M5BP  MIC Poster III

Friday, Nov. 6  10:30-12:30  Grand Exhibit Hall

Session Chair:  Kris Thielemans, University College London, United Kingdom

Show/Hide All Abstracts

M5BP-S2, Development of a Digital Unrestrained Mouse Phantom with Non-Periodic Deformable Motion

I. S. Klyuzhin, G. Stortz, V. Sossi

Department of Physics and Astronomy, University of British Columbia, Vancouver, BC, Canada

We describe a method that can be used to generate a digital 3D phantom of an unrestrained rodent that incorporates complex non-cyclic motion and deformation. The digital phantom is represented by a moving and deforming unstructured point cloud with known activity, attenuation, and anatomical region labels for each point. The simulated motion is generated by applying time-varying mesh deformation operators to the point cloud. To validate the simulated motion, the behavior of an unrestrained live mouse was recorded using a depth camera, and the kinematic parameters were compared between the recorded and simulated rodent motion. We also describe and test a method to correct the local activity concentration for non-rigid compression and expansion of the point cloud, to preserve the total activity in the deformed phantom. To generate the simulated emission data, the point cloud phantom is voxelized and used in the Monte-Carlo emission simulation. The combined emission and motion data can be used for the development and validation of the image reconstruction algorithms with deformable motion correction. Different tracer kinematics can be modeled by making the activity values region- and time-dependent. We have successfully tested the generated phantom data for the evaluation of a novel image reconstruction algorithm with deformable motion correction.

M5BP-3, Performance of the FlexToT V2 ASIC on the Readout of Different Detector Designs for PET

P. Rato-Mendes1, J. M. Cela1, J. M. Fernandez Varea2, L. Freixas1, L. Garrido2, D. Gascon2, R. Graciani2, J. Marin1, G. Martinez1, J. Mauricio2, J. C. Oller1, J. M. Perez1, D. Sanchez2, A. Sanuy2, I. Sarasola1, O. de la Torre2, O. Vela1

1CIEMAT, Madrid, Spain
2Universitat de Barcelona, Barcelona, Spain

A new version of the FlexToT application-specific integrated circuit (ASIC) has been designed and fabricated with an extended dynamic range and improved channel uniformity suitable for readout of different detector block designs in time of flight (TOF) positron emission tomography (PET) applications. The performance of the FlexToT v2 ASIC has been evaluated using segmented, monolithic and phoswich scintillator elements and matrices coupled to silicon photmultiplier (SiPM) arrays. The enhanced dynamic range of FlexToT v2 compared to its previous version allows the correct identification of individual crystals in scintillator matrices, both single layer and phoswich. Operation with monolithic scintillators was also demonstrated, with energy resolutions of 18% (FWHM) at 511 keV and reconstructed PET images of point sources yielding spatial resolutions on the order of 2 mm (FWHM). The results show that the FlexToT v2 ASIC is a flexible solution for the front-end readout of different designs of SiPM-based scintillator detectors in TOF-PET applications.

M5BP-7, Performance of a 0.4 mm Pixelated Ce:GAGG Block Detector with Digital Silicon Photomultiplier

E. Pratiwi1, H. T. Leem2, J. H. Park3, S. Yamamoto4, K.-M. Lim1, J.-Y. Yeom3, Y. Choi2

1Department of Medical IT Convergence, Kumoh National Institute of Technology, Gumi, South Korea
2Department of Electronic Engineering, Sogang University, Seoul, South Korea
3School of Biomedical Engineering, Korea University, Seoul, South Korea
4Radiological and Medical Laboratory Sciences, Nagoya University Graduate School of Medicine, Nagoya, Japan

Spatial resolution is important an aspect for high quality PET image. To achieve a high spatial resolution, small pixelated Cerium doped Gd3Al2Ga3O12 (GAGG) scintillators were coupled to digital silicon photomultiplier (dSiPM) for read out. This scintillator block detector consists of 24 x 24 Ce:GAGG scintillators with 0.4 mm x 0.4 mm x 5 mm elements and a 2 mm thick light guide made of acrylic resin. A dSiPM, DPC-3200-22-44, was used to measure 2-D position histogram, energy resolution and coincidence timing resolution. All measurements were conducted at sensor temperature of ~15°C with settings as follow: the validation scheme was scheme 8 (52.2 photons avg.) with validation length of 45 ns while the integration interval was 165 ns. The 2-dimensional position histogram of the Ce:GAGG block detector for Na-22 gamma photons show that most pixels were clearly resolved with average peak-to-valley (P/V) ratio 3.1. The average energy resolution of Ce:GAGG block detector was 17.7 % FWHM. At trigger scheme 1 (1st photon trigger), the coincidence timing resolution was 442 ps FHWM when acquired in coincidence with a 3 mm x 3 mm x 5 mm LYSO crystal (after skew correction). This study shows that Ce:GAGG block detector coupled with dSiPM provides good position performance and time resolution. From these results we can conclude that Ce:GAGG block detector coupled with dSiPM is a good candidate for implementation in ultra-high resolution nuclear medicine systems.

M5BP-11, 3? Imaging with Liquid Xenon: from XEMIS1 to XEMIS2

L. Gallego Manzano1, S. Acounis1, S. Bassetto2, S. Bouvier1, N. Beaupere1, P. Briend2, T. Carlier3, M. Cherel4, J.-P. Cussonneau1, R. Hamanishi5, F. Kraeber Bodéré3, P. Le Ray1, F. Lefevre1, O. Lemaire1, S. Manen6, J. Masbou1, H. Mathez7, S. Mihara8, E. Morteau1, D. Roy1, L. Royer6, L. Scotto Lavina1, M. Staempflin2, J.-S. Stutzmann1, T. Tauchi8, L. Virone1, D. Thers1

1Subatech UMR 6457, Nantes, France
2Div. of Advanced Technologies, AIR LIQUIDE, Sassenage, France
3Centre Hospitalier Universitaire de Nantes, Nantes, France
4INSERM U892 équipe 13, Nantes, France
5Dept. of Physics, Faculty of Engineering, Yokohama National University, Yokohama, Japan
6LPS Clermont-Ferrand, Clermont-Ferrand, France
7IPNL Université de Lyon, CNRS/IN2P3 UMR5822, Lyon, France
8High En ergy Accelerator Research Organization (KEK), Tsukuba, Japan

We report a new functional medical imaging technique based on the detection in coincidence of 3?-rays. The goal of this innovative medical imaging modality is to obtain a precise 3D location of a radioactive source with high sensitivity, significantly reducing the dose administered to the patient. To exploit the benefits of this technique a new detection device, based on a liquid xenon Compton telescope and a specific (e+,?) emitter radionuclide, 44Sc, are required. A first prototype of a liquid xenon TPC named XEMIS1 has been developed showing very promising results for the energy, spatial and angular resolutions of the ionization signal in liquid xenon. We achieve an energy resolution of 8.9% for a 1MeV ? source and an electric field of 1kV/cm, and a longitudinal spatial resolution smaller than 100µm. This first prototype of small dimension count on an advanced cryogenics system, which has contributed to a high liquid xenon purity with a very good pressure and temperature stability, and an ultra-low noise front-end electronics (below 100 electrons) operating at liquid xenon temperature. Based on the good results obtained with XEMIS1, a second prototype for to small animal imaging, XEMIS2, is now under construction. This new prototype that will hold around 200kg of liquid xenon, is a monolithic liquid xenon cylindrical camera, which totally surrounds the small animal. This innovative geometry will allow the detection of the 3 ?-rays with high sensitivity and high FOV. In parallel, a complete GATE/Geant4 simulation has been developed in order to understand the performances of the 3? imaging technique. The optimization of the geometry and electronics of XEMIS2 has been performed showing very promising results for the sensitivity, energy and spatial resolutions. Moreover, good quality preliminary images obtained by simulated tomographic reconstruction reveal the possibility of imaging the whole animal in a short time with a very low injected activity.

M5BP-15, Evaluation of SensL C-Series Array for PET and SPECT Applications

A. V. Stolin1, R. R. Raylman1, J. Proffitt2

1Radiology, West Virginia University, Morgantown, WV, USA
2AiT Instruments, Newport News, VA, USA

Recent advances in the development of silicon photomultipliers (SiPM) offer new opportunities for medical imaging applications. Specifically, novel imaging devices for positron emission (PET) and single photo-emission computed tomography (SPECT) are becoming feasible. In this investigation, we tested a monolithic array of new generation SiPMs, a C-series array from SensL. 4x4 array of 3-mm square SiPMs was studied with a 4-channel output signal multiplexing readout. C-series detector was coupled to various scintillation arrays and resulting prototypes were evaluated. Testing demonstrated that the new device is capable of resolving 1 mm LYSO at room temperature and 2 mm CsI crystals at 15 degrees Centigrade. Energy resolution of approximately 15 % at 511 keV and 30 % at 140 keV were obtained with LYSO and CsI crystals respectively. It is concluded that new C-series SiPMs from SensL are suitable for use in high spatial resolution nuclear medicine particle detectors.

