Validation of a Monte Carlo simulation of the Inveon PET scanner using GATE

Author’s Accepted Manuscript Validation of a monte Carlo simulation of the Inveon PET scanner using GATE Lijun Lu, Houjin Zhang, Zhaoying Bian, Jianhua Ma, Qiangjin Feng, Wufan Chen www.elsevier.com/locate/nima

PII: DOI: Reference:

S0168-9002(16)30267-4 http://dx.doi.org/10.1016/j.nima.2016.04.059 NIMA58841

To appear in: Nuclear Inst. and Methods in Physics Research, A Received date: 3 October 2015 Revised date: 31 March 2016 Accepted date: 17 April 2016 Cite this article as: Lijun Lu, Houjin Zhang, Zhaoying Bian, Jianhua Ma, Qiangjin Feng and Wufan Chen, Validation of a monte Carlo simulation of the Inveon PET scanner using GATE, Nuclear Inst. and Methods in Physics Research, A, http://dx.doi.org/10.1016/j.nima.2016.04.059 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Validation of a Monte Carlo simulation of the Inveon PET Scanner using GATE Lijun Lu#, Houjin Zhang#, Zhaoying Bian, Jianhua Ma*, Qiangjin Feng and Wufan Chen* School of Biomedical Engineering and Guangdong Provincial Key Laboratory of Medical Image Processing, Southern Medical University, Guangzhou 510515, China #

Contribute equally to this article

*Corresponding author: Jianhua Ma: [email protected], Wufan Chen: [email protected]

Abstract. The purpose of this study is to validate the application of GATE (Geant4 Application for Tomographic Emission) Monte Carlo simulation toolkit in order to model the performance characteristics of Siemens Inveon small animal PET system. The simulation results were validated against experimental/published data in accordance with the NEMA NU-4 2008 protocol for standardized evaluation of spatial resolution, sensitivity, scatter fraction (SF) and noise equivalent counting rate (NECR) of a preclinical PET system. An agreement of less than 18% was obtained between the radial, tangential and axial spatial resolutions of the simulated and experimental results. The simulated peak NECR of mouse-size phantom agreed with the experimental result, while for the rat-size phantom simulated value was higher than experimental result. The simulated and experimental SFs of mouse- and rat- size phantom both reached an agreement of less than 2%. It has been shown the feasibility of our GATE model to accurately simulate, within certain limits, all major performance characteristics of Inveon PET system.

Keywords: PET; PET simulation; Inveon scanner; GATE

1. Introduction

Positron emission tomography (PET), a non-invasive molecular imaging technique, has been widely used for quantification of physiological and biochemical process in preclinical and clinical research [1]. The accuracy of quantitative studies depends on the quality of the acquired dynamic PET images [2,3,4,5,6]. However, the quality of the acquired dynamic PET images is determined by the PET imaging system, which is very sophisticated and complex. The performance of the PET imaging system greatly depends on the system parameters (e.g., the scintillation crystal size, their detection efficiency and geometrical arrangement) and methodological parameters (e.g., the

