Performance Assessment of Pixelated LaBr 3 Detector Modules for TOF PET

Performance Assessment of Pixelated LaBr 3 Detector Modules for TOF PET A. Kuhn, S. Surti, Member, IEEE, J. S. Karp, Senior Member, IEEE, G. Muehllehner, Fellow, IEEE, F.M. Newcomer, R. VanBerg Abstract-- Our recent measurements with pixelated LaBr 3 Angerlogic detectors for use in time-of-flight (TOF) PET have demonstrated excellent energy resolution (5.1% at 511 kev) and coincidence time resolution (313 ps FWHM) with small prototype configurations [1]. A full size detector module suitable for a whole-body 3D PET scanner has been constructed based on the prototype designs and consists of 1620 4x4x30 mm 3 LaBr 3 crystals. We have utilized simulations to guide experimental measurements with the goal of optimizing energy and time resolution in evaluating triggering configurations and pulse shaping needed in a full system. Experimental measurements with the detector module indicate energy and time resolution consistent with our earlier prototypes when measured at low count rate. At very high count rate the energy, time and spatial resolution degrade due to pulse pileup. While it is possible to reduce pulse pileup by using smaller PMTs (i.e., 39 mm instead of 50 mm), we are trying to limit the total number of PMTs needed for a full-scale PET scanner with a large axial field-of-view. Therefore, we have implemented a pulse shaping circuit to improve the detector response and performance at high count rate. Simulations of a complete LaBr 3 scanner indicate significant improvements in NEC and spatial resolution can be achieved using pulse shaping. I. INTRODUCTION ime-of-flight (TOF) measurement in PET has the potential T to significantly improve scanner performance [2]-[4]. However, to realize this improvement requires utilizing a scintillator with high light output, high stopping power, fast decay time, and good linearity [5]-[7]. While scintillator development is on going [8]-[14], the possibility for TOF PET has been proposed for both LSO [15] and lanthanum scintillators [16]. Our current interest is in the LaBr 3 scintillator. LaBr 3 has very high light output (~61,000 photons per MeV), fast decay Manuscript received October 2004. This work was supported in part by DOE Grant No. DE-FG02-88ER60642, NIH Grant No. R33EB001684 and a research agreement with Saint-Gobain Crystals. A. Kuhn, S. Surti, and J. S. Karp are with the Department of Radiology, University of Pennsylvania, Philadelphia, PA 19104 USA (telephone: 215-573-3958, e-mail: akuhn@rad.upenn.edu, surti@rad.upenn.edu, karp@rad.upenn.edu). F. M. Newcomer, and R. Vanberg are with the Department of Physics, University of Pennsylvania, Philadelphia, PA 19104 USA (e-mail: mitch@hep.upenn.edu and rick@hep.upenn.edu). G. Muehllehner is with Philips Medical Systems, Philadelphia, PA 19104 USA (e-mail: gerd.muehllehner@philips.com). time (< 25 ns) and excellent energy resolution (3.2% at 662 kev) [14] making it a viable detection material for TOF PET. We have recently demonstrated excellent performance of pixelated LaBr 3 detectors in small prototype configurations. These detectors consisted of multi-crystal arrays of LaBr 3 (each crystal 4x4x30 mm 3 ) coupled to a continuous light guide and hexagonal arrangement of 50 mm photomultiplier tubes (PMTs). We measured an average energy resolution (at 511 kev) of 5.1% FWHM for individual crystals in the detector and a coincidence time resolution of 313 ps FWHM between two LaBr 3 (5% Ce concentration) Anger-logic detectors [1]. We have extended our design to a full size detector module suitable for use in a whole-body 3D PET scanner. The detector module consists of 1620 individual 4x4x30 mm 3 LaBr 3 crystals (hermetically sealed and packed in a reflective powder by Saint-Gobain Crystals) coupled to a continuous light guide and hexagonal array of Photonis XP20Y0 PMTs (a fast, 8 stage version of the XP2020). Fig. 1 illustrates the arrangement of the detector module. Fig. 1. Top and side view of the LaBr 3 detector module illustrating the arrangement of the PMTs, light guide, and scintillator crystals. Note the hexagonal PMT arrangement in the top view. II. DETECTOR MODULE TESTING In order to test energy resolution, coincidence time resolution and relative light output of individual crystals in the module as well as crystal discrimination, a data acquisition system utilizing both a TDC and ADC was implemented. Fig. 2 illustrates a block diagram layout of the acquisition system. A. Energy Resolution at 511 kev The energy resolution of individual crystals in the detector module was measured by applying software position gating on event positions and generating a histogram of event energies 0-7803-8700-7/04/$20.00 (C) 2004 IEEE 0-7803-8701-5/04/$20.00 3402 (C) 2004 IEEE