M5BP-19, Performances of the MAGICS Compact Gamma Camera

M.-A. Verdier1, L. Pinot1, C. Esnault1, B. Janvier1, B. Aissaoui1, N. Dinu2, Y. Charon1, M.-A. Duval1, L. Menard1

1IN2P3-CNRS, Laboratory of Imaging and Modeling in Neurobiology and Oncology (IMNC), 91405 Orsay, France
2IN2P3-CNRS, Laboratory of Linear Accelerator (LAL), 91898 Orsay, France

We developed a very light hand-held high-performances gamma camera based on the recent SiPM technology and a dedicated readout electronics. The photodetection system consists of a 5 mm thick continuous LaBr3:Ce crystal, optically coupled to a 16 x 16 pixels SiPM array with a size of 3 mm each. The readout electronics allows simultaneous measurements of the 256 pixels including gain adjustments to compensate temperature variations. Position of interaction reconstruction based on analytical model fit yields a useful field of view of 45 x 45 mm² with an average spatial resolution of 0.77 mm FWHM after deconvolution of the pinhole diameter and an average bias of 0.24 mm. The average intrinsic energy resolution is 10.7% at 122 keV in the useful field of view. The photodetection system is encapsulated in a mechanical frame containing lateral lead shielding and a 15 mm thick lead collimator with 1.5 mm hexagonal holes and 250 µm septa coupled to the photodetection system. Recent improvements based on neural network algorithm for position reconstruction gives an improved average bias of 0.14 mm on the total field of view and with a computation time of 10 µs per events making this device very promising to improve the impact of miniaturized gamma cameras for radio-guided cancer surgery.

M5BP-23, (Withdrawn), Investigating the Possibility of Building an Ultra-High Resolution Small Animal PET Scanner Using An Innovative Technique

A. Cardini1, D. Brundu1,2, V. Fanti1,2, S. Pigazzini3,4, T. Tabarelli de Fatis3,4, G. Zavattini5,6

1INFN Sezione di Cagliari, Monserrato, Italy
2Dipartimento di Fisica, Universita' degli Studi di Cagliari, Cagliari, Italy
3INFN sezione di Milano Bicocca, Milano, Italy
4Dipartimento di Fisica, Universita' degli Studi di Milano Bicocca, Milano, Italy
5INFN Sezione di Ferrara, Ferrara, Italy
6Dipartimento di Fisica, Universita' degli Studi di Ferrara, Ferrara, Italy

Abstract withdrawn

M5BP-27, (Withdrawn), Optical Amplifier Based on Electro-Optic Effect for Electrical Signals and Its Application as Semiconductor Radiation Detector Preamplifier

X. Xu

N/A, Vernon Hills, IL, USA

Abstract withdrawn

M5BP-31, A Plastic Scintillator Based Beta Particle Detector for Rhizosphere Phosphorus Transport Visualization in Plants

W. Xi, A. Weisenberger, C. Zorn, S. Lee, B. Kross, J. McKisson

Nuclear Physics/Radiation Detector & Imaging Group, Thomas Jefferson National Accelerator Facility, Newport News, VA, USA

We intended to create a plastic scintillator based ß particle imaging device that could assist in understanding the symbiosis between belowground beneficial fungi and plant roots. These soil fungi living within the roots of most plants could create a mutually beneficial relationship, and help roots scavenge more nutrients and water from the soil in exchange for sugar to make the molecules they need to live and grow. Phosphorus was identified as one of such nutrients, which can be supplied from mycorrhizal fungi to its plant hosts. Phosphorus-32 was chosen as a desirable ß emitting radioisotope for this study, and a scintillation optical fiber detector was designed accordingly. The detector is based on a two-dimensional grid formed by two orthogonal planes of plastic scintillating fiber (PSF) to be readout by a position sensitive photon sensor. The PSF dimensions were optimized to maximize ß emission positioning count rate while maitaining overall detector sensitivity.

M5BP-35, Factors Affecting Spatial Resolution of a Gamma Camera Based on Monolithic Crystals

C. Borrazzo1, M. N. Cinti2, R. Pellegrini2, A. Fabbri2, R. Pani2, M. Colarieti Tosti3, P. Bennati3

1Post-graduate school on Medical Physics, La Sapienza University, Rome, Italy
2Molecular Medicine, La Sapienza University, Rome, Italy
3School for Technology and Health, KTH - Royal institute of technology, Flemingsberg, Sweden

In this work we proposed Monte Carlo simulation and experimental data with the goal to describe the factors that influence the intrinsic spatial resolution in continuous crystals for gamma imaging. Geant4 is used to simulate a crystal of 50x50 mm2 area the gamma detection and the emission of the scintillation light. We studied the spatial resolution and linearity starting from the distribution of the scintillation light. Several factors have been taken into account, such as the different value of index of refraction, optical coupling and internal surfaces treatment. Monte Carlo simulation were compared with experimental data of LaBr3(Ce) and NaI(Tl) crystals with different thickness and surface treatment. In continuous crystals spatial resolution resulted mainly dependent on the number of photoelectrons (statistical fluctuation of the distribution) and on the thickness of the crystal (light spread). The index of refraction was clearly identified as important factor affecting light spread and as consequence spatial resolution. Position algorithms appear to improve the readability of the light distribution strongly improve position linearity and minimize their error in the positioning. In conclusion, our model make easy to identify the predominant factors that influence the spatial resolution so to improve the overall performance of future gamma cameras.

M5BP-39, Design and Development of a Large Size Flat-Panel DOI PET Detector for Advanced Imaging Applications

X. Sun1, Z. Wang1,2, J. Meier1, K. Lou1, Y. Shao1

1Imaging Physics, UT MD Anderson Cancer Center, HOUSTON, TX, United States
2Physical Science and Technology, Sichuan University, Chengdu, Sichuan, China

By using latest “edge-less” SiPM arrays for DOI measurement with dual-ended-scintillator readout, we have successfully developed several compact high performance PET detectors for small animal and proton therapy imaging applications. Targeting for a Brain/Breast PET, it is still challenging to design a large size SiPM based DOI detector unit that requires good detector uniformity, compact size and high density SiPM readout. In this study, a new large size SiPM based DOI PET detector has been developed with advanced detector assembly and readout techniques. Each detector panel consists of an 32x32 array of 2x2x30 mm^3 LYSO scintillators which was optically coupled to a 20x20 array of 3x3 mm^2 SiPMs at both crystal array ends through a 1.5 mm thick optical plate. The overall effective detection volumn of each detector is 65.5x65.5x30 mm^3. Special crystal surface and inter-crystal reflector coupling were used to provide balanced enery, timing, and DOI-measurement performance. A specially designed detector holder allows <1 mm inter-panel insensitive edges, which improves overall system sensitivity when detector panels will be tiled together for assembling a PET. A large scale multiplexing circuit was designed and tested for the detector readout, reducing the output signal channels from 400 to 40 on each detector end. The signal outputs are processed through a customed high performance ASIC and FPGA based data acquisition. Initial detector evaluation showed excedllent crystal identification with all crystals were clearly separated in a Na-22 flood-source image, ~17% energy and ~2.0 mm DOI resoltuons, respectively. In conclusion: after overcoming many technical challenges, a large size DOI PET detector has been developed and partially evaluated, showing desired balanced good imaging performance for developing a high-performance practical PET. The developed detector and readout technologies can be applied for other nuclear imaging and radiation detection applications.

M5BP-43, Conceptual Design of a Combined CLI/PET Scanner via All Event-by-Event Readout

Z. Deng, G. Shao, C. Chen, H. Chen, Z. Wang, H. Hu, Y. Liu, P. Xiao, Q. Xie

BioMedical Engineering Department, Huazhong University of Science and Technology, Wuhan, China

A combined Cerenkov Luminescence Imaging (CLI) and Positron Emission Tomography (PET) scanner for in-vivo imaging of small animals is conceptually designed via all event-by-event readout and multiple-coincidence processing. Strategies and concepts for the design of the combined CLI/PET scanner based on the list-mode scheme are analyzed and modeled in terms of Monte Carlo approach. This allows for a proper inclusion of imaging via CLI modality, PET modality and combined CLI/PET modality. In the CLI modality, the photons are collected in event-by-event mode, which has been widely adopted in PET or SPECT instrument and is effective to reject background and unrelated radiation. In the combined CLI/PET modality, the Cerenkov events in CLI are taken in account for the coincidence processing of the gamma pairs in the PET modality without image fusion after reconstruction. The comparative results of the image quality among the CLI only, PET only and combined CLI/PET are shown after the simulation on a Derenzo phantom. The reconstructed CLI image shows evident blurring in the center of phantom, while the reconstructed PET image is noisy. The combined CLI/PET scanner results in dramatically improved image quality, which is evident from visual inspection of the reconstructed images as well as a quantitative evaluation in Peak Signal to Noise Rate (PSNR), Mean Square Error (MSE), and Maximum Error (ME). The MSE of CLI/PET image has almost an order of magnitude lower than CLI only or PET only image.

