Nuclear Instruments and Methods in Physics Research A 458 (2001) 62}67
A new design for a high resolution, high e$ciency CZT gamma camera detector C. Mestais*, N. Ba!ert, J.P. Bonnefoy, A. Chapuis, A. Koenig, O. Monnet, P. Ouvrier Bu!et, J.P. Rostaing, F. Sauvage, L. Verger LETI (CEA-Technologies Avance& es), CEA/GRE, 17 rue des Martyrs, Grenoble Cedex 9, France
Abstract We have designed a CZT gamma camera detector that provides an array of CZT pixels and associated front-end electronics } including an ASIC } and permits gamma camera measurements using the method patented by CEA-LETI and reported by Verger et al. [1]. Electron response in each CZT pixel is registered by correcting pulse height for position of interaction based on fast rise-time information. This method brings advantages of high scatter rejection while allowing high detection e$ciency. These techniques and the systems approach have been developed at CEA-LETI in an exclusive joint development with BICRON and CRISMATEC who in turn are commercializing the technology. The initial system is implemented in an array framework with 1920 pixels, approximately 180;215 mm in dimension, but the system architecture expands readily to 4096 pixels, and these arrays can be ganged into groups of up to 8 for pixel planes totaling over 32 000 pixels without architecture changes. The overall system design is described and brain phantom images are presented that were obtained by scanning with a small number of pixels. 2001 Elsevier Science B.V. All rights reserved. Keywords: Gamma camera detector; CdZnTe; Energy resolution; Detection e$ciency; Compactness; Biparametric spectrum
1. Introduction Scintillation cameras based on NaI(Tl) crystals and photomultiplier tubes (PMTs) [2] are reaching fundamental performance limits, especially in terms of energy resolution. Today's resolutions of 9}10% at 140 keV limit scatter rejection and degrade image contrast. Semiconductor detectors o!er resolutions of a few percent, but present a number of other design and performance challenges, in particular, achieving resolution at room temperature * Corresponding author. Tel.: #33-4-76-88-31-23; fax: #334-76-88-51-64 E-mail address:
[email protected] (C. Mestais).
without loss of detection e$ciency [3,4]. Roomtemperature pixelated semiconductor arrays also o!er other advantages including improved system count rate performance, compactness, light weight, live edges, and the #exibility to construct various detector shapes [5}7]. System spatial resolution is expected to be similar to scintillation cameras because this parameter depends primarily on the geometric characteristics of the collimator. 2. Answering the challenge: energy resolution with detection e7ciency CdZnTe has electron mobility which is much higher than its hole mobility, a fact which makes
0168-9002/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 9 0 0 2 ( 0 0 ) 0 0 8 5 2 - 4
C. Mestais et al. / Nuclear Instruments and Methods in Physics Research A 458 (2001) 62}67
63
Fig. 1. BP spectrum acquired on characterization bench.
the pulse height depend on the interaction position along the detector and leads to asymmetric spectra with the so-called `hole-tailinga. As a result, many full-energy events are outside the full energy peak, even though its width at half-maximum may be quite narrow. This dependence on position can be corrected by measuring both pulse height and fast rise time as seen in Fig. 1 (BP spectrum acquired on characterization bench), and detailed elsewhere [1,8,9,10]. Our system is based on this method and provides an imaging array that achieves good energy resolution while preserving detection e$ciency. The project, named `PEGASEa for `Projet d'Etude de GAmma cameH ra a` SEmiconducteura, is worked out at CEA-LETI in an exclusive joint development with BICRON and CRISMATEC.
at 10 cm from the collimator face. The pixel thickness is taken at 6 mm to preserve detection e$ciency.
4. Detector technology We use CdZnTe grown with the High Pressure Bridgman (HPB) method including material from both BICRON in Cleveland and LETI in Grenoble. Devices have been fabricated at LETI, including contact deposition, edge passivation and coating. They are tested for detection e$ciency and energy resolution at LETI and assembled on platforms at MATRA in Compie`gne. Processes for device manufacturing and testing are being implemented at BICRON for production.
3. Pixel size 5. Front-end ASIC Pixels of size approximately 4 mm and pitch of about 4.5 mm were chosen to provide system spatial resolution similar to the spatial resolution of a scintillation camera equipped with a Low Energy All Purpose (LEAP) collimator, typically 8}9 mm
To preserve semiconductor detector energy resolution the preampli"ers must be near the detectors, and space and cost considerations dictate application-speci"c integrated circuit (ASIC) technology.
SESSION II.
64
C. Mestais et al. / Nuclear Instruments and Methods in Physics Research A 458 (2001) 62}67
Fig. 2. A 16-channel multiplexed ASIC preampli"er.