1

choice of parameters for image acquisition and reconstruction) [7]. The optimization of these parameters need numerous and extensive experiments using the PET scanner. This takes huge time and economic cost (such as the optimization of crystal size and their geometrical arrangement). Furthermore, the ground truth of the data (e.g., images) cannot be experimentally measured by the PET scanner; this poses a huge challenge for the reconstruction parameter optimization [7]. The simulation of a PET system using Monte Carlo (MC) algorithms could define an accurate model for the real PET scanner enabling acquisition of realistic simulated data that, in practice, cannot be measured experimentally. Such a model can be used for the optimization of the system geometry, the acquisition protocol, the reconstruction algorithm as well as for enhanced image quantification [8]. Geant4 Application for Tomographic Emission (GATE) is an advanced open source MC simulation platform developed by the international OpenGATE collaboration and dedicated to numerical simulations in medical imaging [9]. It is specified designed to provide realistic models of complete PET/SPECT/CT imaging systems through a user-friendly macro-command interface [9]. Recently, the GATE is also extended to support simulations of optical imaging such as bioluminescence or fluorescence imaging [10]. Using an easy-to-learn macro mechanism to configurate simple or highly sophisticated experimental settings, GATE now plays a key role in the design of new medical imaging devices, in the optimization of acquisition protocols and in the development and assessment of image reconstruction algorithms and correction techniques [11,12,13]. Several validation studies of GATE models for clinical PET and pre-clinical small animal PET have recently been completed [14,15,16,17,18,19]. For the clinical PET scanner, Gonias et al [14] validated a GATE model for the simulation of Siemens PET BiographTM 6 scanner, and compared the results with experimental data, obtained in accordance with the National Electrical Mannufacturers Association (NEMA) NU 2-2001 performance measurement protocol (National Electrical Manufacturers Association 2001). Karakatsanis et al [15] evaluated two GATE models of the commercial available PET scanner HR+ and the PET/CT Biograph 2. Furthermore, the development of an approximate dead-time model at the level of single and coincidence events was also carried out, so that the simulated count rate curve can satisfactorily match the experimental count rate performance curve for each scanner. Lamare et al [16] validated the model of the Philips Allegro/GEMINI PET scanner using a number of NEMA NU2-2001 performance protocols including spatial resolution, sensitivity and scatter fraction. In the same paper a dead

time model was also developed and validated using count rate loss and noise equivalent count rates. Schmidtlein et al [17] validated a model within GATE of the General Electric (GE) Advance/Discovery Light Speed (LS) PET scanner in accordance with the NEMA NU 2-1994 (National Electrical Manufacturers Association 1994) and NEMA NU 2-2001

2

(National Electrical Manufacturers Association 2001) protocols. In addition, several small animal PET scanners have been validated using GATE. For example, Rannou et al [18] reported the simulation results of the PET part of optical positron emission tomography (OPET) system using GATE and investigated the definition of the geometric parameters of the OPET tomography. Yang et al [19] validated the application of GATE to model a GE eXplore VISTA small animal PET system and used a realistic voxelized mouse phantom to present the overall quality of the model. Overall, all these GATE models for clinical PET and pre-clinical small animal PET have been validated based on the NEMA NU 2-1994 and NEMA NU 2-2001 for whole body PET scanner. The Siemens Inveon dedicated PET scanner is the latest generation of commercial tomography from Siemens Preclinical Solutions, Inc. Kemp et al (2009) reported the performance characteristics of the Inveon PET scanner based on the modified NEMA NU 2-2007 performance standards (National Electrical Manufacturers Association 2007). Bao et al [20] first reported its performance (including energy resolution, spatial resolution, sensitivity, scatter fraction, counting-rate and imaging capability) based on NEMA NU-4 2008 standards for small animal PET scanner (National Electrical Manufacturers Association 2008). Visser et al [21] later experimentally measured the spatial resolution using various reconstruction algorithms and sensitivity for its whole field of view (FOV) using NEMA NU-4 2008 standards. Disselhorst et al [22] further accessed the image-quality for several positron emitters using the NEMA NU-4 standards in the Inveon PET scanner. Magota et al [23] investigated the performance of the Inveon small-animal PET/SPECT/CT systems and compared the imaging capabilities of the SPECT and PET components. In contrast to the experimental performance evaluation, the validation of the Inveon PET Scanner using Monte Carlo simulation model was also reported. Boisson et al [24] validated the PET-SORTEO Monte Carlo simulation model for the Inveon PET following the NEMA NU 4-2008 standards. Furthermore, Lee et al [25] presented and validated a GATE model of the Siemens Inveon trimodal imaging platforms (including PET, SPECT, and CT). The aim of this study is to evaluate the performance characteristics of the Inveon PET scanner using a validated GATE model based on NEMA NU-4 2008 standards [26]. GATE version 6.1 is employed for the modeling while system performance evaluation is based on National Electrical Manufactures Association (NEMA) NU-4 standards, as it provides a standardized methodology for small animal PET performance evaluation [26].