M5BP-47, Geometric Calibration Workflow for High Resolution Cone Beam Micro-Computed Tomography

A. Marcos1, A. Ortega1,2, M. Abella1,2, M. Desco1,2, J. J. Vaquero1

1Departamento de Bioingeniería e Ingeniería Aeroespacial, Universidad Carlos III de Madrid, Madrid, Spain
2Instituto de Investigación Sanitaria Gregorio Marañón, madrid, spain

High-resolution CBµCT imaging requires complex mechanical configurations which need advanced calibration procedures in order to obtain the best image quality. This work proposes a workflow for the complete geometrical calibration and characterization of a cone beam micro-CT system. Our implementation modifies methods described in previous works with updated enhancements. Also, an adjustment tool is introduced to facilitate the calibration task. The result is a whole protocol to characterize the geometry of the scanner that can be incorporated in the device global workflow as a periodic calibration procedure. We also describe how the calibration parameters were integrated in the acquisition protocol of the scanner. The method was tested on a new high-resolution CBµCT system designed to produce very high resolution and high soft-tissue contrast imaging in mice.

M5BP-51, A Dual-Head Germanium SPECT System

D. L. Campbell1, L. C. Johnson2, A. J. Gearhart1, R. S. Perea1, S. Shokouhi1, T. E. Peterson1

1Institute of Imaging Science, Vanderbilt University, Nashville, TN, USA
2Dept. of Radiology, University of Pennsylvania, Philadelphia, PA, USA

Position-sensitive High-Purity Germanium (HPGe) detectors are an emerging technology for biomedical applications with properties relevant to high quality imaging capability (1% FWHM at 140 keV, 1.5 mm intrinsic spatial resolution, and 1-mm depth of interaction estimation). In this work, we build a two-camera system using opposing-view HPGe cameras on a rotating gantry. The outstanding energy resolution allows for the use of narrow energy windows, facilitating multi-isotope imaging and superior scatter rejection capability, removing the necessity for scatter correction during image reconstruction. This system builds upon a previously demonstrated proof-of-concept preclinical SPECT system using a single 90-mm diameter, 10-mm thick HPGe detector with a single, 1-mm diameter tungsten pinhole collimator. The imaging capability and performance of the dual-head pinhole SPECT system is evaluated by adapting the NEMA NU-4 small-animal PET phantom for single-photon emitters.

M5BP-55, Feasibility Study for the Use of Cerenkov Radiators in Preclinical Optical Imaging

C. R. Gigliotti, L. Altabella, A. E. Spinelli

Medical Physics Department, San Raffaele Scientific Institute, Milan, Italy

Aim: Cerenkov luminescence imaging (CLI) is a technique based on detection of photons produced by ß emitting radionuclides. Cerenkov signal exiting from small animals could be improved using a Cerenkov radiator (CR), which could also reduce acquisition time. A straightforward approach to enhance Cerenkov photons production (CP) is using a high refractive index (n) material exploiting ß particles exiting from tissue, since increasing n, energy threshold for CP decreases and Cerenkov photons number produced for a fixed energy rises. However, for low energy beta particles commonly used for CLI, n is not the only parameter to take into account for CR optimization. We investigate here the efficiency of different materials using Monte Carlo (MC) simulations the effects of the material properties. Methods: GAMOS plug-in for GEANT4 was used for MC simulations. We simulated a tissue slab with a point ß source placed at a variable depth and a radiator, of variable thickness, on top of the slab. Three different radionuclides with different end-point energies were considered; 18F, 32P and 90Y. PMMA and transparent ceramic with high n were simulated as CR materials. We studied the effect on CP of n and of the density material modifying them. Results and conclusion: We observed that the simulated low density ceramic, with same n of the real ceramic, produces the highest percentage of photons. PMMA, which has lower n, shows intermediate efficiency. Considering our results, it seems that n is not to be the only crucial parameter affecting the efficiency, since the density seems to have the main role in CP at the considered energies. This is well explained considering the Bethe-Bloch equation. Comparing different ß-emitters, we observed the highest production gain for the 32P source. We conclude that ideal material for CR optimized for CLI has low density and high n and that the major enhancement of the signal is expected for medium energy ß-emitter.

M5BP-59, Monte Carlo Feasibility Study for in Vivo Small Animals Beta Detection: from Beta to Cerenkov Luminescence Imaging

L. Altabella, C. R. Gigliotti, A. E. Spinelli

Medical Physics Department, San Raffaele Scientific Institute, Milan, Italy

Cerenkov luminesce imaging (CLI) is a novel tool to image in vivo beta emitters. Bremsstrahlung (BREM) has been used for imaging and dosimetry in radiometabolic therapy. Here we investigate with Monte Carlo simulations the feasibility of beta imaging of small animals with CLI, direct (DB) and indirect beta (IB) detection and BREM using beta emitters. CLI is based on photons generated when beta particles travel with a velocity greater than speed of light in the medium. DB and IB detection are based on direct interaction with the detector or on conversion into light by means of a scintillator. This study shows pro and cons of the different physical approaches for beta emitter imaging for small animals. Monte Carlo simulations were implemented using GAMOS plug-in for GEANT4. We considered a slab of mouse muscle and a radioactive source (32P or 90Y) placed at different depths. For IB detection a 0.2 mm CsI(Tl) scintillator was superimposed to slab to convert beta in optical photons. BREM signal is the 1.7% of the number of beta exiting from slab. CLI signal is 10 times greater than beta signal. Scintillator produces a number of optical photons about 1800 times greater than beta counts. DB detection show FWHM=2.32mm for 32P and 2.44mm for 90Y at 1mm source depth, while CLI shows FWHM=3.69mm for 32P and 3.98mm for 90Y. For IB detection, we obtained a FWHM of 2.71mm with 1 mm source depth for 32P and 3.37mm for Y90. DB detection showed the best solution in term of resolution because of short beta path, but total number of collected particles is small. The use of a thin scintillator increases signal without strongly affecting the spatial resolution. BREM presents lower signal and it does not represent the best choice in small animal imaging. CLI is a good alternative to DB or IB detection, since it allows larger FOV using standard optical imaging systems with respect to beta detection. For these reasons CLI is a more flexible approach for in vivo beta emitter imaging.

M5BP-63, Fisher Information Analysis of Digital Pulse Timing

M. Ruiz-Gonzalez, L. R. Furenlid

College of Optical Sciences/CGRI, The University of Arizona, Tucson, AZ, USA

PET cameras are used to measure the concentration of a tracer injected to the body. The radionuclides emit positrons that annihilate with electrons and produce gamma-ray pairs. One important factor that determines the estimation of the position of the event is the timing resolution in the detection of pairs of gamma rays. There is a trade-off between the amount of data acquired and the timing resolution of a detector. In order to develop an efficient data acquisition system, the acquired data must be the minimum number of digital samples that contains the information required for the desired timing resolution. A simulation package was created and implemented to perform a Fisher Information analysis of simulated scintillation pulses. The result allow us to perform the timing estimation to determine the minimum amount of data that achieves the required timing resolution.

M5BP-67, Simulation Study of Sensitivity and Resolution for a Small Animal PET Ring Based on Continuous Crystals

A. Etxebeste, J. Barrio, C. Lacasta, G. Llosá, E. Muñoz, C. Solaz, P. Solevi, J. F. Oliver

Instituto de Física Corpuscular (CSIC/UV), Valencia, Spain

A first prototype of a high resolution small animal PET scanner based on continuous LYSO crystals and SiPMs was developed at IFIC. At the moment, a second version of that prototype is under development with the aim of building a full ring in the near future. In this work, we present the ongoing studies for enhancing sensitivity as well as spatial resolution. For maximizing the sensitivity, a scanner based on tapered crystals is considered. In addition, the reduction of data truncation effects is expected to improve the image quality. The scanner based on tapered crystals was characterized through Monte Carlo simulations and compared to a scanner based on cuboid detectors. Simulation results show that using transaxial tapered crystals increases the average sensitivity over the entire FOV by 14% when using a Low Energy Threshold (LET) of 400 keV. An homogeneous cylinder with 1 MBq activity covering approximately 72% of the PET scanner was simulated and reconstructed for both systems. Image Reconstruction was implemented with Filtered Back-Projection (FBP) method. Loop like structures in the transverse plane of the image reconstruction can be seen due to the dead space between detector modules for the scanner based on cuboid crystals. In the case of the scanner based on tapered crystals those structures are barely visible due to the tighter packing of detectors and the image is more uniform. Moreover, with respect to spatial resolution and following previous studies on position determination, the impact of the crystal reflector coating is also being investigated. A significant improvement in the spatial resolution including depth of interaction was shown using a specular reflector compared to diffuse reflector. Experimental studies are ongoing. In the future, we plan to extend the work to the study of spatial resolution of the scanner.

