contents lists available at sciencedirect nuclear instruments

6
Free-running ADC- and FPGA-based signal processing method for brain PET using GAPD arrays Wei Hu a,b , Yong Choi a,n , Key Jo Hong a , Jihoon Kang a,b , Jin Ho Jung a , Youn Suk Huh a,b , Hyun Keong Lim a , Sang Su Kim a , Byung-Tae Kim b , Yonghyun Chung c a Department of Electronic Engineering, Sogang University, 1 Shinsu-Dong, Mapo-Gu, Seoul 121-742, Republic of Korea b Department of Nuclear Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Ilwon-Dong, Gangnam-Gu, Seoul 135-710, Republic of Korea c Department of Radiological Science, Yonsei University College of Health Science, 234 Meaji, Heungup Wonju, Kangwon-Do 220-710, Republic of Korea article info Keywords: Positron emission tomography Gamma ray signal processing Field-programmable gate array (FPGA) Data acquisition (DAQ) Analog-to-digital converter (ADC) Geiger mode avalanche photodiodes (GAPDs) abstract Currently, for most photomultiplier tube (PMT)-based PET systems, constant fraction discriminators (CFD) and time to digital converters (TDC) have been employed to detect gamma ray signal arrival time, whereas anger logic circuits and peak detection analog-to-digital converters (ADCs) have been implemented to acquire position and energy information of detected events. As compared to PMT the Geiger-mode avalanche photodiodes (GAPDs) have a variety of advantages, such as compactness, low bias voltage requirement and MRI compatibility. Furthermore, the individual read-out method using a GAPD array coupled 1:1 with an array scintillator can provide better image uniformity than can be achieved using PMT and anger logic circuits. Recently, a brain PET using 72 GAPD arrays (4 4 array, pixel size: 3 mm 3 mm) coupled 1:1 with LYSO scintillators (4 4 array, pixel size: 3 mm 3 mm 20 mm) has been developed for simultaneous PET/MRI imaging in our laboratory. Eighteen 64:1 position decoder circuits (PDCs) were used to reduce GAPD channel number and three off-the-shelf free-running ADC and field programmable gate array (FPGA) combined data acquisition (DAQ) cards were used for data acquisition and processing. In this study, a free-running ADC- and FPGA-based signal processing method was developed for the detection of gamma ray signal arrival time, energy and position information all together for each GAPD channel. For the method developed herein, three DAQ cards continuously acquired 18 channels of pre-amplified analog gamma ray signals and 108-bit digital addresses from 18 PDCs. In the FPGA, the digitized gamma ray pulses and digital addresses were processed to generate data packages containing pulse arrival time, baseline value, energy value and GAPD channel ID. Finally, these data packages were saved to a 128 Mbyte on-board synchronous dynamic random access memory (SDRAM) and then transferred to a host computer for coincidence sorting and image reconstruction. In order to evaluate the functionality of the developed signal processing method, energy and timing resolutions for brain PET were measured via the placement of a 6 mCi 22 Na point source at the center of the PET scanner. Furthermore the PET image of the hot rod phantom (rod diameter: from 2.5 mm to 6.5 mm) with activity of 1 mCi was simulated, and then image acquisition experiment was performed using the brain PET. Measured average energy resolution for 1152 GAPD channels and system timing resolution were 19.5% (FWHM%) and 2.7 ns (FWHM), respectively. With regard to the acquisition of the hot rod phantom image, rods could be resolved down to a diameter of 2.5 mm, which was similar to simulated results. The experimental results demonstrated that the signal processing method developed herein was successfully implemented for brain PET. This reduced the complexity, cost and developing duration for PET system relative to normal PET electronics, and it will obviously be useful for the development of high-performance investigational PET systems. & 2011 Elsevier B.V. All rights reserved. 1. Introduction Gamma ray signal processing performs an important role in positron emission tomography (PET), which is based on the detection of paired gamma rays emitted by radiotracers. Normally gamma rays are detected using a photo-sensor coupled with scintillator to generate a signal, which is then amplified and processed by PET electronics to obtain position, energy and time information of the detected gamma rays [1,2]. Currently, for most photomultiplier tube (PMT)-based PET systems, constant fraction discriminators (CFD) and time to digital converters (TDC) are used to detect gamma ray signal arrival time, whereas anger logic circuits and peak detection analog-to-digital converters (ADC) are implemented to acquire Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/nima Nuclear Instruments and Methods in Physics Research A 0168-9002/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2011.05.053 n Corresponding author. Tel.: þ82 2 705 8910; fax: þ82 2 706 4216. E-mail address: [email protected] (Y. Choi). Please cite this article as: W. Hu, et al., Nucl. Instr. and Meth. A (2011), doi:10.1016/j.nima.2011.05.053 Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]]]]]