In the present study, we used the NEMA NU-4 2008 sensitivity, scatter fraction, count rates, and spatial resolution protocols to validate the simulation model. The GATE output data were histogrammed into 3D sinograms and subsequently processed by the Software of

3

Tomography Reconstruction (STIR) to reconstruct 3D PET images [27]. In addition, we had studied the sensitivity for the whole FOV of the scanner. The validation was carried out against actual measurements performed on the Inveon PET scanner by strictly following the NEMA NU 4-2008 standards. Parts of this work were presented at the 2013 IEEE Nuclear Science Symposium and Medical Imaging Conference [28]. The paper is organized as follows: We start with a brief description of the system and then describe the methods for measuring spatial resolution, sensitivity, scatter fraction and counting-rate. Section 3 presents the spatial resolution, sensitivity, scatter fraction and counting-rate performances measured from simulated data and experimental data. The discussion of the measured results is given in Section 4. Finally conclusions are drawn in Section 5

2. Methods and materials

2.1 Description of the system

The Inveon PET system is a lutetium oxyorthosilicate (LSO)–based, high-sensitivity, high-resolution preclinical PET scanner The system has been installed at the PET Center, Nanfang Hospital of Southern Medical University at Guangzhou, Guangdong (Lu et al. 2014). It consists of 64 detector blocks arranged in 4 contiguous rings. Each detector block has a 20  20 LSO crystal array configuration. Each crystal element is 10.0 mm long and has a cross-sectional area of 1.511.51 mm. The crystal pitch is 1.59 mm in both axial and transverse directions. The crystal

ring diameter is 16.1cm while the transaxial and axial FOV are 10.0cm and 12.7cm, respectively. Figure 1 shows the graphical representation of the geometry of the Inveon PET system as visualized by GATE. The advantage of this system is equipped with the quicksilver asynchronous event processing architecture allowing higher speed event processing [29]. Figure 2 shows typical signal processing chain simulated by GATE used to convert the particle interactions within the detectors into coincidence counts. Raw data are acquired and stored in list-mode format. From the list-mode data, coincidence events can then be retro-actively histogrammed into two-dimensional (2D) using FORE or SSRB or into three-dimensional (3D) sinograms with different combinations of span and maximum ring differences (MRD). Random coincidence events are measured using a delayed-window technique and can be either subtracted to the prompts events or saved separately. GATE is a MC simulation platform designed to utilize the underlying well-validated physics components of Geant4. We selected to incorporate into our GATE simulation the low energy Geant4 models for the Compton, Rayleigh and photoelectric photon interactions. Another interesting feature of GATE is the ability to simulate the conversion of photon interactions into

4

digital counts in an attempt to model the detector and electronic responses of a real scanner. The energy resolution of the 511keV photopeak was set to 14.6% [20]. We applied a paralyzable approximate dead-time model in order to simulate the dead time both at singles and the coincidences level of the signal processor chain of the scanner. The dead time is set to 300ns at the singles and 500ns at the coincidence level.

Fig.1 Graphical representation of the geometry of the Inveon PET system as visualized by GATE. The LSO layer is depicted with yellow color

Fig.2 Typical signal processing chain simulated by GATE used to convert the particle interactions within the detectors into coincidence counts