M5BP-71, Improving Spatial Resolution in Small Animal PET by Reducing the Non-Collinearity Effect: Results of Simulations and Measurements

K. Bolwin1, D. Vernekohl2, F. Büther3, J. Lühder4, K. Schäfers1

1European Institute for Molecular Imaging, University of Muenster, Münster, Germany
2Department of Radiation Oncology, Stanford School of Medicine, Stanford, USA
3Institute for Nuclear Physics, University of Muenster, Münster, Germany
4Department of Nuclear Medicine, University Hospital of Münster, Münster, Germany

Abstract—Spatial resolution of small animal PET system plays a major role in studying molecular processes in vivo. Beside other physical effects, the photon non-collinearity effect degrades reconstructed images. Since this effect is strongly depending on the detectorto- object distance we investigated in simulations and real measurements whether an optimized detector setup will improve the system performance. Therefore, we have built a prototype PET scanner by minimizing the detector distance. It could be demonstrated that this approach improves the spatial resolution both at the level of the reconstructed images (simulation) as well as in real point source measurements. Index Terms—Sub-millimeter small animal PET, Monte Carlo Simulation, wire chamber

M5BP-75, An Algorithm for Automatic Flood Segmentation Using Thin Plate Splines and Gaussian Mixture Models

G. R. Schellenberg1, G. F. Stortz2, A. L. Goertzen1,3

1Physics and Astronomy, University of Manitoba, Winnipeg, MB, Canada
2Physics and Astronomy, University of British Columbia, Vancouver, BC, Canada
3Radiology, University of Manitoba, Winnipeg, MB, Canada

A typical Positron Emission Tomography (PET) detector is comprised of a scintillator crystal array coupled to a photodetector array or other position sensitive detector. Such detectors using light sharing to read out crystal elements require the creation of a crystal lookup table (CLUT) that maps the detector response to the crystal of interaction based on the x-y position of the event calculated through Anger-type logic. It is vital for system performance that these CLUTs be accurate so that the location of events can be accurately identified and so that crystal-specific corrections, such as energy thresholding or time alignment, can be applied. While using manual segmentation of the flood image to create the CLUT is a simple and reliable approach, it is both tedious and time consuming for systems with large numbers of crystal elements. In this work we describe the development of an automated algorithm for CLUT generation that uses a Gaussian Mixture Model (GMM) paired with Thin Plate Splines (TPS). Starting from a region of stability, Gaussians are individually fit to data corresponding to crystal locations while simultaneously updating a TPS for predicting future Gaussian locations at the edge of a region of interest that grows as individual Guassians converge to crystal locations. The algorithm was tested with flood image data collected from 16 detector modules, each consisting of a 409 crystal dual-layer offset array readout by a 32 pixel SiPM array. For these detector flood images the algorithm required an average of 31.3 ± 1.2 seconds per detector to run on a single core of an Intel i7 processor. The accuracy of the method was 99.91%, with only 6 out of 6,544 crystals requiring manual adjustment. This method can be easily extended for use with other detector types through adjustment of the initial template model used.

M5BP-79, Calibration Stability in a 1mm3 Resolution, Clinical PET System and Its Impact on Real-Time Data Processing and Coincidence Sorting

D. L. Freese1, A. Vandenbroucke2, D. F. C. Hsu1, P. D. Reynolds1, D. Innes2, C. S. Levin1,2,3,4

1Electrical Engineering, Stanford University, Stanford, United States
2Radiology, Stanford University, Stanford, United States
3Physics, Stanford University, Stanford, United States
4Bioengineering, Stanford University, Stanford, United States

Real-time feedback of system performance, such as a non-diagonstic quality image to the operator, is important for clinical imaging where time is constrained. This task becomes more difficult in systems with large numbers of components. We are constructing a two-panel clinical PET system dedicated to imaging the breast that has 294,912 LYSO crystals read out by 4608 Position-Sensitive Avalanche Photodiodes (PSAPD). The system requires 907,776 calibration parameters to be estimated for an image to be produced. Applying stored parameters in real-time requires understanding their stability. We show that using previously stored calibration estimates causes no significant degradation of system measured energy resolution at 12.2±0.1%, with a photopeak shift of 3.6±0.2keV. Timing resolution is degraded by 0.3±0.2ns to 15.0±0.1ns. Crystals are correctly identified for 93.5% of events, with 3.5% of events being mispositioned within 1mm. A singles event data structure is proposed and implemented given the measured calibration stability and used to benchmark the speed of singles calibration and coincidence sorting. We demonstrate we that we are able to process our system’s output at a rate of 9.24 Mevents/s or 15.4 % slower than expected real-time. We show that this is I/O limited and that with storage upgrades we can process 16.5% faster than expected real-time at an event rate of 12.7Mevents/s.

M5BP-83, Evaluation of the Effects of PET Modules on the RF Field Distribution of an Integrated PET/RF-Coil Modality

M. S. H. Akram1, T. Obata1, M. Suga2, F. Nishikido1, E. Yoshida1, T. Yamaya1

1Molecular Imaging Center/Imaging Physics Team, National Institute of Radiological Sciences, Inage, Chiba, Japan
2Faculty of Engineering/Department of Medical System Engineering, University of Chiba, Chiba, Chiba, Japan

In this study we conducted experimental evaluation of MRI radiofrequency (RF) coil field transmission (B1) quality for an integrated PET/RF-coil modality. In this case, PET modules are very close to the imaging region (radially 3.2cm from the 22cm dia.×12cm length cylindrical phantom) and each module is positioned in between the coil elements - almost in contact with coil wires. To provide an acceptable solution for quality PET data acquisition, in our previous studies we reported design of shielding box to shield PET circuits from possible interferences from RF field. Also because of the inclusion of PET modules very close to the RF coil, it is very challenging to provide quality MR imaging. Previously we reported on quality of MR signal-to-noise ratio (SNR), static field (Bo) homogeneity and gradient coil effects. MR imaging implements three magnetic fields: (1) static field, (2) RF field and (3) gradient field. In this study we have performed RF B1 field measurements of our integrated PET/RF-coil modality in a 3 T Siemens Magnetom Verio MRI system by implementing magnitude-based double-angle method (DAM). A four-layer depth-of-interaction (DOI) detector was developed by LYSO scintillation crystal (2×2×4mm3, array size: 16×16×4; axial FOV: 32mm). 8×8 array single channel multi-pixel photon counters (MPPC, S12641PA-050; 4.1mm pitch; 1mm dead spaces) with 3×3mm2 sensitive area and ASIC readout circuit was designed. In the experiment, we took two magnitude images by implementing two gradient-echo (GRE) sequences with ?? and 2?? flip-angle RF pulses. The ratio of two images provides us the actual flip angle distribution in the imaging region. From the actual flip angle distribution we calculated the B1 field map. We have given results of B1 map for three cases: RF coil (1) without PET modules; (2) with copper shielded PET; and (3) with carbon fiber shielded PET module. We have found very similar results for copper shielded and carbon fiber shielded PET modules.

M5BP-87, Highly Multiplexed DOI PET Detector Based on SiPM Sensors

R. Chil1, G. Konstantinou1, M. Desco1,2,3, J. J. Vaquero1

1Dept. Bioingenieri?a e Ingenieria Aeroespacial, Universidad Carlos III de Madrid, Leganes, Madrid
2Instituto de Investigacion Sanitaria Gregorio Maranonn, Madrid, Madrid
3Centro de investigacion en red en salud mental (CIBERSAM), Madrid, Madrid

We present a readout circuit to read the output from two SiPM detectors, Hamamatsu-S11830-3344MF, forming an 8x4matrix of pixels, able of efficiently encoding event position, energy and depth of interaction information in five analog output channels. Position and energy resolution were measured using part of a 32x32 matrix of 1.3x1.3x12 mm3 LYSO crystals illuminated with a 22Na source. Regarding the position resolution, the system yielded a mean resolvability index of 0.31 for the center columns, and an energy resolution of 20% on the 511 keV peak. Depth of interaction was measured using a GSO/BGO phoswhich of monolithic crystals, providing clear distinction of the two different crystals. These results prove that the proposed system reduces the number of analog outputs by multiplexing the outputs from several detectors, while separating 1.3x1.3 mm2 crystals and preserving the individual energy resolution. The results also indicate that this read out topology should be able of identifying even smaller crystals, increasing the overall resolution of our system. We also prove that this system is capable of separating different crystals from a phoswich. It seems feasible to design a cost effective PET detector using this technique, with a performance equivalent or superior to that offered by standard PS-PMT.