Upload: others

Post on 04-Feb-2022

13 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Contents lists available at ScienceDirect Nuclear Instruments

Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]]–]]]

Contents lists available at ScienceDirect

Nuclear Instruments and Methods inPhysics Research A

0168-90

doi:10.1

n Corr

E-m

Pleas

journal homepage: www.elsevier.com/locate/nima

Free-running ADC- and FPGA-based signal processing method for brain PETusing GAPD arrays

Wei Hu a,b, Yong Choi a,n, Key Jo Hong a, Jihoon Kang a,b, Jin Ho Jung a, Youn Suk Huh a,b,Hyun Keong Lim a, Sang Su Kim a, Byung-Tae Kim b, Yonghyun Chung c

a Department of Electronic Engineering, Sogang University, 1 Shinsu-Dong, Mapo-Gu, Seoul 121-742, Republic of Koreab Department of Nuclear Medicine, Samsung Medical Center, Sungkyunkwan University School of Medicine, 50 Ilwon-Dong, Gangnam-Gu, Seoul 135-710, Republic of Koreac Department of Radiological Science, Yonsei University College of Health Science, 234 Meaji, Heungup Wonju, Kangwon-Do 220-710, Republic of Korea

a r t i c l e i n f o

Keywords:

Positron emission tomography

Gamma ray signal processing

Field-programmable gate array (FPGA)

Data acquisition (DAQ)

Analog-to-digital converter (ADC)

Geiger mode avalanche photodiodes

(GAPDs)

02/$ - see front matter & 2011 Elsevier B.V. A

016/j.nima.2011.05.053

esponding author. Tel.: þ82 2 705 8910; fax:

ail address: [email protected] (Y. Choi)

e cite this article as: W. Hu, et al., N

a b s t r a c t

Currently, for most photomultiplier tube (PMT)-based PET systems, constant fraction discriminators (CFD)

and time to digital converters (TDC) have been employed to detect gamma ray signal arrival time, whereas

anger logic circuits and peak detection analog-to-digital converters (ADCs) have been implemented to

acquire position and energy information of detected events. As compared to PMT the Geiger-mode

avalanche photodiodes (GAPDs) have a variety of advantages, such as compactness, low bias voltage

requirement and MRI compatibility. Furthermore, the individual read-out method using a GAPD array

coupled 1:1 with an array scintillator can provide better image uniformity than can be achieved using PMT

and anger logic circuits. Recently, a brain PET using 72 GAPD arrays (4�4 array, pixel size: 3 mm�3 mm)

coupled 1:1 with LYSO scintillators (4�4 array, pixel size: 3 mm�3 mm�20 mm) has been developed for

simultaneous PET/MRI imaging in our laboratory. Eighteen 64:1 position decoder circuits (PDCs) were used

to reduce GAPD channel number and three off-the-shelf free-running ADC and field programmable gate

array (FPGA) combined data acquisition (DAQ) cards were used for data acquisition and processing. In this

study, a free-running ADC- and FPGA-based signal processing method was developed for the detection of