2.2 Spatial resolution

In accordance with the NEMA NU-4 2008 standards, we simulated a spherical 22Na point source with a nominal size (0.3 mm) to measure the spatial resolution. The activity of the source was 198kBq. The energy window was set to 350-625keV and the coincidences time window (CTW) was 3.432ns. The source was fixed in the tomography, and located at two axial positions: (i) the center of the axial FOV and (ii) one-fourth of the axial FOV (31.75 mm away from the center along axial direction). For each of the two axial positions, the source was stepped towards the edge of the transverse FOV. For the central 5 mm of the transverse FOV, the source was stepped at 1 mm increments and then at 5 mm steps up to the edge of the FOV. The 3-dimensional (3D) sinograms were sorted into 2-dimensional (2D) sinograms by Fourier rebinning (FORE). Subsequently, the images were reconstructed using an analytic 2D filtered backprojection (FBP) (FBP2D in STIR), with the ramp filter cut off at the Nyquist frequency [27]. A zoom factor was 5

selected to achieve a 0.39 mm-pixel in-plane resolution. The axial plane separation was 0.796 mm. The dimension of the reconstructed image was 256  256  159 .

2.3 Sensitivity

An 18F point source was used to measure the absolute sensitivity of the system for different energy window settings. The 18F source was positioned at the center of the FOV and scanned for 5 minutes. The CTW and dead time were set to 3.42ns and 500ns respectively. Two classes of energy windows were examined: a) one with a fixed 350keV low-level discriminator (LLD) and a set of upper-level discriminators (ULDs) of 600, 625, 650, 700keV respectively and b) the other with a fixed 625keV ULD and LLDs of 300, 350, 400, 450 respectively. Furthermore, to provide a sensitivity illustration over the complete FOV, we used a transaxial

step size of: (i) 1 mm for the range from -5 mm up to 5 mm and (ii) 5 mm for the range from 5 mm up to 45 mm. The axial step size was 1 mm for the range of -5 to 5 mm and 5 mm elsewhere. For each position of the point source, the scan time was 5 seconds and the energy widows were 350-625keV.

2.4 Scatter Fraction and Counting-Rate Performance

Scatter fraction and counting-rate performance were measured using 2 different cylindrical phantoms of high density polyethylene to model the geometries of a mouse and a rat. The design of the phantoms conformed to the NEMA NU-4 standards. The mouse-like phantom was a 70 mm long solid cylinder with a 25 mm diameter. A cylindrical hole (diameter, 3.2 mm) was drilled parallel to the central axis, at a radial distance of 10 mm. The rat-like phantom was also cylindrical but with larger dimensions (length 150 mm; diameter 50 mm). An 18F source with a variable total activity from 10MBq to 500MBq was utilized for both mouse- and rat- size phantom, respectively. The scatter fraction was calculated by equation (1): SF 

Rs . R s +R t

(1)

While the corresponding noise equivalent counting rate (NECR) was given by equation (2): NECR=

R 2t ， R t +R s +2R r

where R s , R t , R r are the scatter, true, and random counting rates, respectively.

3. Results

6

(2)

3.1 Spatial Resolution

The spatial resolutions (FWHM and FWTM) obtained from the FORE- and 2D FBP-reconstructed images are shown in figure 3. According to NEMA-NU4 2008 standards, spatial resolution should be reported for the source positions at the center and one-fourth of the axial FOV of the scanner. The FWHM at the center and one-fourth of the axial FOV are shown in figure 3(a) and (b), respectively. For each, the experimental and simulated axial, radial and tangential spatial resolutions are measured. It is clearly seen that the radial FWHM deteriorates with the transverse offset increasing for both simulated and experimental results. In contrast to this, the tangential FWHM is relatively constant with the transverse offset increasing for both simulated and experimental results. Thus, both the radial and tangential FWHMs agreed well with the previous published work [21]. For the axial FWHM, there are some differences between the simulated and experimental results. For the range of 0-20 mm of the transverse offset, the simulated resolution is better than the experimental results. While the source position steps up to the edge of the FOV, the simulated results match well with the experimental results. Overall, the spatial resolution values obtained using the simulated model are similar with the measured results or published data with 5–18% error. The similar results of FWTM can be observed in figure 3(c) and (d).