M5BP-91, Implementation of Precise Bed Motion Control and Super-Sampling Acquisition in LaPET Scanner

Y. Li, S. Matej, J. S. Karp, S. D. Metzler

Department of Radiology, University of Pennsylvania, Philadelphia, PA, USA

A typical whole-body PET scanner performs data acquisition of a patient from multiple bed positions with overlap between adjacent bed positions. An alternative acquisition mode, continuous or near-continuous bed motion acquisition, has a number of advantages over the traditional step-and-shoot mode, including uniform axial contrast-to-noise and elimination of axial resolution artifacts. In this study, we investigate the incorporation of super-sampling techniques (in both axial and transverse directions) in the continuous bed motion acquisition to further take advantage of the acquisition mode. To implement the super-sampling data acquisition, precise motion control and synchronized data acquisition are crucial. We develop a real-time precise bed motion control based on the NI SoftMotion architecture to upgrade the manual control, and we generate digital pulses and recorded them as control-events in list-mode data for motion synchronization. A LabVIEW graphical user interface is developed to coordinate the super-sampling acquisitions with the synchronized bed motion. We adapt our blob-based list-mode reconstruction package for super-sampling reconstructions by incorporating different offsets for different data sets in the forward- and backward-projection at each iteration or sub-iteration.

M5BP-95, A Modular Hybrid PSPMT/SiPM Depth Encoding Detector for High Resolution Positron Emission Tomography

M. G. B. Sumanasena1, F. Godinez1, M. S. Judenhofer1, Q. Peng2, R. D. Badawi1

1University of California, Davis, Sacramento CA, USA
2Lawrence Berkley National Laboratory, Berkeley CA, USA

We developed a compact, modular, depth encoding detector for high resolution positron emission tomography (PET). It consists of a 3×3 array of Sensl C series 6 mm Silicon Photo-multipliers (SiPM) and the Hamamatsu C12 position sensitive photo-multiplier tube (PSPMT) coupled to either end of a 14×14 element array of unpolished 1.55 mm pitch, 20 mm thick lutetium yttrium orthosilicate (LYSO) scintillation crystals. The PSPMT provides six position encoding signals in each of the X and Y directions, which were combined by a resistive network to form two position encoding signals in each direction. Signals from the 3×3 array of SiPMs were combined to a single signal using a resistor network. The resulting five analog signals were amplified by five pre-amplifiers. In the SiPM array there is a 1.2mm gap between the active areas of two Adjacent SiPMs due to device packaging. A 1.4mm thick acrylic light guide is sandwiched between the crystal array and SiPM array to improve light collection from crystals coinciding with the dead-area. A thermistor is used to readout the temperature at the SiPM array. The detector provides FWHM DOI resolution of 2.15mm, FWHM energy resolution of 20.5% and FWHM timing resolution of 1.7ns. All crystals can be identified at all depths.

M5BP-99, PET Performance Evaluation of a RF-Penetrable PET Insert for Simultaneous PET/MR Imaging

C.-M. Chang1, A. M. Grant2, B. J. Lee3, C. S. Levin2,4,5,6

1Dept. of Applied Physics, Stanford University, Stanford, California, USA
2Dept. of BioEngineering, Stanford University, Stanford, California, USA
3Dept. of Mechanical Engineering, Stanford University, Stanford, California, USA
4Dept. of Radiology, Stanford University, Stanford, California, USA
5Dept. of Physics, Stanford University, Stanford, California, USA
6Dept. of Electrical Engineering, Stanford University, Stanford, California, USA

Integrated PET/MRI enables simultaneous measurement of molecular, functional and anatomical information of the body in one combined scan, providing physicians and researchers with multi-parameter information. However, the long-term impact of integrated PET/MRI is limited by the high cost of the current commercial systems, which require the users to purchase both PET and MR subsystems, which are permanently integrated. We are developing an RF-penetrable PET insert technology to address this challenge, and a prototype brain-dedicated insert system has been built to evaluate the technology. The insert system consists of 16 detector modules assembled into a ring of an inner diameter of 32 cm. A total number of 2,048 3.2 × 3.2 × 20 mm3 LYSO crystals coupled one-to-one to 2,048 SiPM pixels are implemented in the system. An intrinsic spatial resolution of below 2.3 mm has been achieved by measuring the coincidence point spread functions of a 500 µm positron-emitting point source with two electronically collimated detectors. A custom resolution phantom with hot rods ranging from 2.8 to 5.2 mm diameter has been acquired and reconstructed to evaluate the image spatial resolution of the system. The sizes of the smallest resolvable hot rods in the reconstructed images were 2.8 mm and 4.2 mm, when the phantom was placed at the center of the field-of-view and 9-cm off-center trans-axially, respectively. An energy resolution of 15.6 % FWHM at 511 keV and a coincidence time resolution of 5.2 ns FWHM have been achieved, limited by the outdated 2008 SiPM technology employed. The variation of the energy resolution and coincidence time resolution stays within a range of 0.5 % and 80 ps over 3 hours, demonstrating the stability of the system.

M5BP-103, A SiPM-Based PET-ToF Demonstrator Featuring a Highly Integrated Readout and DAQ System

R. Bugalho1, L. Ferramacho1, A. Di Francesco2, C. Leong1, T. Niknejad2, M. Rolo2, J. C. Silva2, R. Silva2, M. Silveira1, S. Tavernier3, J. Varela2, C. Zorraquino2

1PETsys Electronics, Oeiras, Portugal
2LIP, Lisbon, Portugal
3Vrije Universiteit, Brussel, Belgium

A SiPM-based PET-ToF demonstrator featuring a highly integrated readout and DAQ system has been developed. The PET ring comprises 24 compact Detector Modules. Each module integrates 128 gamma-ray detection pixels of 3.1x3.1x15 mm3 LYSO crystals associated to MPPC photosensors. The module has two 64-channel readout TOFPET ASICs integrating signal amplification and discrimination circuitry and high-performance TDCs for each channel, featuring 25 ps r.m.s intrinsic resolution and fully digital output. The data acquisition system comprises three frontend digital boards, each collecting the data of 1024 channels (8 Detector Modules) and transmitting assembled data frames through a serial link (1.6 Gb/s), and a single DAQ board connected to the PCIe bus of the acquisition PC. In this work, we present results on the performance of the system with a partially assembled ring (16 Detector Modules) corresponding to 2048 SiPM readout channels.

M5BP-107, Improved Image Quality Using Monolithic Scintillator Detectors with Dual-Sided Readout in a Whole-Body TOF-PET Ring: a Simulation Study

V. Tabacchini1, S. Surti2, G. Borghi1, B. J. Peet1, J. S. Karp2, D. R. Schaart1

1Dept. of Radiation Science and Technology, Delft University of Technology, Delft, The Netherlands
2Dept. of Radiology, University of Pennsylvania, Philadelphia, PA, United States

We have built and characterized a monolithic detector based on a 32 mm x 32 mm x 22 mm LYSO:Ce crystal read out by digital photon counter (DPC) arrays coupled to the crystal front- and back surface in so-called dual-sided readout (DSR) configuration. In this work we evaluate its potential for high resolution whole-body time-of-flight (TOF) PET applications and compare it to conventional back-side readout (BSR). For this purpose we perform Monte Carlo simulations of clinical PET rings based on such detectors and use the results of our experimental characterization to model their spatial response. Using clinically relevant metrics, we quantify the improvement in image quality obtained with DSR over conventional BSR. Finally, for a comparison with traditional pixelated PET detector design, we simulate rings of pixelated detectors. By varying the cross section of the individual crystals we determine what reduction in pixel size would be necessary to obtain an improvement in image quality equivalent to that provided by the additional photosensor in the DSR monolithic detector.

M5BP-111, Development of a Capacitive Charge Division Circuit for Time-of-Flight (TOF) PET Detector

H.-J. Choe, Y. Choi, J. H. Jung

Department of Electronic Engineering, Sogang University, Seoul, South Korea

An individual channel readout method is used to achieve excellent timing performance with employing a pixelated photosensor, however, it leads to complex signal readout and processing schemes. Various multiplexing methods have been proposed to reduce the number of readout channels, but traditional multiplexing methods, such as a resistive charge division network, severely degrade a coincidence resolving time. The purpose of this study was to develop a multiplexing circuit that can reduce the number of readout channels without substantial degradation of timing performance. A proof-of-principle TOF-PET detector module was constructed using a 6 × 6 array of 2 × 2 × 20 mm3 LYSO scintillators and a 4 × 4 array of 3 × 3 mm2 MPPCs. A capacitive charge division (CCD) circuit, which reduced 64 anode signals of the four detector modules to four outputs, was designed and fabricated. Energy spectra and flood image were measured to evaluate the performance of the TOF-PET detectors employing the CCD circuit developed in this study. Coincidence timing spectrum was also acquired and compared with that measured by the individual readout method. The average energy resolution was 14.1±2.3% (n = 144). One-hundred forty-four crystal pixels were clearly identified in the flood image. Coincidence resolving times were 475±2 ps FWHM and 345±1 ps FWHM with and without the CCD circuit, respectively. Experimental results indicate that the developed CCD circuit is useful to measure the TOF information as well as to reduce the number of readout channels for the development of TOF-PET.