gamma ray signal arrival time, energy and position information all together for each GAPD channel. For the

method developed herein, three DAQ cards continuously acquired 18 channels of pre-amplified analog

gamma ray signals and 108-bit digital addresses from 18 PDCs. In the FPGA, the digitized gamma ray pulses

and digital addresses were processed to generate data packages containing pulse arrival time, baseline

value, energy value and GAPD channel ID. Finally, these data packages were saved to a 128 Mbyte on-board

synchronous dynamic random access memory (SDRAM) and then transferred to a host computer for

coincidence sorting and image reconstruction. In order to evaluate the functionality of the developed signal

processing method, energy and timing resolutions for brain PET were measured via the placement of a

6 mCi 22Na point source at the center of the PET scanner. Furthermore the PET image of the hot rod

phantom (rod diameter: from 2.5 mm to 6.5 mm) with activity of 1 mCi was simulated, and then image

acquisition experiment was performed using the brain PET. Measured average energy resolution for 1152

GAPD channels and system timing resolution were 19.5% (FWHM%) and 2.7 ns (FWHM), respectively. With

regard to the acquisition of the hot rod phantom image, rods could be resolved down to a diameter of

2.5 mm, which was similar to simulated results. The experimental results demonstrated that the signal

processing method developed herein was successfully implemented for brain PET. This reduced the

complexity, cost and developing duration for PET system relative to normal PET electronics, and it will

obviously be useful for the development of high-performance investigational PET systems.

& 2011 Elsevier B.V. All rights reserved.

1. Introduction

Gamma ray signal processing performs an important role inpositron emission tomography (PET), which is based on thedetection of paired gamma rays emitted by radiotracers. Normally

ll rights reserved.

þ82 2 706 4216.

.

ucl. Instr. and Meth. A (20

gamma rays are detected using a photo-sensor coupled withscintillator to generate a signal, which is then amplified andprocessed by PET electronics to obtain position, energy and timeinformation of the detected gamma rays [1,2].

Currently, for most photomultiplier tube (PMT)-based PETsystems, constant fraction discriminators (CFD) and time todigital converters (TDC) are used to detect gamma ray signalarrival time, whereas anger logic circuits and peak detectionanalog-to-digital converters (ADC) are implemented to acquire

11), doi:10.1016/j.nima.2011.05.053

Page 2: Contents lists available at ScienceDirect Nuclear Instruments

W. Hu et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]]–]]]2

the position and energy information of detected gamma rayevents [3–8]. Recently, free-running ADC and FPGA have increas-ingly been used in combination for PET systems [9–13] and avariety of digital signal processing algorithms can be implemen-ted for gamma ray signal processing [14–19].

Geiger-mode avalanche photodiodes (GAPDs) have a varietyof advantages over PMT, most notably compactness, low biasvoltage requirement and MRI compatibility. Recently, a brain PEThas been developed using 72 GAPD arrays coupled 1:1 with LYSOarray scintillators for simultaneous PET/MRI imaging in ourlaboratory [20]. As this brain PET is based on a 1:1 individualread-out method, which differed from that of PMT-based PETsystems, normal PET electronics such as CFD, TDC, anger logiccircuits and peak detection ADC are replaced by eighteen 64:1position decoder circuits (PDCs) and three off-the-shelf free-running ADC and field programmable gate array (FPGA) combineddata acquisition (DAQ) cards [21–25].

In this study, a free-running ADC- and FPGA-based signalprocessing method is developed to detect gamma ray signalarrival time, energy and position information all together foreach GAPD channel. To evaluate the functionality of the devel-oped signal processing method, energy and timing resolution forthe brain PET have been measured. Furthermore, the hot rodphantom images have been acquired.

Fig. 1. Basic parameters for GAPD array and LYSO array.

