90 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 52, NO. 1, FEBRUARY 2005 The ECAT HRRT: An Example of NEMA Scatter Estimation Issues for LSO-Based PET Systems L. Eriksson, Member, IEEE, C. C. Watson, Member, IEEE, K. Wienhard, M. Eriksson, M. E. Casey, Member, IEEE, C. Knoess, M. Lenox, Z. Burbar, M. Conti, B. Bendriem, Member, IEEE, W. D. Heiss, and R. Nutt, Fellow, IEEE Abstract—The ECAT high-resolution research tomograph (HRRT) is a three-dimensional (3-D)-only dedicated brain positron emission tomograph with LSO and GSO scintillators. In this paper, the system has been looked at as an example of issues that need to be addressed when evaluating LSO-based system following the recent NEMA NU 2-2001 protocols. The LSO scintillators contain the isotope which is radioactive and creates a small amount of single counts and, depending on the low level discriminator threshold for validated singles, also true coincidence events from a cascade of gamma rays following the beta decay of into . The presence of intrinsic random and true coincidence events has an effect on the scatter fraction determination if using the guidelines according to NU 2-2001. We show here that these guidelines have to be changed in order to ob- tain accurate determination of the scatter fraction. In this paper, we have determined the scatter fraction for different low level discriminator settings. Since the scatter fraction determinations also comprise full count rate studies for the NEMA phantom, we have in addition extracted the sensitivity information for singles, true scatter, and for the NEC data. Index Terms—Intrinsic radiation, LSO, NEMA NU 2-2001, positron emission tomography (PET). I. INTRODUCTION T HE high-resolution research tomograph (HRRT) is a three- dimensional (3-D)-only dedicated brain tomograph em- ploying the new scintillator LSO. To provide depth-of-interac- tion (DOI) information, the detectors are based on two scintilla- tors (phoswich), separable due to differences in decay time. The second scintillator may be GSO , or more recently, LYSO. The system has been previously evaluated [1] regarding sen- sitivity, spatial resolution, and scatter contributions. The spa- tial resolution has been shown to be better than 2.5 mm within the imaged field-of-view (FOV). The sensitivity was determined based on a 20-cm diameter phantom, 20-cm long and filled with activity, and with a point source. An initial study regarding the count rate performance and sensitivity measure- ments based on the NEMA NU 2-2001 standard [2] has been previously reported [5]. Manuscript received November 14, 2003; revised September 7, 2004. This work was supported in part by the Swedish Science Research Council. L. Eriksson and M. Eriksson are with CPS Innovations, Knoxville, TN 37932 USA, and also with the Karolinska Institute, SE-17177 Stockholm, Sweden (e-mail: Lars.Eriksson@cpspet.com). C.C. Watson, M.E. Casey, M. Lenox, Z. Burbar, M. Conti, and B. Bendriem are with CPS Innovations, Knoxville, TN 37932 USA. K. Wienhard, C. Knoess, and W.D. Heiss are with the Max-Planck Institute for Neurological Research, D-50931 Cologne, Germany. R. Nutt is with CTI Molecular Imaging, Knoxville, TN 37932 USA. Digital Object Identifier 10.1109/TNS.2004.843139 Fig. 1. Beta decay of to energy levels of [3]. Fig. 2. Energy spectrum of Lu 176 from a piece of LSO placed on a diameter, long NaI(Tl) detector. The 511-keV energy peak is shown as a reference. The has an abundance of 2.6% of the Lu in LSO and creates approximately 250 c/s per cubic centimeter of LSO, using the information from the full energy spectrum. In addi- tion to singles and randoms coincidence count rates created by the , a certain number of true coincidence counts may be present due to the cascading gamma rays. Fig. 1 shows the decay of into and Fig. 2 shows the energy spectrum ac- quired with a NaI(Tl) ( diameter long) detector from a 4 4 LSO crystal placed on the detector. 0018-9499/$20.00 © 2005 IEEE