for the individual crystals. This was measured using an integration time of 60 ns in the charge integrating ADC. To obtain the energy resolution as a function of event rate in the module, realistic source geometry was chosen which consisted of a 20 cm diameter by 30 cm long cylinder filled with 18 F and water placed 35 cm from the front face of the detector. Fig. 3(a) illustrates the source and detector module arrangement. The average measured energy resolution for a group of 25 crystals in the detector module as a function of activity concentration in the cylinder is shown in Fig. 3(b). The measurement error bars represent the standard deviation in the average. Excellent energy resolution of better than 5.5% is achieved for activity concentrations < 0.5 µci/cc well above the clinical range. At very high activity concentration the energy resolution degrades due to pulse pileup in the detector module. To illustrate pulse pileup Fig 4(a) shows the detector module and relative light spread from two events. Due to the design of our detector, the two events spread their light to a local group of 7 PMTs (marked in blue and red for the primary and pileup event, respectively). However, as a result of their proximity, two PMTs share light from both events (marked in purple). Fig. 4(b) shows the pulses from the PMTs that share light from both events. Because the pulse from the pileup event occurs within the integration period of the primary pulse additional signal is added to the primary event. This causes degradation in energy resolution as well as shifting of the photopeak to higher energy. Fig. 5(a) shows the measured shift in photopeak energy as a function of activity concentration in the cylinder. No shift in the photopeak or energy resolution (Fig. 3) is measured at activity concentrations up to ~0.4 µci/cc. Fig. 5(b) shows the energy spectrum representative of activity concentrations up to 0.4 µci/cc and the spectrum at 0.65 µci/cc. Fig. 2. Block diagram of the electronics layout used in data acquisition. Fig. 3. (a) Source and detector module arrangement. (b) Average measured energy resolution for 25 crystals as a function of activity concentration. Fig. 4. (a) Light spread from two events in the detector module. Blue PMTs indicate light spread from primary event, red PMTs from pileup event, and purple PMTs share light from both events. (b) Pulse shapes in the PMTs sharing light from both events. Shaded region indicates the additional signal in the primary event. 0-7803-8700-7/04/$20.00 (C) 2004 IEEE 0-7803-8701-5/04/$20.00 3403 (C) 2004 IEEE

Fig. 7. Reduction in pulse width with shaping reduces the overlap of event pulses and thus the effect of pileup at high event rate. Fig. 5. (a) Measured shift in photopeak energy as a function of activity concentration. (b) Energy spectrum representative of activity concentrations up to 0.4 µci/cc and the spectrum at 0.65 µci/cc. B. Pulse Shaping We have constructed a pole-zero cancellation circuit with the goal of reducing the effect of pulse pileup, and thus improving the detector response at very high event rate. A diagram of the circuit is shown in Fig 6(a). Measured pulse shapes (at 511 kev) from a single LaBr 3 crystal coupled directly the PMT photocathode with and without pulse shaping are shown in Fig. 6(b). The shaping circuit reduces the pulse width by ~1/2. This reduces the overlap in pulses at high event rate as shown in Fig. 7. It is important to point out that only a small loss in energy resolution was measured at low event rate with the pulse shaping. The energy resolution was 4.25% at 511 kev with pulse shaping as compared to 3.5% without shaping, for a single LaBr 3 crystal directly coupled to a PMT. Additionally, the fast rise of the pulse is preserved with the shaping circuit thus maintaining excellent timing characteristics. C. Coincidence Time Resolution The coincidence time resolution for crystals in the detector module was measured by utilizing the electronics setup illustrated in Fig. 2, however, a BC-418 plastic scintillator coupled to a XP2020 PMT formed the start channel to the TDC. With a positron point source placed between the two detectors and placing software position gates on individual crystals in the detector module, coincidence time spectra were generated. The time resolution was measured for crystals at various locations in the detector module. Fig. 8(a) illustrates three measured positions in the module. By utilizing the signals from a group of 7 PMTs in the analog sum (see Fig. 2) light collection is maximized and thus time resolution improved. For example, the