Fig. 2. System architecture of the developed brain PET: PET scanner, GAPD–LYSO P

Please cite this article as: W. Hu, et al., Nucl. Instr. and Meth. A (20

2. Materials and methods

2.1. Brain PET using GAPD arrays

A brain PET using GAPD arrays has been developed forsimultaneous PET/MRI imaging in our laboratory [20]. Seventytwo 4�4, 3 mm�3 mm GAPD arrays (SensL, Ireland) coupled1:1 with 4�4, 3 mm�3 mm�20 mm LYSO array scintillators(Sinocera, China) constituted the full PET scanner with a ringdiameter of 330 mm and an axial field of view (FOV) of approxi-mately 14 mm. Fig. 1 shows the basic parameters for GAPD arrayand LYSO array.

In order to minimize the local heat production inside the MRIbore and the potential RF interference on PET signals for simulta-neous PET/MRI imaging, 3-m FFC cables were used to send chargesignals from GAPD arrays (inside MRI bore) to preamps (outsidethe MRI bore). To reduce system cost, particularly that of the ADCchannels, every 64 channels of GAPD output signal were sent to a64:1 position decoder circuit (PDC) [25], which detected the first-arriving signal of 64 input channels, and then sent 1 analog pulseand a 6-bit digital address to the VHS-ADC Virtex-4 DAQ cards(Lyrtech, Canada) [21–24]. Fig. 2 depicts the system architectureof the developed brain PET.

2.2. Off-the-shelf DAQ cards

For the brain PET system developed herein, a total of threeDAQ cards (VHS-ADC Virtex-4, Lyrtech, Canada) were employedto acquire 18 PDCs input signals. These DAQ cards were installed

ET detector, preamps, PDCs, DAQ cards and host computer (from left to right).

Fig. 3. Front view of free-running ADC– and FPGA-combined DAQ card.

11), doi:10.1016/j.nima.2011.05.053

Page 3: Contents lists available at ScienceDirect Nuclear Instruments

Fig. 5. Synchronization scheme of three DAQ cards.

W. Hu et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]]–]]] 3

on a host computer via a compact PCI (cPCI) interface. For eachDAQ card, there were eight free-running ADCs with a samplingrate of 100 MHz and input range of �1.25 V to þ1.25 V. An FPGA(Xilinx, USA) and a 128 MByte on-board synchronous dynamicrandom access memory (SDRAM) were used for digital signalprocessing and data recording, respectively. Two rapid channellow-voltage differential signaling (LVDS) connectors were placedon the DAQ card—one in the reception direction and one in thetransmission direction — yielding full-duplex, 1-GBps data trans-fers. The rapid channel buses were used for data transfer andsynchronization between multiple DAQ cards. The front paneldata port (FPDP) and general purpose input/output (GPIO) portwere used together to receive digital addresses from PDCs.

Fig. 3 shows a front view of the free-running ADC- and FPGA-combined DAQ card.

Fig. 4 shows the FPGA developing and data acquisition con-trolling scheme for the brain PET system. Firstly, a graphicalprogramming software (Mathworks, USA) based timing diagramsimulation was carried out to evaluate the developed FPGAalgorithm. Following a successful evaluation, the system genera-tor (Xilinx, USA) automatically converted the designed FPGA codeinto a hardware-recognizable bitstream file with optimized hard-ware resource distribution. Finally the bitstream file was down-loaded into FPGA using the ‘‘VHS control utility’’, which alsocontrolled PET data acquisition.

Fig. 4. FPGA developing and data acquisition controlling scheme for the brain PET.

Please cite this article as: W. Hu, et al., Nucl. Instr. and Meth. A (20

2.3. Synchronization for multiple DAQ cards

To acquire accurate time information of PET signals fromdifferent PDCs, three DAQ cards must be synchronized. Thesynchronization scheme is shown in Fig. 5: A 100 MHz externalclock and an external trigger were sent to three DAQ cardsthrough splitters to synchronize clock and start times for eachDAQ card. After the ‘‘data acquisition begin’’ button is pressedon the VHS control utility, the host computer sends counter-initialization signals to three DAQ cards to set all counters’ valueto 0 and initiate data acquisition. As all three of the DAQ cardswere inserted on the same cPCI interface, data on three SDRAMswere sent to the host computer sequentially at a maximumtransfer rate of approximately 70 Mbyte/s.