M5BP-115, Development of Stationary Dedicated Cardiac SPECT with Multi-Pinhole Collimators on a Clinical Scanner

H. Liu1,2, T. Ma1,2, S. Yao3, J. Wu4, S. Chen1,2, S. Wang1,2, Y. Liu1,2

1Department of Engineering Physics, Tsinghua University, Beijing, China
2Key Laboratory of Particle & Radiation Imaging (Tsinghua University), Ministry of Education, Beijing, China
3Department of nuclear medicine, Chinese PLA General Hospital, Beijing, China
4Department of Diagnostic Radiology, Yale University, New Haven, CT, USA

The existing SPECT system for nuclear cardiac imaging suffer from low sensitivity or high cost. We propose a novel design of the stationary dedicated cardiac SPECT combined with multi-pinhole collimators for high performance, based on the clinical dual-head scanner with low cost. Previous simulation study has demonstrated the feasibility of the design. In this study we report the system development and experimental results. Two tungsten collimators were fabricated and mounted to variable-angle dual-head SPECT ImageE NET 632. Accurate geometrical calibration was implemented. The projection of a 2-mm-diameter sphere source, mounted and moved through three certain gridding scanning orbits, was acquired. The geometrical parameters were estimated by fitting the experimental projection centroids to the geometrical model. The reconstruction adopted MLEM algorithm with GPU acceleration based on CUDA. According to the large pinhole size, two analytical pinhole model were proposed to derive the system matrix on the fly during reconstruction. The sphere source projection of one view was reconstructed with the two analytical pinhole models and delta model. The reconstruction results of different views’ projection were compared. With geometrical calibration, the reasonable reconstruction images were achieved with 10.6 × 12.9 mm2 spatial resolution (FWHM). Analytical pinhole models improved the spatial resolution to 6.8 × 8.0 mm2 and suppressed the deformation while asymmetric Gaussian model were slightly better than sampling pinhole model. There was no significant difference between the reconstruction images with one-view and three-view projection, which indicated the stationary imaging property of the SPECT system. In conclusion, our initial results show reasonable imaging performance could be achieved on the developed stationary multi-pinhole dedicated cardiac SPECT. In the future work, the proper phantom and patient experiments would be implemented to evaluate the SPECT system.

M5BP-119, Impact of DOI Resolution and Ring Diameter on the Imaging Performance of DOI-PET

Q. Ye1,2, T. Ma1,2, X. Wang3, Q. Wei1,2, H. Liu1,2, Y. Liu1,2

1Department of Engineering Physics, Tsinghua University, Beijing, China
2Key Laboratory of Particle & Radiation Imaging, Ministry of Education, Tsinghua University, Beijing, China
3Department of Nuclear Medicine, Navy General Hospital of PLA, Beijing, China

DOI-PET is capable of achieving high resolution by reducing the impact of the parallax effect. One of the most cost-efficient and common methods to achieve DOI information is using multi-layer crystal arrays. A quantitative cost-effect performance of the multi-layer DOI-PET system require would help to optimize the design of the DOI detectors between the imaging performance and hardware complication. In this paper, we simulated the DOI-PET systems of different DOI resolution and different ring diameters to evaluate the spatial resolution performance in radial, tangential and axial directions. To minimize the impact of reconstruction algorithm parameters, The images were reconstructed using a simple fully 3D list-mode back-projection method based on Siddon algorithm. The spatial resolution is defined as the full width at half maximum (FWHM) of point spread function (PSF). The radial resolution booms with the radial distance increases, while the tangential resolution changes slightly and the axial resolution is almost the same. The system with more layers shows better spatial resolution performance. The 2-layer and 3-layer DOI-PET have good enough performance, 18.4 % and 22.6% averagely better than the conventional PET system in the radial direction. A 2-layer DOI-PET with ring diameter of 750 mm has similar spatial resolution with the conventional PET with the ring diameter of 802 mm. Taking the two factors into account, to acquire better performance and lower cost, we propose to choose 2 or 3-layer DOI-PET system with ring diameter smaller than 802 mm.

M5BP-123, Three-Dimensional Angiography Using Mobile C-Arm with IMU Sensor Attached: Initial Study

A. Moataz1, A. Soliman1, A. M. Ghanem1, M. al-Shatouri2, A. Atia3, E. A. Rashed4

1Faculty of Engineering, Suez Canal University, Ismailia, Egypt
2Dept. of Radiology, Faculty of Medicine, Suez Canal University, Ismailia, Egypt
3HCI-LAB, Faculty of Computers and Informatics, Helwan University, Helwan, Egypt
4Image Science Lab., Dept. of Mathematics, Faculty of Science, Suez Canal University, Ismailia, Egypt

Three-dimensional (3D) computed tomography (CT) imaging is becoming an essential demand in several clinical procedures. Mobile C-arm is a useful imaging tool for image-guided interventional radiology. C-arm systems are provided with X-ray image intensifier (XRII) or flat-panel detectors. Essentially, C-arm CT systems requires scanners with flat-panel detectors for its ability to provide homogenous image quality and improve the resolution of low-contrast subjects compared to those equipped with XRII. However, C-arm systems with XRIIs are widely used in several interventional procedures. Such systems can provide a high quality two-dimensional (2D) fluoroscopic images that facilitates minimal invasive surgery. However, it is unable to provide depth information for 3D imaging due to several factors. First, the gantry of XRII-based C-arms is usually operated manually, where the rotation angle is determined using printed angle scale attached to the scanner gantry. Second, the gantry orbital rotation is normally limited to angular range less than theoretically required for exact 3D reconstruction. Third, considering the offset-scan geometry, which is common configuration in mobile C-arm with XRII, the number of rays passing through field-of-view (FOV) is limited. In this paper, we develop a 3D angiographic imaging system using commercial C-arm system equipped with XRII. First, an in-house made gantry rotation unit is developed to control the scanner orbital rotation. Second, the gantry rotation is traced using inertial measurement unit (IMU) sensor attached to the scanner gantry. Geometry information obtained from IMU sensor are used to define the gantry position in the 3D space and synchronized with detector measurements. The SCAN algorithm is used for the 3D reconstruction and achieved results are of high quality.

M5BP-127, Estimation of Effective Atomic Number by Large-Energy Gap Dual Energy CT Reconstruction

D. Sakata1,2, A. Haga2, S. Kida3, Y. Masutani4, K. Nakagawa2

1Department of Physics, Japanease Foundation for Cancer Research, Tokyo, Japan
2Department of Radiology, The University of Tokyo Hospital, Tokyo, Japan
3Faculty of medicine, Gunma Univsesity, Gunma, Japan
4Faculty of Information Sciences, Hiroshima City University, Hiroshima, Japan

Computed Tomography (CT) is a method to produce slice image of specific volume from the x-ray projection images. The contrast of CT image is correlated with the attenuation coefficient of the x-ray in the object. The attenuation coefficient is strongly dependent on the x-ray energy, the effective atomic number and density of the object. Therefor it is possible to reconstruct effective atomic number and density by two attenuation coefficient with different energies. In this study, we will propose method to extract information of atomic number and density using “large-energy gap” dual energy CT reconstruction. We can calculate effective atomic number by solving the coupled equation between the two attenuation coefficients with different energies. It is important to reconstruct attenuation coefficient correctly. In this study, we reconstruct expected attenuation coefficient via calibration of attenuation coefficient for each energy. To improve accuracy of effective atomic number and density, we take into account energy spectrum of CT modality in the calibration. For this study, we use the megavoltage (MV) and kilovoltage (kV) x-rays of Elekta Agility system as the dual source x-ray. And we estimate energy spectrum of these modality by GEANT4. We use two types of phantoms, one is a phantom of CIRS for calibration of attenuation coefficient and the other is CatPhan to check performance of dual-energy CT reconstruction. To reconstruct attenuation coefficient distribution it is used MLEM algorithm for CBCT. We report effective atomic number distribution and electron density distribution by well-calibrated dual-energy CT reconstruction method. The accuracy of atomic number and electron density is in 10%. We developed the MV and kV dual-source CBCT reconstruction to yield the effective atomic number and electron density distribution. This technics will arrow to calculate realistic absorbed dose on Monte Carlo simulation.

M5BP-131, Response Function Estimation for the Pilatus 200K Photon Counting Detector

L. Zhang1,2, D. Wu1,2, X. Zhu1,2, X. Xu1,2

1Key Laboratory of Particle & Radiation Imaging (Tsinghua University), Ministry of Education, Beijing, China
2Department of Enginerring Physics, Tsinghua University, Beijing, China

The energy response function of photon counting detector is not ideal due to charge sharing effect and limited energy resolution. Its measurement and estimation are useful for the analysis of system noise and optimization of the imaging system. In this work we estimate the response function of a Pilatus 200K silicon photon counting detector in the energy range of 5keV – 30keV by X-ray fluorescence of thin metal sheets composed of Cu, Zr, Mo, Ag, and Sn. Threshold scan for each sample is carried out from 3keV to 30keV with 0.2keV intervals and the integral spectrum is fitted to the experienced response function model with differential evolution algorithm. Pre-calculated basis response functions are used to accelerate the calculation of the response function. Fitting results show good linear relationship between the charge sharing effect, the square of energy resolution and the incident energy. It also indicates that charge sharing effect grows stronger as the incident energy increases. The response function model provides a good way to interpolate spectrums from the measured ones and can be further used for system analysis.