(a)

(b)

7

(c)

(d) Fig.3 Experimental (solid lines) and simulated (dotted lines) radial, tangential and axial FWHM ((a) and (b)) and FWTM ((c) and (d)) resolutions obtained using FORE and 2D FBP. In (a) and (c) the point source was located at the center of the axial FOV and in (b) and (d) the point source was at one-fourth of the axial FOV. For each of the two axial positions, the source was stepped towards the edge of the transverse FOV

3.2 Sensitivity

Table 1 comparatively presents the simulated and experimental (reported in [20]) absolute sensitivity for different energy windows. It can be observed that the absolute sensitivity is dropped to 5.93% when the LLD reaches 450keV. Figure 4 presents the simulated absolute sensitivity for complete FOV of the Inveon scanner. We can note that the value of the absolute sensitivity is maximal at the center of the FOV. Also this result is similar to the experimental result reported in [21].

Table 1. Comparison between the simulated and experimental absolute sensitivity for different energy window settings. Energy window (keV)

Experimental results

Simulated results

350-600

6.64%

6.79%

8

350-625

6.72%

6.82%

350-650

6.74%

6.83%

350-700

6.85%

6.86%

300-625

7.86%

7.63%

350-625

6.72%

6.82%

400-625

5.95%

6.52%

450-625

4.19%

5.97%

Fig.4 Simulated sensitivity for complete FOV of the Inveon scanner. (The transaxial step size was 1 mm for the range of -5 to 5 mm and 5 mm for the range of 5-45 mm; the axial step size was 1 mm for the range of -5 to 5 mm and 5 mm elsewhere).

3.3 Scatter Fraction and Counting Rate Performance

Table 2 lists the simulated and experimental scatter fraction (reported in) for the mouse- and rat-size phantom, respectively. The simulated value is a little higher than the value determined experimentally. Figure 5 shows the plots of simulated and experimental NECR as a function of total activity for mouse- and rat-size phantom. It can be observed that, for the mouse-size phantom, the simulated and experimental NECR is very close at the range 0-300Mbq. However, for the high activity the simulated result is a little higher than the experimental result. For the rat-size phantom, the simulated result is obviously higher than the experimental result. Table 3 lists the simulated and experimental NECR peak value with the corresponding total activity (reported in [20]) for mouse- and rat-size phantom, respectively. The simulated NECR curve reaches the peak value 1666kcps with the total activity approximately 130Mbq, while the experimental NECR curve reaches the peak value 1670kcps with the total activity approximately 120MBq. Though there are some different, the results indicate a good agreement between our simulated and experimental 9

NECR peak value for mouse-size phantom. However, for the rat-size phantom, the simulated peak NECR value is obviously higher than the experimental result [20].

Table 2. Comparison of the intrinsic scatter fraction simulated and experimental values for both mouse- and rat-size phantom. SF

Mouse phantom

Rat phantom

Experimental results

7.8%

17.2%

Simulated results

9.7%

18.8%

Fig 5. Plots of simulated and experimental NECR as a function of total activity for mouse- and rat-size phantom.

Table 3. Comparison of the peak NECR (total activity) simulated and experimental values for both mouse- and rat-size phantom. Peak NECR (total activity)

Mouse phantom

Rat phantom

Experimental results

1670kcps (120Mbq)

590kcps (110Mbq)

Simulated results

1666kcps (130Mbq)

835kcps (100Mbq)

4. Discussion

4.1 Spatial Resolution

The spatial resolution of the PET imaging system depends on a number of factors from various sources: the physical characteristic of positron, the detector design (e.g., scintillation crystal size and their geometrical arrangement), and reconstruction processes (Zaidi et al. 2006). In our prevent study, the spatial resolution values obtained using the simulated model are similar with the experimental measured results or published data with 5–18% error. The systematic

10

difference can be related to: (1) the absence of any modeling within GATE during the simulation of the detection process on effects such as light sharing between PMTs and block decoding, (2) an inadequate relative placement of the point source between the experimental and simulated designs, (3) the use of different image reconstruction procedures. The underestimation of the spatial resolution is in agreement with other works [16, 17].