For these reasons, a 16-channel ASIC has been designed with input transistor size and consumption adapted to the detector characteristics. Previous experience was acquired at LETI with the ASIC design of ISGRI, the CdTe gamma camera on board the Integral satellite [11,12]. Special attention was paid to gain bandwidth product in order to optimize the signal-to-noise ratio and preserve rise-time information. In addition, isolation between analog and digital functions was emphasized to avoid cross-talk, and immunity to electromagnetic disturbances were carefully taken into account in the ASIC, platform and electric boards design (see Fig. 2).
and pixels has the appearance of a stand-alone electrical component, with a size of approximately 18;18 mm, and is easy to handle, to manufacture and to replace in the detector plane. The platform structure has been designed in cooperation with Matra BAe Dynamics (Fig. 3). To acquire the pulse height and rise time for each event detected, groups of 64 pixels are associated with a spectrometry circuit that measures and reports pulse height and rise time for each event, and a data management circuit that tracks and reports pixel location, pulse height and rise time. These functions are accomplished today with surfacemount boards attached to an interconnection board, though they could be implemented with ASICs, one per platform for instance, for better compactness and lower cost. A dialogue board providing overall data management and interface with the computer is also included. The full contiguous prototype array is currently designed to include 10;12 platforms, i.e. 40;48 pixels, so that the complete array is approximately 180;215 mm in dimension with 1920 pixels (Fig. 4). The system architecture is structured to accommodate a board array of up to 64;64" 4096 pixels without change in system logic, and up to 8 boards can be combined to provide up to 32 000 pixels without changes in system logic. The individual 1920 pixel boards are buttable on three sides maintaining pixel size, pitch and spacing tolerance from one board to the other which allows
6. Imager structure and system architecture The imager system is designed to provide a modular, #exible architecture adaptable to various applications in terms of pixel count, "eld of view and detector shape while also being consistent with cost-e!ective manufacture and maintenance. The pixel array is composed of 4;4 building blocks of 16 pixels, each including a 16-channel multiplexed ASIC preampli"er all mounted to a platform. One of these units complete with ASIC
Fig. 3. Platform.
C. Mestais et al. / Nuclear Instruments and Methods in Physics Research A 458 (2001) 62}67
65
Fig. 4. Imager prototype.
various "elds of view by arraying the boards: 180;215 mm (single), 360;215 mm (double), 180; 430 mm (double), 540;215 mm (triple), 360; 430 mm (four) and so forth up to 720;430 mm.
7. Operation During a calibration phase, the system acquires one BP spectrum per pixel, and then computes a two-dimensional (2D) window around the correlated events, with a width chosen by the operator. In the examination phase, all events of a given pixel whose pulse height and rise time fall within the 2D window contribute to the image, all the others are rejected. This patented scheme allows discrimination of low-energy events coming from scatter from those higher energy events that have low pulse heights because they have interacted close to the anode in the detector (see Fig. 5). Moreover, the 2D window width can also be adjusted to compensate for non-uniform detection e$ciency from pixel to pixel in the image plane.
8. Preliminary results The BP spectrum of Fig. 5 was acquired on the imager and gives 70% detection e$ciency in a $6.5% window at 122 keV. Better results are expected in the future when noise is reduced with
Fig. 5. BP spectrum acquired on imager prototype and 2D window.
Version 2 of the ASIC and platform. Count rate of the system is greater than 200 counts/pixel/s. Images have been obtained by scanning phantoms with a limited number of operating pixels (Fig. 6) and were compared to a standard scintillation camera. A brain phantom was "lled with a total activity of 10 mCi of Tc with local concentrations varying up to #70%}50% in various cavities. The scintillation gamma camera and the scanning pixel array were "t with similar collimators and each acquired images for 120 s. For the scintillation gamma camera, the energy acceptance window was set at $10%, and the total number of counts in the image was close to 2.7 millions (Fig. 7). For the CZT image, the 2D energy acceptance window was set at $6.5%, and the total number of counts in the image was close to 1.7 millions (Fig. 8). The CZT pixelated image has been
SESSION II.
66
C. Mestais et al. / Nuclear Instruments and Methods in Physics Research A 458 (2001) 62}67
Fig. 6. Test bench.
Fig. 7. Brain phantom image with scintillation camera.
processed with special patented image interpolation software that allows an image presentation similar to scintillation cameras (Fig. 9). The images are quite similar, which is encouraging for this early work. Nevertheless, the measured detection e$ciency with the scanning pixel array is close to 60% to be compared with 96% for the scintillation camera equipped with a 3/8 NaI(Tl) crystal. This is mainly due to the fact that the pixel detection e$ciencies ranged between 63% and 85% as measured on the characterization test bench. For the image, we used a rough uniformity correction which aligned all e$ciencies on the less e$cient pixel. Moreover the preampli"er noise of Version 1 of the ASIC widened the BP correlation.