2.4. Implementation of digital signal processing method using

free-running ADC and FPGA

For the PET system, accurate time, energy and positioninformation for each gamma ray signal should be acquired.Recently, a simple and improved digital timing method has beendeveloped for the PET [21]. The accurate gamma ray pulse arrivaltime can be estimated by calculating the intersection of the initialrise line with the baseline.

Based on this digital timing method, CFD and TDC werereplaced by a free-running ADC– and FPGA-based signal proces-sing method as shown in Fig. 6:

1.

11)

100 MHz free-running ADCs continuously digitized GAPD out-put signals and sent them to FPGA;

2.

Parallel signal processing (arrival time detection, baselinecalculation, pulse energy calculation and channel ID genera-tion) was carried out on FPGA for gamma ray signal informa-tion generation;

3.

Delay modification was carried out to transfer generatedgamma ray signal information with the same delay timesbetween one another;

4.

Delay-modified gamma ray signal information was packagedinto one list mode format (LMF) data package;

5.

Sampling rate of packaged data was changed from 100 to25 MHz by 4:1 down-sampling to avoid exceeding SDRAMtransfer bandwidth;

6.

Energy window was applied to discriminate 511 keV eventsfrom scatter events prior to saving data packages to SDRAM.

The signal processing method was implemented as hardwareusing graphical programming-based FPGA code design as shownin Fig. 7.

By post-processing the LMF data saved on the host computer,coincidence event pairs were sorted for final PET image recon-struction. Additionally, random correction and scatter correctioncan be performed to improve image performance.

, doi:10.1016/j.nima.2011.05.053

Page 4: Contents lists available at ScienceDirect Nuclear Instruments

Fig. 6. Free-running ADC– and FPGA-based signal processing method for brain PET.

Fig. 7. Graphical programming-based FPGA code design for the implementation of the signal processing method.

W. Hu et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]]–]]]4

2.5. Evaluation of performance of the developed signal processing

method

To evaluate the functionality of the developed signal proces-sing method, the energy and timing resolutions of the brain PETwere measured by placing a 6 mCi 22Na point source in the center

Please cite this article as: W. Hu, et al., Nucl. Instr. and Meth. A (20

of the PET scanner. Furthermore the PET image of the hot rodphantom (rod diameter: from 2.5 mm to 6.5 mm) with activity of1 mCi was simulated by Geant4 application for tomographicemission (GATE) tool, and then image acquisition experimentwas performed using the brain PET. Fig. 8 shows the brain PETsetup for the evaluation experiments.

11), doi:10.1016/j.nima.2011.05.053

Page 5: Contents lists available at ScienceDirect Nuclear Instruments

W. Hu et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]]–]]] 5

3. Results and discussion

For the 22Na-22 point source experiment, after about 5 min ofdata acquisition, totally 1,000,000 single events were detectedand acquired. Fig. 9 shows representative 64-channel energyspectra of one PDC output.

Among 1152 GAPD channels, total 17 dead GAPD channelswere found. Measured average energy resolution for 1152�17¼1135 GAPD channels was 19.5% (FWHM%).

By calculating the time differences of coincidence events,system timing spectrum for the brain PET was acquired as shownin Fig. 10. The measured system timing resolution was 2.7 ns(FWHM). Due to the long cable transmission between GAPDarrays and preamps, the GAPD signal’s rise time became slower

Fig. 9. Representative 64-channel en

Fig. 8. Brain PET setup for evaluation experiments.

Please cite this article as: W. Hu, et al., Nucl. Instr. and Meth. A (20

(from 30 to 50 ns). To improve system timing resolution, a pulserise time calibration circuit is being developed to recover theGAPD signal’s rise time.