crystal at position 1 distributes all of its light to the 7 PMT group (blue + red) and only 73% to the single PMT (blue). The corresponding time resolution measured for the group of 7 PMTs is 295 ps FWHM and 350 ps FWHM for the single PMT as shown in Fig 8(b). Using a group of seven PMTs for timing measurements, only a small variation in time resolution is measured over the detector module (~ 40 ps). Fig. 6. (a) A Diagram of the pulse shaping circuit. (b) Measured pulse shapes (at 511 kev) from a single LaBr 3 crystal coupled directly to a PMT photocathode with and without pulse shaping. Fig. 8. (a) Detector module indicating three measured positions in the module. (b) Time resolution and light collected for 1 and 7 PMTs at the three positions. 0-7803-8700-7/04/$20.00 (C) 2004 IEEE 0-7803-8701-5/04/$20.00 3404 (C) 2004 IEEE

Additionally, the time resolution was measured as a function of event rate. By utilizing additional 511 kev activity, the event rate in the detector module was increased incrementally. Fig. 9 shows the average measured time resolution for a group of nine crystals in coincidence with the reference detector. An additional scale on the X-axis is shown to correlate the event rate to the activity concentration in the 20 cm diameter by 30 cm cylinder. At very high event rates the time resolution degrades due to pulse pileup. However, by utilizing the pulse shaping circuit described in Sec. II.B, excellent time resolution of better than 310 ps FWHM is maintained up to ~1.2 µci/cc activity concentrations well beyond the clinical range. shown in Fig. 11 for four scanner configurations as a function of activity concentration in a 20 cm diameter by 70 cm cylinder. A secondary Y-axis is shown indicating the effective NEC with TOF for 300 ps FWHM time resolution as calculated from reference (NEC gain is 2D/c t, where D is the object diameter, c the speed of light, and t the timing resolution) [16]. Fig. 10. Scanner energy spectra at low and high event rates. True events are pushed out and additional scattered events are pushed into the fixed energy window at high rate due to pulse pileup. Fig. 9. Measured time resolution as a function of event rate > 400 kev in a 7 PMT cluster. Secondary X-axis scale correlates the event rate to the activity concentration in a 20 cm dia. by 30 cm cylinder. III. SIMULATED LABR 3 SCANNER PERFORMANCE In order to relate the measured detector module performance to that of the full system, simulations of a full LaBr 3 scanner were performed. The details of the simulation process are similar to those outlined in reference [17]. The scanner simulated has an axial field-of-view of 25 cm and a diameter of 84 cm (with a 65 cm port). The detector was modeled as a pixelated Anger-logic detector using 4x4x30 mm 3 LaBr 3 crystals with a crystal pitch of 4.3mm. A fixed energy window of 470 to 665 kev was used to determine valid events, with a coincidence window of 5 ns and no electronics dead time. The system was simulated with and without pulse shaping in two PMT size configurations (50 mm and 39 mm) Pulse pileup at high event rates degrades the scanner performance by pushing true events out and additional scattered events into the fixed energy window. This is illustrated in Fig. 10 showing the scanner energy spectra at low and high rates. This is the same effect measured for individual crystals in the detector module and shown in Sec. II.A. The resulting simulated noise equivalent count-rate (NEC) is Fig. 11. Simulated noise equivalent count-rate (NEC) for four scanner configurations as a function of activity concentration in a 20 cm diameter by 70 cm cylinder. A secondary Y-axis is shown indicating the effective NEC with TOF for 300 ps FWHM time resolution as calculated from reference (gain is 2D/c t ) [16]. While it is possible to reduce the effect of pulse pileup at high event rates and improve NEC by using smaller PMTs (i.e., 39 mm instead of 50 mm), it is a practical advantage to limit the total number of PMTs needed in a full scanner with a large axial field-of-view. Therefore, by utilizing 50 mm PMTs 0-7803-8700-7/04/$20.00 (C) 2004 IEEE 0-7803-8701-5/04/$20.00 3405 (C) 2004 IEEE