M5BP-135, Efficient Spectral CT Material Decomposition Using the Gauss-Newton Algorithm

D. S. Rigie, P. J. La Riviere

Radiology, University of Chicago, Chicago, IL, United States

We present an efficient, data-domain material decomposition method for spectral CT that achieves nearly the same estimation accuracy as maximum-likelihood methods and converges in only a few iterations. By setting up the material decomposition as a nonlinear, least-squares problem in the logged data, we can apply the highly efficient Gauss-Newton algorithm. Surprisingly, we find that this algorithm even outperforms Newton's method in this task, despite only requiring first-order derivatives. This scheme allows for very rapid material decomposition with essentially no sacrifice in estimation accuracy, which is important for clinical utilization. Additionally, this extremely fast solver may be a useful component in developing more general image-reconstruction algorithms.

M5BP-139, A Study of Sparse Detector Designs with Interpolation for Multi-Slice Spiral CT

D. M. Coelho, K. Mueller

Computer Science, Stony Brook University, New York, United States of America

In multi-slice spiral CT, z-sampling has a great effect on image quality, dose and speed. In our work, we explore using sparse detectors of various designs in multi-slice spiral CT. These designs have fewer detector elements as compared to a regular detector. Or alternatively, they have the same number of detectors but provide more axial coverage for greater scanning speed of longer objects. The sparser detector elements also affect the reconstruction in a similar manner as z-sampling does. To account for the missing detector elements we interpolate the acquired data to simulate a full detector. We apply both bilinear interpolation and directional interpolation and compare them. We also report the quality of the reconstruction, the amount of dose and scan speed for each detector pattern.

M5BP-143, Metal Artifact Reduction Algorithm Using Directional Interpolation by Sinogram Decomposed Derivatives

H. Nam1, J. Baek1,2

1Yonsei Institute of Convergence Technology, Yonsei University, Incheon, Korea
2School of Integrated Technology, Yonsei University, Incheon, Korea

Abstract: High-density objects such as metal prostheses, surgical clips, or dental fillings generate streak-like artifacts in computed tomography images. In this study, we present a novel method for cone-beam CT metal artifact reduction by approximate missing information into the corrupted sinogram. Methods: We used sinogram decomposition to compute the structural tensor of the sinogram to estimate the direction. The proposed algorithm consists of two main steps. The first step decomposes the sinogram to estimate rough structure of the sinogram on the metal corrupted part. Using the decomposed sinogram, we estimate the missing sinogram on the second part. Results: To validate the superiority of the proposed algorithm, we used both numerical simulated data and the experimental phantom data. For simplicity, we used 2 dimensional Shepp-Logan phantom for the numerical simulation with parallel beam geometry. For reality, we used 3 dimensional cone-beam CT geometry for the Rando phantom. We implanted two metal implants in the Rando phantom, and scanned both with and without metal implants for the quantitative evaluation. Conclusion: We compared the result of our algorithm with metal artifact reduction with different interpolation methods. We showed that our algorithm outperformed other interpolation algorithms such as linear, moving least squares, and adaptive moving least squares algorithms.

M5BP-147, A Geometrical Calibration Method by Using Only Two Small Balls for X-Ray Intraoral Digital Tomosynthesis

L. Li, Y. Yang, Z. Chen

Department of Engineering Physics, Tsinghua University, Beijing, China

The X-ray tomosynthesis technique has become a popular clinical imaging in clinical imaging modality, such as the breast and lung tomosynthesis. For high quality image reconstruction, it is essential to know the accurate geometric definitions of the scanning system which are referred to the relative positions between the focal spot of the x-ray tube and the center of the detector. Some methods have reported to measure or calibrate the geometric parameters of the tomosynthesis and CT. However, most of these methods needed complicated calibration phantoms and were not available for real-time calibration during patient scanning. This paper focuses on the geometrical calibration problem for X-ray intraoral digital tomosynthesis where the X-ray tube moves along the straight line or arc, and the flat-panel detector and the object stay static. A new calibration method is presented which only uses two positioning balls at least locating in the plane parallel to the detector. With three preconditions or prior information, all of the X-ray source positions can be calculated. It is easy to use and also suitable for other tomosynthesis applications like breast tomosynthesis. The results of the numerical simulation validate this calibration method.

M5BP-151, X-Ray Induced Luminescence Imaging with Europium-Doped Nanophosphors: Simulation and Experimental Validation

B. Quigley, S.-H. Cheng, J. Souris, C.-T. Chen, C. Pelizzari, S. Kron, L.-W. Lo, P. J. La Rivière

Medical Physics, The University of Chicago, Chicago, IL, USA

In this work, we develop and validate an imaging model for x-ray induced luminescence (XIL). In x-ray induced luminescence, lanthanide-doped nanoparticles emit optical photons under x-ray irradiation. Physical x-ray induced luminescence measurements were acquired by inserting a capillary tube containing 10 mg of Yttrium oxide nanophosphors doped with europium 3+ into a mouse phantom. The x-rays were generated with a small animal IGRT unit and the luminescence measured with a macro lens attached to a cooled CCD monochrome camera.

Our simulation sought to model the process by calculating x-ray dose to a given concentration of nanophosphors at depth in tissue. Photon diffusion modeling determined the fraction of luminescence that escapes the mouse phantom. The optics of the lens and the CCD camera parameters determined the fraction of luminescent signal that was detected and converted into ADU. The simulation predicted a measured luminescent signal of 45 ADU while the average measured voxel value was 114 ADU. The relatively good agreement of these values indicates that the general structure of the model is sound. The remaining factor of two discrepancy is likely due to inaccurate assumptions regarding luminescence efficiency that we will seek to measure by other means.

M5BP-155, Energy Spectrum Extraction and Optimal Imaging via Dual-Energy Material Decomposition

W. Zhao1, L. Wan2,3, B. Zhang4, Q. Zhang2,3, Z. Xiong2,3, T. Niu5

1Wuhan Jiubang Technology Co., Ltd, Wuhan, Hubei, China
2Wuhan Riverine Technology Co., Ltd, Wuhan, Hubei, China
3Department of Biomedical Engineering, Huazhong University of Science and Technology, Wuhan, Hubei, China
4Wuhan digital PET Technology Co., Ltd, Wuhan, Hubei, China
5Institute of Translational Medicine, Zhejiang University, Hangzhou, Zhejiang, China

Limited soft-tissue contrast resolution is a major limitation of current CT scanners. The aim of the study is to improve the contrast resolution of CT scanners using dual-energy acquisition. Based on dual-energy material decomposition, the proposed method starts with extracting the outgoing energy spectrum by polychromatic forward projecting the material-selective images. The extracted spectrum is then reweighted to boost the soft-tissue contrast. A simulated water cylinder phantom with inserts that contain a series of six solutions of varying iodine concentration (range, 0-20 mg/mL) is used to evaluate the proposed method. Results show the root mean square error (RMSE) and mean energy difference between the extracted energy spectrum and the spectrum acquired using an energy-resolved photon counting detector(PCD), are 0.044 and 0.01 keV, respectively. Compared to the method using the standard energy-integrating detectors, dose normalized contrast-to-noise ratio (CNRD) for the proposed method are improved from 1 to 2.15 and from 1 to 1.88 for the 8 mg/mL and 16 mg/mL iodine concentration inserts, respectively. The results show CT image reconstructed using the proposed method is superior to the image reconstructed using the standard method that using an energy-integrating detector.

M5BP-159, Energy Dispersive X-Ray Diffraction System as a Promising Virtual Biopsy in Mammography

F. Marticke1,2, G. Montemont1, C. Paulus1, J. I. Mars2, O. Michel2, L. Verger1

1CEA LETI, MINATEC Campus, F-38054 Grenoble, France
2CNRS, Gipsa-Lab, F-38000 Grenoble, France

X-ray diffraction is an imaging technique, which provides information about the molecular structure of the sample. This kind of information is more tissue specific than information that is obtained by conventional mammography. Hence, a virtual biopsy based on X-ray diffraction instead of an invasive breast biopsy seems to be promising. Since X-ray diffraction suffers from low sensitivity, it is important to optimize the acquisition system for the given application. We propose to use energy dispersive X-ray diffraction (EDXRD) with the incident X-ray beam to be a very thin pencil beam. Thus, the principal element to optimize is the secondary collimation. In this study, we present a strategy to optimize the system and the performances of two types of optimized collimation systems. Sensitivity and different kinds of resolution were assessed by detective quantum efficiency (DQE) calculations. In order to evaluate the discrimination power between fibroglandular tissue and pure carcinoma, different EDXRD spectra of different types of phantoms were simulated, and contrast to noise ratio (CNR) and receiver operation characteristics (ROC) calculations were realized. It was shown in particular that fibroglandular tissue could be distinguished from cancerous tissue by respecting the delivered dose constraints. To conclude, EDXRD seems very promising as a virtual biopsy tool in mammography.