Fig. 11 shows hot rod phantom images simulated using GATE,and acquired by the brain PET. PET images were reconstructed viathe filtered back projection (FBP) method. Comparing to GATEsimulation result, there were even some ring artifacts in the outerfield of view of the acquired image using the brain PET, the rodscan be resolved down to a diameter of 2.5 mm.

Compared to PMT the Geiger-mode avalanche photodiodes(GAPDs) have a variety of advantages, including compactness, lowbias voltage requirement and MRI compatibility. Furthermore, the

ergy spectra of one PDC output.

Fig. 10. System timing spectrum of the brain PET.

11), doi:10.1016/j.nima.2011.05.053

Page 6: Contents lists available at ScienceDirect Nuclear Instruments

Fig. 11. Hot rod phantom images simulated using GATE (left), and acquired by the

brain PET (right).

W. Hu et al. / Nuclear Instruments and Methods in Physics Research A ] (]]]]) ]]]–]]]6

individual readout method using a GAPD array coupled 1:1 withan array scintillator can provide better image uniformity thanusing anger logic circuits.

The signal processing method developed in this study hasseveral advantages: the cost and the complexity of the PET systemwere reduced using PDCs and off-the-shelf DAQ cards; PETsystem development duration was reduced using the off-the-shelf DAQ cards and related development toolkits compared tothe use of custom-designed boards and hardware language(VHDL, Verilog and so on)-based development methods; variousdigital signal processing algorithms such as digital time pick-off,baseline restoration and pile-up rejection or recovery can bereadily implemented on FPGA to improve PET system perfor-mance. The principal disadvantage of this signal processingmethod was that the PET system timing resolution was affectedby GAPD output signal rise time, delay variation of PDC channels,sampling rate of free-running ADC and accuracy of digital timepick-off.

4. Conclusions

In this study, a free-running ADC– and FPGA-based signalprocessing method was developed for a brain PET system usingGAPD arrays coupled 1:1 with LYSO array scintillators. NormalPET electronics such as CFD, TDC, anger logic circuits and peakdetection ADC were replaced with PDCs and off-the-shelf DAQcards. Accurate time, baseline, energy and position information ofPET signals were detected via digital signal processing algorithmsimplemented on FPGA. Experimental results demonstrated thatthe signal processing method developed herein was implementedsuccessfully for the brain PET and high-performance phantomimages were acquired. The method herein reduced the complex-ity, cost and development duration for the PET system ascompared to normal PET electronics, and it will obviously beuseful for the development of high-performance investigationalPET systems.

Please cite this article as: W. Hu, et al., Nucl. Instr. and Meth. A (20

Acknowledgments

This study was supported by a grant of the ConvergingResearch Center Program through the National Research Founda-tion of Korea (NRF) funded by the Ministry of Education, Scienceand Technology (2010K001109), and by the Technology Innova-tion Program funded by the Ministry of Knowledge Economy(10030029), Republic of Korea.

References

[1] J.A. Sorenson, M. Phelps, Physics in Nuclear Medicine, second ed., W.B.Saunders Company, Philadelphia, PA, 1987.

[2] S.R. Cherry, J.A. Sorenson, M. Phelps, Physics in Nuclear Medicine, third ed.,W. B. Saunders Company, Philadelphia, PA, 2002.