4.2 Sensitivity

The sensitivity shown in figure 4 was obtained for the whole FOV for an MRD of 79. The sensitivity in figure 4 is in accordance with straightforward geometric considerations based on total number of LORs contained in the FOV of the scanner. The sensitivity value decreases with the value of MRD decreasing, as illustrated in paper [21]. From figure 4, we can also observe that the sensitivity decreases differently at a fixed axial position for different radial position. This can be explained by the fact that oblique photons may be less effectively detected by crystals at the edge of detector blocks because of the gaps between blocks. Coincidences are detected even when the point source is located beyond the FOV. This effect should be attributed mostly to scatter effects in the detector crystals. The sensitivity of whole FOV, as shown in figure 4, also provide reference for researchers who wish to optimize their experiments by scanning more than 1 animal at the same time. On the basis of only sensitivity, it should be preferable to place 2 animals (e. g., mouse) at different position that have similar sensitivity. However, the sensitivity should also be balanced with attenuation and scatter correction, if these two factors are also affected by the position [21].

4.3 Scatter Fraction and Counting Rate Performance

For the scatter fraction, shown in Table 2, simulated results were consistently higher than the experimental results. The counting rate performance of our GATE model was also validated in comparison with the experimental results. Though there are some different, the results indicate a good agreement between our simulated and experimental NECR peak value for mouse-size phantom. However, for the rat-size phantom, the simulated peak NECR value is obviously higher than the experimental result. Several factors may account for this behavior: (1) the inaccuracy modeling of the scattering material, in particular the bed and the shields, in the scanner, (2) phenomena within a real detector-signal processing chain, not explicitly modeled in the simulation, includes block non-uniform energy resolutions, light spread and leakage, and PMT and optical coupling efficiencies.

11

4.4 Future work

GATE is a powerful Monte Carlo simulator that combines the advantages of the general-purpose GEANT4 simulation code and the specific software tool implementations dedicated to emission tomography. It is easy to set up the simulation models of PET/SPECT imaging systems through a user-friendly macro-command interface. The GATE has been validated for a wide range of clinical and small animal scanners. In particular, Popota et al first validate the GATE model following the NEMA NU 4-2008 standards for small animal systems [30]. The limitation of GATE is highly computationally demanding, especially when tracking particles through voxelized phantoms. To circumvent the relatively slow simulation, SimSET is combined with GATE to simulate photon interactions and transport inside a voxelized phantom [31]. Also, there are methods using parallel computational model to accelerate the simulation [32]. The present work has focused on the performance of a GATE simulated inveon PET model based on NEMA NU-4 standards. There are some discrepancies between the simulated results and the experimental measured results or published data. Future work consists of application to an more accuracy model, incorporating intrinsic activity of LSO in the PET ring, the scattering material (e.g., the bed and the shields), the attenuation materials et al.

5. Conclusion

In this study it has been shown that modeling of a commercial high-end PET preclinical system is feasible with GATE by validating simulation against experimental results. Subsequently the validated GATE model was employed to evaluate the performance of a real PET system based on NEMA NU-4 standards. The results demonstrated the ability as well as the limitations of GATE in accurately modeling the standard performance characteristics of the Siemens Inveon small animal PET system. The constructed model will be utilized in future to validate, assess and advance reconstruction algorithms and acquisition protocols of the Inveon PET system

Acknowledgment

This work was supported by the National Natural Science Foundation of China under grants No. 81501541, the Natural Science Foundation of Guangdong Province under grant no. 2014A030310243 and the Specialized Research Fund for the Doctoral Program of Higher Education under grant 20134433120017. The authors would wish to thank Profs. Uwe Pietrzyk

12

and Michaela Gaens from Institute of Neuroscience and Medicine of Jülich Research Centre, for offering the Radioactive decay patches for GATE V6.1. The authors also wish to thank Dr. Nicolas A. Karakatsanis from PET Instrumentation and Neuroimaging laboratory of Geneva University Hospital for helpful suggestions and discussions.

