘‘donor level’’ (or extra holes for an ‘‘acceptor level’’). The doped material presents an extra absorption tail in the visible radiation region. Of course the same absorption could be achieved by a material with a low (around 1–3 eV) energy gap, but such material has structural properties similar to those of a doped large gap material. In particular CsI is a magic material: the energy that could be supplied by phonon absorption is similar to the energy loss by scattering of secondary electrons and it results into a large QE with the maximum around 500 nm of thickness. Instead visible photomaterials have the maximum for 100–200 nm. Furthermore the electron affinity is very low (0.01 eV) allowing a very efficient escape process for photoelectrons. Our group is starting a research program in this direction. We are looking for an optimal doping of CsI, the first candidate will be the Cs itself.
4. Conclusions 1. Selection of a different scintillator, emitting a more significant amount of light in the UV part of the spectrum. Tavernier [9] reports on the properties of LaF3:Nd. The emission spectrum of this scintillator better than BaF2 overlaps the CsI QE response. But LaF3 has a poorer light output than BaF2. Nevertheless, 250 p.e./MeV can be obtained from LuAP:Nd. However the possibility of growing such kind of scintillators
Gaseous photodetectors are attractive alternatives to the solid state ones for some bio-medical application. The main advantage is related to their low cost and to the possibility of building large area photodetectors not affected from dead area. The disadvantage of this technique is the small number of photoelectrons attainable, but it is not critical for some application where the energy
ARTICLE IN PRESS F. Garibaldi et al. / Nuclear Instruments and Methods in Physics Research A 525 (2004) 263–267
267
Fig. 3. How to produce a photoconverter sensible to the visible light.
resolution does not play a significant role. One way to improve the technique is the use of more efficient scintillators and photoconverters sensitive to wavelengths in the visible range (Fig. 3).
[9]
References [1] R.J. Ott, et al., Nucl. Instr. and Meth. A 477 (2002) 475 references quoted therein. [2] Liu Xuan, S. Rajeswaran, W. Smolik, S. Tavernier, Zhang Shuping, P. Bruyndonckx, Nucl. Instr. and Meth. A 382 (1996) 589. [3] A. Divoli, Volume imaging with a novel positron camera (PETTRA), Proceeding of the ITBS2001 Conference. [4] ALICE Collaboration, Technical Design Report of the high momentum particle identification detector, CERN/ LHCC98-19, Alice TDR 1, 14 August 1998. [5] F. Garibaldi, et al., Nucl. Instr. and Meth. A 502 (2003) 117. [6] F. Cusanno, et al., Nucl. Instr. and Meth. A 502 (1) (2003) 252. [7] M.V. Green, et al., Comput. Med. Imaging Graphics 25 (2001) 79. [8] C.W.E. Van Eijk, Experimental and theoretical studies of cross luminescence, in: F. De Notaristefani, P. Lecoq
[10]
[11] [12] [13]
[14] [15]
[16]
(Eds.), Proceedings of the Conference on Heavy Scintillators for Scientific and Industrial Applications, Editions Frontieres, M. Schneegans, pp. 161–166. S. Tavernier, P. Bruyndockx, Readout of scintillation light with gas gain devices, in: Gaspar Barreira et al. (Eds.), Proceedings of the VIII International Conference on Calorimetry in High Energy Physics, Lisbon Portugal, June 13–19, World Scientific, Singapore, 1999, pp. 65–72. J. Seguinot, G. Charpak, Y. Giomataris, V. Peskov, J. Tischhauser, T. Ypsilantis, Nucl. Instr. and Meth. A 297 (1990) 133. A. Di Mauro, Nucl. Instr. and Meth. A, (2004) these Proceedings. M. Artuso, et al., Nucl. Instr. and Meth. A 502 (1) (2003) 91. F. Cusanno, et al., Proc. of ITBS 2003, Athens/Milos Island, Greece, May 26–30, 2003, Nucl. Instr. and Meth. A, (2004) in press. R. Accorsi, et al., IEEE Trans. Nucl. Sci. NS-48 (6) (2001) 2411. F. Garibaldi, et al., Small animal imaging by single photon emission using pin hole and coded aperture collimation, IEEE2003 MIC Proceedings, Trans. Nucl. Sci., submitted for publication. A. Braem, et al., Nucl. Instr. and Meth. A 502 (2003) 205.