[3] M.E. Casey, R. Nutt, IEEE Trans. Nucl. Sci. NS-33 (1) (1986) 460.[4] P.D. Cutler, E.J. Hoffman, IEEE Trans. Med. Imag 13 (2) (1994) 408.[5] M.L. Simpson, G.R. Young, R.G. Jackson, M. Xu, IEEE Trans. Nucl. Sci. NS-43 (3)

(1996) 1695.[6] J.W. Young, J.C. Moyers, M. Lenox, IEEE Trans. Nucl. Sci. NS-47 (4) (2000)

1676.[7] J.F. Pratte, G. Degeronimo, S. Junnarkar, P. O’Connor, B. Yu, S. Robert, et al.,

IEEE Trans. Nucl. Sci. NS-51 (4) (2004) 1318.[8] V.Ch. Spanoudaki, D.P. McElroy, S.I. Ziegler, Nucl. Instr. and Meth. A 564

(2006) 451.[9] M. Streun, G. Brandenburg, H. Larue, E. Zimmermann, K. Ziemons, H. Halling,

Nucl. Instr. and Meth. A 486 (2002) 18.[10] K. Ziemons, E. Auffray, R. Barbier, G. Brandenburg, P. Bruyndonckx, Y. Choi,

et al., Nucl. Instr. and Meth. A 537 (2005) 307.[11] A. Mann, B. Grube, I. Konorov, S. Paul, L. Schmitt, D.P. McElroy, et al., IEEE

Trans. Nucl. Sci. NS-53 (1) (2006) 297.[12] J. Imrek, D. Novak, Gy. Hegyesi, G. Kalinka, J. Molnar, J. Vegh, et al., IEEE Trans.

Nucl. Sci. NS-53 (5) (2006) 2698.[13] P. Guerra, J. Espinosa, J.E. Ortuno, G. Kontaxakis, J.J. Vaquero, M. Desco, et al.,

IEEE Trans. Nucl. Sci. NS-53 (1) (2006) 770.[14] R. Fontaine, M.A. Tetrault, F. Belanger, N. Viscogliosi, R. Himmich,

J.B. Michaud, et al., IEEE Trans. Nucl. Sci. NS-53 (3) (2006) 784.[15] M. Streun, G. Brandenburg, H. Larue, E. Zimmermann, K. Ziemons, H. Halling,

IEEE Trans. Nucl. Sci. NS-48 (3) (2001) 524.[16] A. Fallu-Labruyere, H. Tan, W. Hennig, W.K. Warburton, Nucl. Instr. and Meth.

A 579 (2007) 247.[17] M. Streun, G. Brandenburg, H. Larue, E. Zimmermann, K. Ziemons, H. Halling,

Nucl. Instr. and Meth. A 487 (2002) 530.[18] S.-J. Park, S. Southekal, M. Purschke, S.S. Junnarkar, J.-F. Pratte, S.P. Stoll, et al.,

IEEE Trans. Nucl. Sci. NS-55 (1) (2008) 510.[19] Q.G. Xie, C.M. Kao, X. Wang, N. Guo, C.G. Zhu, H. Frisch, et al., IEEE Trans. Nucl.

Sci. NS-56 (5) (2009) 2607.[20] J.H. Jung, Y. Choi, K.J. Hong, J.H. Kang, W. Hu, B.J. Min, et al., NSS-MIC Conf.

Rec (2009) 3556.[21] W. Hu, Y. Choi, K.J. Hong, J.H. Kang, J.H. Jung, Y.S. Huh, et al., Nucl. Instr. and

Meth. A 622 (2010) 219.[22] J. Kang, Y. Choi, K.J. Hong, J.H. Jung, W. Hu, B.T. Kim, Med. Phys. 37 (11)

(2010) 5655.[23] A. Douraghy, F.R. Rannou, R.W. Silverman, A.F. Chatziioannou, IEEE Trans.

Nucl. Sci. NS-55 (5) (2008) 2541.[24] H. Zhang, N.T. Vu, Q. Bao, R.W. Silverman, B.N. Berry-Pusey, A. Douraghy,

et al., IEEE Trans. Nucl. Sci. NS-57 (3) (2010) 1038.[25] J.H. Jung, Y. Choi, K.J. Hong, W. Hu, J.H. Kang, B.J. Min, et al., Nucl. Instr. and

Meth. A 621 (2010) 310.

11), doi:10.1016/j.nima.2011.05.053