References: [1]. Rahmim A, Zaidi H. PET versus SPECT: strengths, limitations and challenges. Nucl Med Commun ,2008, 29(3): 193-207. [2]. Leahy RM, Qi J. Statistical approaches in quantitative positron emission tomography. Statistics and Computing. 2000, 10(2): 147-165. [3]. Reader AJ, Zaidi H. Advances in PET Image Reconstruction. PET Clinics, 2007, 2: 173-190 [4]. Lu L, Karakatsanis NA, Tang J, Chen W, Rahmim A. 3.5D dynamic PET image reconstruction incorporating kinetics-based clusters. Phys Med Biol 2012, 57(15): 5035-5055. [5]. Rahmim A, Zhou Y, Tang J, Lu L, Sossi V, et al. Direct 4D parametric imaging for linearized models of reversibly binding PET tracers using generalized AB-EM reconstruction. Phys Med Biol, 2012, 57(3): 733-755. [6]. Lu L, Ma J, Feng Q, Chen W, Rahmim A. Anatomy-guided brain PET imaging incorporating a joint prior model. Phys Med Biol, 2015, 60(6):2145-2166 [7]. Zaidi H. Quantitative Analysis in Nuclear Medicine Imaging: Springer US. 2006 [8]. Stute S, Carlier T, Cristina K, Noblet C, Martineau A, et al. Monte Carlo simulations of clinical PET and SPECT scans: impact of the input data on the simulated images. Phys Med Biol, 2011, 56(19): 6441-6457. [9]. Jan S, Santin G, Strul D, Staelens S, Assie K, et al. GATE: a simulation toolkit for PET and SPECT. Phys Med Biol, 2004, 49(19): 4543-4561. [10]. Cuplov V, Buvat I, Mesradi M, Pain F, Jan S. Optical imaging simulation using GATE. Nuclear Science Symp. Conf. Record (Anaheim, CA, 27 October - 03 November 2012), 2012 IEEE. 2614-2617. [11]. Santin G, Strul D, Lazaro D, Simon L, Krieguer M, et al. GATE: a Geant4-based simulation platform for PET and SPECT integrating movement and time management. IEEE Trans Nucl Sci, 2003, 50(5): 1516-1521. [12]. Karakatsanis NA, Lodge MA, Tahari AK, Zhou Y, Wahl RL, et al. Dynamic whole-body PET parametric imaging: I. Concept, acquisition protocol optimization and clinical application. Phys Med Biol, 2013, 58(20): 7391-7418.

