80 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 1, FEBRUARY 2004 A Study of Intrinsic Crystal-Pixel Light-Output Spread for Discrete Scintigraphic Imagers Modeling Raffaele Scafè, Rosanna Pellegrini, Alessandro Soluri, Livia Montani, Angelo Tatì, Maria Nerina Cinti, Francesco Cusanno, Gianfranco Trotta, Roberto Pani, and Franco Garibaldi Abstract—This paper is focused on the discrete scintillation imaging devices, consisting of crystal arrays and metal-channel dynode Hamamatsu 1” and 2” square position sensitive photo- multiplier tubes (PSPMTs). These devices are suitable for nuclear medicine based high resolution applications, and, particularly, for single photon emission computed tomography (SPECT). The model of scintillation light distribution (SLD) previously developed was able to distinguish the responses from crystal-pixels with different side, but it was not detailed enough to explain the influence of crystal-thickness. For this reason the experimental data were reviewed to find a new and more adequate analyt- ical model. The improved SLD model explains the influence both of crystal-side and crystal-thickness on the scintillation light-output spread. The SLD expression is quite simple and its spread depends only on one q-parameter. This expression is well adaptable over the range of examined crystal arrays. Furthermore, in the considered experiments, the SLD q-param- eter was found linearly dependent on crystal-pixel shape factor . An overview of discrete scintillation imager simulator (DISIS) computer code is reported. Major outcomes of this work are: 1) the improved expression of SLD, which consolidates the DISIS performances, and 2) a tool for local SLD-spread control in the imager field of view (FOV). Index Terms—High resolution, nuclear medicine, position sensi- tive photomultiplier tube, scintillation array, scintillation imager, spatial light distribution. I. INTRODUCTION T O improve the performances of discrete scintillation im- agers up to the level required for high-resolution nuclear medicine applications, it is mandatory to gain further insight into the scintillation light distribution (SLD), emerging from a crystal-pixel of the array. The most popular choice of planar light sensors for crystal-pixels light-output readout concerns devices measuring illuminance by charge-integrals obtained by anode-stripes or -matrices. Many applications, reported in the literature about the light-output readout, involve PSPMTs, based on cross-wire, cross-plate, or anode-matrix structures [1]–[10]. Manuscript received March 24, 2003; revised September 25, 2003. R. Scafè, L. Montani, and A. Tatì are with the Italian National Council for New Technologies, Energy and the Environment (ENEA), C.R. Casaccia, I-00060 S.Maria di Galeria (Rome), Italy (e-mail: scafe@casaccia.enea.it). R. Pellegrini, M. N. Cinti, G. Trotta, and R. Pani are with the Department of Experimental Medicine, University of Rome “La Sapienza,” 00100 Rome, Italy. A. Soluri is with the Institute of Biomedical Technologies, National Research Council (CNR), 00100 Rome, Italy. F. Cusanno and F. Garibaldi are with the Laboratory of Physics, Italian Na- tional Institute of Health (ISS), 00100 Rome, Italy. Digital Object Identifier 10.1109/TNS.2003.823039 Other devices use photodiodes for pixel-by-pixel light-output readout [11]–[15]. Presently, the former solution is widely pre- ferred to the latter one, basically because the same light-sensor may be used coupling different crystal-pixel sizes. Our attention has been focused on imaging devices based on metal-channel dynode Hamamatsu 1” and 2” square posi- tion sensitive photomultiplier tubes (PSPMTs) [16], which are suitable for nuclear medicine based high-resolution applications and, particularly, for single photon emission computed tomog- raphy (SPECT). They allow to perform imaging by applying the Anger Principle [17] and using narrow SLD emerging from few millimeters crystal-pixel side. The small PSPMT internal charge-spread shown by metal- channel technology can produce a pixel spread function (PSF) having a full width half maximum (FWHM) significantly nar- rower than the pixel-side. Provided this condition is satisfied, a design effort can be devoted to obtain the best pixel identifica- tion over the field of view (FOV) (i.e., to obtain the best linear correlation between planar coordinates of primary interaction in the FOV and the event localization in the corresponding recon- structed image). To optimize the image contrast and the signal-to-noise ratio (SNR), the imager spatial linearity over the FOV seems the prin- cipal factor to be taken under control. The previous model was not detailed enough to explain the influence of crystal-pixel thickness on SLD spread, therefore, the experimental data were reassessed to find a new and more adequate analytical model. II. EQUIPMENT AND METHODS To evaluate the SLD expression two independent methods were identified. Firstly, a photographic technique was utilized to obtain qualitative information about SLD, particularly about the symmetry around the pixel-axis and the overall light-spread. Secondarily, an accurate method was applied for SLD-spread single-event estimation using charge-strip data measured by a linear array PSPMT. Ten 5 5 CsI(Tl) scintillation arrays with different pixel size were individually investigated. Crystal-pixel side was in the range from 1.0 to 2.0 mm and thickness ranged from 1.0 to 5.0 mm (Table I). Such arrays were manufactured by Hilger Crystals, U.K., using a white-epoxy coating (typically 0.1 mm thick) for the crystal-pixel optical insulation and the light-output improvement. 0018-9499/04$20.00 © 2004 IEEE