Fig. 8. Brain phantom rough image with CZT imager prototype.
Fig. 9. Brain phantom interpolated image with CZT imager prototype.
9. Conclusion We have designed a CZT imager system who answers the challenge good energy resolution with high detection e$ciency by using a 2D spectrometric window around the correlated events in the biparametric spectrum, and gives encouraging results for this stage of development. Progresses are expected with noise reduction in Version 2 of the ASIC and platform. The modular and #exible architecture is adaptable to various "elds of view, pixel counts and detector shapes. The system design is consistent with cost-e!ective manufacture and easy maintenance.
C. Mestais et al. / Nuclear Instruments and Methods in Physics Research A 458 (2001) 62}67
Acknowledgements This project would not have existed without the perspicacity and tenacity of R. Allemand and J. Mareschal, both now retired, and the help of O. Peyret and M. Wolny. Part of the material research was sponsored by a grant of the French Ministry of Research. The authors wish to thank R. DupreH et al. from Matra BAe Dynamics, C. Massit, and P. Trystram from LETI for collaboration on platform design, F. Boucly, G. Chamming's, F. Josso, and P. Villard from LETI for collaboration on ASIC design, A. Glie`re and F. Mathy from LETI for collaboration on software design, F. De La Barre for collaboration on collimator design, A. Bergot from La Fonderie de Gentilly for manufacturing the collimator and imager lead envelope, D. Michon from Art Technique for photographs, and M.R. Mayhugh for commenting on the manuscript. References [1] L. Verger, J.-P. Bonnefoy, F. Glasser, P. Ouvrier-Bu!et, J. Electron. Mater. 26 (6) (1997) 738. [2] H.O. Anger, Rev. Sci. Instr. 29 (1958) 27. [3] A. Shor, Y. Eisen, I. Mardor, Nucl. Instr. and Meth. A 428 (1999) 182.
67
[4] C.L. Lingren, B. Apotovsky, J.F. Butler, F.P. Doty, S.J. Friesenhahn, A. Oganesyan, B. Pi, S. Zhao, Materials Research Society Symposium Proceedings Vol. 487, Materials Research Society, Boston, 1998, p. 263. [5] J.F. Butler, C.L. Lingren, S.J. Friesenhahn, F.P. Doty, W.L. Ashburn, R.L. Conwell, F.L. Augustine, B. Apotovski, B. Pi, S. Zhao, C. Isaacson, IEEE Trans. Nucl. Sci. NS-45 (3) (1998) 359. [6] C. Scheiber, B. Eclancher, J. Chambron, V. Prat, A. Kazandjan, A. Jahnke, R. Matz, S. Thomas, S. Warre, M. Hage-Hali, R. Regal, P. Si!ert, M. Karman, Nucl. Instr. and Meth. A 428 (1999) 138. [7] Y. Eisen, A. Shor, I. Mardor, Nucl. Instr. and Meth. A 428 (1999) 158. [8] L. Verger, J.-P. Bonnefoy, A. Glie`re, P. Ouvrier-Bu!et, M. Rosaz, Materials Research Society Symposium Proceedings, Vol. 487, Materials Research Society, Boston, 1988, p. 171. [9] L. Verger, M. Boitel, M.C. Gentet, R. Hamelin, C. Mestais, F. Mongellaz, J. Rustique, G. Sanchez, Invited paper IX.2, These proceedings, Nucl. Instr. and Meth. A 458 (2001) 297. [10] F. Mathy, J.P. Bonnefoy, A. Glie`re, C. Mestais, L. Verger, Poster P13, These proceedings, Nucl. Instr. and Meth. A 458 (2001) 484. [11] M. Arque`s, N. Ba!ert, D. Lattard, J.L. Martin, G. Masson, F. Mathy, A. Noca, J.P. Rostaing, P. Trystram, P. Villard, J. CreH tolle, F. Lebrun, J.P. Leray, O. Limousin, IEEE Trans. Nucl. Sci, NS-46 (3) (1999), p. 181. [12] O. Limousin, J.P. Leray, M. Arque`s, N. Ba!ert, P. Baron, C. Blondel, J. CreH tolle, E. Delagnes, A. Jourdan, P. Laurent, F. Mathy, J.P. Rostaing, P. Trystram, P. Villard, Nucl. Instr. and Meth. A 458 (2001) 551.
SESSION II.
View publication stats