13

[13]. Karakatsanis NA, Lodge MA, Zhou Y, Wahl RL, Rahmim A. Dynamic whole-body PET parametric imaging: II. Task-oriented statistical estimation. Phys Med Biol, 2013, 58(20): 7419-7445. [14]. Gonias P, Bertsekas N, Karakatsanis NA, Saatsakis G, Nikolopoulos D, et al. Validation of a GATE model for the simulation of the Siemens Biograph 6 PET scanner. Nucl Instrum Methods Phys Res A. 2007, 571: 263-266 [15]. Karakatsanis NA, Sakellios N, Tsantilas X, Dikaios N, Tsoumpas C, et al. A Comparative Evaluation of two commercial PET Scanners using GATE. Nucl Instrum Methods Phys Res A. 2006, 569: 368-372. [16]. Lamare F, Turzo A, Bizais Y, Le Rest CC, Visvikis D. Validation of a Monte Carlo simulation of the Philips Allegro/GEMINI PET systems using GATE. Phys Med Biol, 2006, 51(4): 943-962. [17]. Schmidtlein CR, Kirov AS, Nehmeh SA, Erdi YE, Humm JL, et al. Validation of GATE Monte Carlo simulations of the GE Advance/Discovery LS PET scanners. Med Phys, 2006, 33(1): 198-208. [18]. Rannou FR, Kohli V, Prout DL, Chatziioannou AF. Investigation of OPET performance using GATE, a Geant4-based Simulation software. IEEE Trans Nucl Sci, 2004, 51(5): 2713-2717. [19]. Yang CC, Seidel J, Wang Y, Lee JS, Pomper MG, et al. Validation of GATE Monte Carlo simulation of the performance characteristics of a GE eXplore VISTA small animal PET system. In: Nuclear Science Symposium and Medical Imaging Conference(NSS/MIC), 2007 IEEE. 3187-3190. [20]. Bao Q, Newport D, Chen M, Stout DB, Chatziioannou AF. Performance evaluation of the inveon dedicated PET preclinical tomograph based on the NEMA NU-4 standards. J Nucl Med, 2009, 50(3): 401-408. [21]. Visser EP, Disselhorst JA, Brom M, Laverman P, Gotthardt M, et al. Spatial resolution and sensitivity of the Inveon small-animal PET scanner. J Nucl Med, 2009, 50(1): 139-147. [22]. Disselhorst JA, Brom M, Laverman P, Slump CH, Boerman OC, et al. Image-quality assessment for several positron emitters using the NEMA NU 4-2008 standards in the Siemens Inveon small-animal PET scanner. J Nucl Med, 2010, 51(4): 610-617. [23]. Magota K, Kubo N, Kuge Y, Nishijima K, Zhao S, et al. Performance characterization of the Inveon preclinical small-animal PET/SPECT/CT system for multimodality imaging. Eur J Nucl Med Mol Imaging. 2011, 38(4): 742-752. [24]. Boisson F, Wimberley CJ, Lehnert W, Zahra D, Pham T, et al. NEMA NU 4-2008 validation and applications of the PET-SORTEO Monte Carlo simulations platform for the geometry of the Inveon PET preclinical scanner. Phys Med Biol, 2013, 58(19): 6749-6763.

14

[25]. Lee S, Gregor J, Osborne D. Development and validation of a complete GATE model of the Siemens Inveon trimodal imaging platform. Mol Imaging, 2013, 12(7): 1-13. [26]. National Electrical Manufacturers Association. NEMA Standards Publication NU 4-2008. Performance Measurements for Small Animal Positron Emission Tomographs (PETs).2008. [27]. Thielemans K, Tsoumpas C, Mustafovic S, Beisel T, Aguiar P, et al. STIR: software for tomographic image reconstruction release 2. Phys Med Biol, 2012, 57(4): 867-883. [28]. Lu L, Karakatsanis N, Ma J, Bian Z, Han Y, et al. Performance evaluation of the Inveon PET scanner using GATE based on the NEMA NU-4 standards. In: Nuclear Science Symposium and Medical Imaging Conference(NSS/MIC), 2013 IEEE. 1-4. [29]. Newport DF, Siegel SB, Swann BK, Atkins BE, McFarland AR, et al. QuickSilver: A Flexible, Extensible, and High-Speed Architecture for Multi-Modality Imaging. In: Nuclear Science Symposium and Medical Imaging Conference(NSS/MIC), 2006 IEEE. 2333-2334. [30] Popota FD, Aguiar P, Espana S, Lois C, Udias JM, Ros D, Pavia J, and Gispert JD, Monte Carlo simulations versus experimental measurements in a small animal PET system. A comparison in the NEMA NU 4-2008 framework, Phys Med Biol, 2015, 60 (1): 151-162. [31] Lin H, Chuang K, Lin Y, Ni Y, Wu J, Jan M. Efficient simulation of voxelized phantom in GATE with embedded SimSET multiple photon history generator, Phys Med Biol, 2014, 59(20): 6231-6250. [32] Rannou FR, Vega-Acevedo N, Bitar ZE, A parallel computational model for GATE simulations, Comput Methods Programs Biomed. 2013, 112(3): 673-683

15