2424 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 57, NO. 5, OCTOBER 2010 Characterization of Silicon Detectors for the SiliPET Project: A Small Animal PET Scanner Based on Stacks of Silicon Detectors N. Auricchio, G. Di Domenico, L. Milano, R. Malaguti, G. Ambrosi, M. Ionica, E. Fiandrini, and G. Zavattini Abstract—In this paper we propose a new scanner for small animal positron emission tomography (PET) based on stacks of double sided silicon detectors. Each stack is composed of 40 planar detectors with dimension 60 mm 60 mm 1 mm and 128 orthogonal strips on both sides to read the two coordinates of interaction, the third being the detector number in the stack. Mul- tiple interactions in a stack are discarded. In this way we achieve a precise determination of the first interaction point of the two 511 keV photons. The price to pay is an efficiency reduction for each stack of about 50%. The reduced dimensions of the scanner also improve the solid angle coverage resulting in a high sensitivity. Preliminary results were obtained with the MEGA prototype tracker. Here, we report on the results obtained with double sided silicon prototype detectors, manufactured by ITC-FBK, having an active area of 3 cm 3 cm and a strip pitch of 500 m. Two dif- ferent strip widths of 300 m and 200 m, and two thicknesses of 1 mm and 1.5 mm, equipped with 64 orthogonal p and n strips on opposite sides were read out with the VATAGP2.5 ASIC, a 128- channel “general purpose” charge sensitive amplifier. We describe the experimental setup, the measurements and the results in terms of spatial resolution, spectral and timing performances obtained with the prototype detectors. Index Terms—Positron emission tomography, silicon detectors, timing. I. INTRODUCTION S MALL animal Positron Emission Tomography (PET) scanners have been developed to test the safety and effectiveness of a drug in preclinical animal models before conducting clinical trials. High spatial resolution and high sensitivity are the major goals in this kind of medical imaging applications. Positron emission tomography is a noninvasive molecular imaging technique that measures in vivo the biodistribution of a compound labeled with specific positron—emitting ra- dionuclides and it is based on the determination of the lines of Manuscript received July 14, 2009; revised December 29, 2009 and March 04, 2010; accepted April 13, 2010. Date of publication July 19, 2010; date of current version October 15, 2010. This work was supported by the Italian National In- stitute for Nuclear Physics (INFN) under the framework project “SIGESPES”. N. Auricchio, G. Di Domenico, L. Milano, and G. Zavattini are with the De- partment of Physics, University of Ferrara, 44100 Ferrara, Italy and also with INFN/Sezione di Ferrara, 44100 Ferrara, Italy (e-mail: auricchio@fe.infn.it). R. Malaguti is with INFN/Sezione di Ferrara, 44100 Ferrara, Italy. G. Ambrosi and M. Ionica are with INFN/Sezione di Perugia, 06123 Perugia, Italy. E. Fiandrini is with the Department of Physics, University of Perugia, 06123 Perugia, Italy and also with INFN/Sezione di Perugia, 06123 Perugia, Italy. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TNS.2010.2052067 response (LORs) generated by the interaction coordinates of the two back-to-back 511 keV annihilation photons. The requirements of this technique are: — to detect as many LORs as possible in order to obtain the highest sensitivity; — to measure exactly the first interaction coordinate of both annihilation photons, in order to have the finest spatial res- olution. Two physical effects, intrinsic to this application, limit the spatial resolution in the final images: the non collinearity of the 511 keV photons and the positron range. A positron emitted from a decay will thermalise, after having traveled a certain distance (typically referred to as positron range), then annihi- late with an electron. In most annihilations two almost back to back photons will be emitted each carrying 511 keV. There is an angular dispersion from collinearity of about 0.5 , due to the residual momentum associated with the electron-positron center of mass. Currently small animal PET scanners are based commonly on pixellated scintillator crystals coupled to position sensitive photomultipliers with a center of mass position determination algorithm. This approach makes it difficult to improve scanner performances due to technological limits: — multiple interactions in thick scintillators; — fewer scintillator photons reaching the photocathode due to the smaller matrix elements with a high aspect ratio; — parallax error due to the indetermination of the depth of interaction (DOI). Scintillators adopted for this application are usually mate- rials with a high atomic number Z [1], [2] both to maximize photofraction and to increase detection efficiency. The requirements of a detection system for small animal imaging are: — small scanner size to reduce the effect of non collinearity and to increase solid angle coverage; — detection of two -rays with a single interaction or with the capability of distinguishing the first interaction; — thick detectors for high efficiency. In small animal scanners the measurement of the energy de- posited in the detectors, for scattered coincidence rejection, is not necessary because the amount of scattering of 511 keV is negligible inside the animal [3]. The ultimate spatial resolution is limited by the positron range. Monte Carlo simulations show that an intrinsic res- olution of 0.5 mm FWHM can be achieved with detectors consisting of 250 m pixels [4]. But due to inter-crystal scatter an increasing fraction of events will be assigned to the wrong 0018-9499/$26.00 © 2010 IEEE