with pulse shaping the expected NEC can be increased ~20% at high event rates becoming equal to using 39 mm PMTs. Additionally, the reconstructed spatial resolution for the scanner was simulated. Fig. 12 shows the spatial resolution as a function of activity concentration calculated using NEMA NU2-2001 analysis [18]. Again the use of pulse shaping with 50 mm PMTs improves the predicted spatial resolution to the level achieved with 39 mm PMTs. Fig. 12. Simulated reconstructed spatial resolution as a function of activity concentration for the LaBr 3 scanner. IV. CONCLUSIONS Our results indicate that the excellent energy and time resolution we obtained with small LaBr 3 prototype detectors can be extended to a 1620 crystal detector module suitable for use in a whole body 3D TOF PET scanner. Measured energy resolution of better than 5.5% at 511 kev was achieved for activity concentrations up to ~0.5 µci/cc. Coincidence time resolution of better than 310 ps FWHM was measured for event rates well beyond those expected clinically (rates corresponding to activity concentrations up to 1.2 µci/cc). By utilizing the signal from a hexagonal cluster of 7 PMTs, uniform light collection as well as small variations in time resolution was measured. A pole-zero cancellation circuit was implemented which reduced the pulse shape width by ~1/2 while preserving the measured energy and time resolution of the detector. At high event rate, the measured time resolution was improved by the pulse shaping. Simulations of a full LaBr 3 scanner indicate a significant improvement in NEC and spatial resolution is expected at high event rate using pulse shaping. The excellent energy, time, and spatial resolution measured with the detector module indicate that a large, continuous, LaBr 3 Anger-logic detector can be used to construct a high performance 3D whole-body TOF PET scanner. V. ACKNOWLEDGMENT We would like to thank Godwin Mayers and Randall Kulp from the University of Pennsylvania for constructing the pulse shaping circuits, and the research members at Saint-Gobain for their continued support. VI. REFERENCES [1] A. Kuhn, S. Surti, J. S. Karp, et al., Design of a Lanthanum Bromide Detector for Time-of-Flight PET, to appear in IEEE Trans. Nucl. Sci., vol. NS-51, pp. 2550-2557, Oct. 2004 [2] R. Allemand, C. Gresset, and J. Vacher, "Potential advantages of a Cesium Flouride scintillator for time-of-flight positron camera," J. Nucl. Med., vol. 21, pp. 153-155, 1980. [3] N. A. Mullani, D. C. Ficke, R. Hartz, J. Markham, G. Wong, System design of a fast PET scanner utilizing time-of-flight, IEEE Trans. Nucl. Sci., vol. NS-28, pp. 104-107, 1981. [4] T. F. Budinger, "Time-of-flight positron emission tomography: Status relative to conventional PET," J. Nucl. Med., vol. 24, pp. 73-78, 1983. [5] C. W. E. van Eijk, "Inorganic scintillators in medical imaging," Phys. Med. Biol., vol. 47, pp. R85-R106, 2002. [6] S. E. Derenzo, M. J. Weber, E. Bourret-Courchesne, M. K. Klintenberg, "The quest for the ideal inorganic scintillator," Nucl. Instrum. Methods A., vol. 505, pp. 111-117, 2003. [7] W. W. Moses, Current trends in scintillator detectors and materials, Nucl. Instrum. Methods A., vol. 487, pp. 123-128, 2002. [8] A. Lempicki, J. Glodo, Ce-doped scintillators: LSO and LuAP, Nucl. Instrum. Methods A., vol. 416, pp. 333-344, 1998. [9] K.S. Shah, J. Glodo, M. Klugerman, L. Cirignano, W. W. Moses, S. E. Derenzo, M. J. Weber, LaCl 3:Ce scintillator for γ-ray detection, Nucl. Instrum. Methods A., vol. 505, pp. 76-81, 2003. [10] E. van Loef, P. Dorenbos, C. W. E. van Eijk, K. W. Kramer, H. U. Gudel, "High-energy-resolution scintillator: Ce 3+ activated LaCl 3," Appl. Phys. Lett., vol. 77, pp. 1467-1468, 2000. [11] E. van Loef, P. Dorenbos, C. W. E. van Eijk, K. W. Kramer, H. U. Gudel, "Scintillation properties of LaBr 3:Ce 3+ crystals: fast, efficient, and high-energy-resolution scintillators," Nucl. Instrum. Methods A., vol. 486, pp. 254-258, 2002. [12] E. van Loef, P. Dorenbos, C. W. E. van Eijk, K. W. Kramer, H. U. Gudel, "High-energy-resolution scintillator: Ce 3+ activated LaBr 3," Appl. Phys. Lett., vol. 79, pp. 1573-1575, 2001. [13] C. W. E. van Eijk, "New inorganic scintillators- aspects of energy resolution, Nucl. Instrum. Methods A., vol. 471, pp. 244-248, 2001. [14] K.S. Shah, J. Glodo, M. Klugerman, W. W. Moses, S. E. Derenzo, M. J. Weber, "LaBr 3:Ce Scintillators for Gamma Ray Spectroscopy, " IEEE Trans. Nucl. Sci., vol. NS-50, pp. 2410-2413, 2003. [15] W. W. Moses and S. E. Derenzo, "Prospects for time-of-flight PET using LSO scintillator," IEEE Trans. Nucl. Sci., vol. NS-46, pp. 474-478, 1999. [16] S. Surti, J. S. Karp, G. Muehllehner, and P.S. Raby, "Investigation of Lanthanum Scintillators for 3-D PET," IEEE Trans. Nucl. Sci., vol. NS- 50, pp. 348-354, 2003. [17] S. Surti, J. S. Karp and G. Muehllehner, Image quality assessment of LaBr3-based whole-body 3D PET scanners: A Monte Carlo evaluation, Phys. Med. Biol., 49, pp. 4593-4610, 2004. [18] NEMA Standards Publication NU 2-2001, Performance Measurements of Positron Emission Tomographs. Rosslyn, VA: National Electrical Manufacturers Association; 2001. 0-7803-8700-7/04/$20.00 (C) 2004 IEEE 0-7803-8701-5/04/$20.00 3406 (C) 2004 IEEE