M5BP-163, A Monte Carlo Simulator Dedicated to a Time-Resolved Optical Tomographic Modality Based on the Henyey-Greenstein Phase Function

A.-N. Rapsomanikis1, A. Eleftheriou1, M. Mikeli1, C. Pafilis1,2, M. Zioga1, E. Stiliaris1,3

1Physics Department, National & Kapodistrian University of Athens, Athens, Greece
2EEAE, Greek Atomic Energy Commission, Agia Paraskevi, Athens, Greece
3Institute of Accelerating Systems & Applications (IASA), Athens, Greece

The current work focuses on the Time-Resolved Optical Tomographic (TROT) modality and mainly studies the scattering process that takes place in a diffusing medium like tissue when Near-Infrared radiation propagates through it. Additionally, it handles the time filtering which is necessary to be performed on the detected photons in order to gain the anatomical information of an area under investigation. The Near-Infrared radiation that is utilized in this type of imaging technique is mainly followed by Mie (forward peaked) type of scattering which causes the majority of the photons to get diffused. A small fraction of those photons remain forward scattered and only by proper time filtering can be resolved; they provide the correct anatomical information of the medium. In order to investigate this mechanism, a Monte Carlo simulator program, named PhoSim, has been developed. This software package is based on the Henyey-Greenstein phase function which in a macroscopic way can undoubtedly approximate the microscopic character of the process that gives rise to Mie type of scattering.

M5BP-167, In Vivo and in Vitro Imaging Using a Multimodal Optical System

C. R. Gigliotti1, L. Altabella1, F. Boschi2, M. Crippa3, A. E. Spinelli1

1Medical Physics Department, San Raffaele Scientific Institute, Milan, Italy
2Department of Computer Science, University of Verona, Verona, Italy
3Division of Genetics and Cell Biology, San Raffaele Scientific Institute, Milan, Italy

Aim: Multimodal imaging systems are becoming necessary tools for preclinical studies. Different combined machines are commercially available, but they are not fully customizable according to specific needs. Preclinical studies often involve not only in vivo imaging, but also in vitro cells analyses, usually performed with dedicated microscope. We proposed a system composed of a cooled EMCCD mounted on a light-tight box that allows multimodal detection (Spinelli et al., Biomed. Opt. Exp., in press) allowing modifying the experimental set-up. We performed here a feasibility study using Monte Carlo simulation (MCs) of a prototype made of CCD with a macro lens, to perform bioluminescence microscopy, considering an acquisition method and a data processing to improve cells resolution. Methods: Images of bioluminescent cells, in particular several frames of few seconds and long exposure (300 sec) benchmark acquisition were simulated using GAMOS plug-in for GEANT4. The data are then processed as follow. The signal integrated over each frame is used as a prior for spatial localization of the main sources. After a Gaussian smoothing of each frame, a Gaussian fit of each source is performed. The location and intensity of the peak for each fit were then exploited to reconstruct a high-resolution image. Results and conclusion: Processing of the simulated frames allowed halving the FWHM of cells signal with respect to the image that simulates a long acquisition. In particular, FWHM shrank to physical dimension of the simulated cells (50µm). The results obtained with MCs showed that the proposed acquisition method and analysis procedure allow reducing FWHM of a factor 2. More importantly, it is feasible to use a unified optical platform for small animal and microscopic bioluminescence imaging.

M5BP-171, Performance Assessment of CdZnTe Pixellated Detectors for Quantitative Multi-Spectral Photon-Counting X-Ray Imaging

D. G. Darambara

Physics/Radiotherapy and Imaging, Institute of Cancer Research / The Royal Marsden Foundation Trust, London, UK

We have been investigating – theoretically and experimentally - to which extent novel, energy-sensitive CdZnTe pixellated detectors and a dedicated fast photon-counting readout ASIC with dynamic energy binning are capable of overcoming fundamental performance limits and challenges of photon-counting multi-spectral x-ray imaging, and therefore, to reveal the real benefits from the additional spectral information and facilitate the realisation of this application type in clinical practice. Photon-counting multi-spectral x-ray imaging can become an important technique with added clinical value providing material-specific quantitative information and providing molecular information of tissue structure superimposed on anatomy. This dynamic hardware approach provided us with the means to investigate the optimal geometrical and electronic configuration, the number of energy bins required for task specific imaging, the trade-off between spatial and energy resolution and high count rate and the imaging and dosimetric performance of a CZT-based multispectral medical x-ray imaging system. Further, we have been exploring the feasibility of using CZT detectors for Cone Beam CT, assessing the detector performance for spectral AuNPs imaging within a pre-clinical small-animal radiation research platform and demonstrating the potential of Cone Beam Spectral Tomography as an advanced imaging technique for diagnostic and radiotherapy imaging. Results from both theoretical and experimental studies will be presented and the feasibility and the challenges of implementing CZT pixelated detectors and advanced photon-counting readout electronics for multi-spectral photon-counting medical x-ray imaging will be extensively discussed and their performance and potential to be implemented for AuNP cone-beam spectral tomography imaging will be assessed.

M5BP-175, Imitation Methods for Study of Micro-Calcification with Phase-Contrast X-Ray Mammography

X. Zhu1,2, L. Zhang1,2, S. Wang3,4, X. Jiang1,2, X. Li1,2, W. Peng3,4, Z. Chen1,2

1Department of Engineering Physics, Tsinghua University, Beijing, China
2Key Laboratory of Particle and Radiation Imaging, Ministry of Education, Beijing, China
3Department of Diagnostic Radiology, Cancer Center Fudan University, Shanghai, China
4Department of Oncology, Shanghai Medical College, Fudan University, Shanghai, China

X-ray phase contrast imaging with grating interferometry, as a potential nondestructive imaging tool in biology and medical tests, works well with conventional X-ray tubes, and provides three kinds of information with higher contrast. X-ray phase-contrast mammography was already proved to have potential in identification and classification of microcalcifications, which are indicative in the diagnosis of early breast cancer. However, ideal breast tissue sample with microcalcifications for scientific study are limited, since the identification and confirmation of microcalcifications are stiil requiring and costly. In this paper, we propose a very simple imitation method for some potential study of microcalcifications. Granule and powder of similar composition with maicrocalcifications and substances of similar structure with human breast tissues are used to imitate real breast tissue sample. Experimental results verify the similarity with real samples and the feasibility of the method.

M5BP-179, Optimization Techniques of Radiation Dose for Dedicated Breast CT

Z. Wang1,2, C. Wei1,2, Y. Wang1,2, X. Zhang1,2, M. Li1,2, L. Wei1,2

1Division of Nuclear Technology and Applications, Institute of High Energy Physics, Beijing, China
2Beijing Engineering Research Center of Radiographic Techniques and Equipment, Beijing, China

Reducing radiation dose level, suppressing image artifacts and improving image quality are issues under intensive research in the dedicated breast CT (DBCT) development. In this study, a breast CT imaging platform and a Monte-Carlo simulation platform are built in the first place. Then the effects of tube voltage, mAs, filter and projection angle on the radiation dose level, artifacts and image quality are studied by using parameters such as radiation dose, signal-to-noise ratio and quality factor. Experiment results indicate that the quality factor is optimized when the tube voltage is chosen as 50kVp. Moreover, by reducing mAs, the DBCT can still obtain a good image quality in low radiation scan. Bowtie filters with appropriate shape can lead to a better dose efficiency than normal flat filters. Besides, short scan mode can reduce the radiation dose level by 40% with similar image quality compared with conventional scan mode. In conclusion, choosing proper tube voltage, using low radiation scan mode, X-ray uneven modulation, finite angle projection are effective in reducing radiation dose and improving image quality.

M5BP-183, Evaluation of the Random Fraction Contribution to Error in Quantitative Y-90 PET

J. Strydhorst1, T. Carlier2,3, A. Dieudonné4,5, I. Buvat1

1IMIV, U1023 Inserm/CEA/Paris Sud University and ERL 9218 CNRS, CEA/SHFJ, Orsay, France
2Nuclear Medicine Department, University Hospital of Nantes, Nantes, France
3CRCNA, Inserm U892, Nantes, France
4Department of Nuclear Medicine, Beaujon Hospital, Assistance Publique-Hôpitaux de Paris (APHP), Clichy, France
5Inserm U1149, Clichy, France

Objective: Y-90 PET can be used to assess the dose distribution associated with radioembolization using microspheres in liver cancer treatment. Yet, Y-90 PET produces images with poorer contrast recovery than F-18 PET. Our aim is to evaluate the factors limiting the quantitative accuracy of Y-90 PET. In this work, we focus on the contribution of the relatively high random fraction (RF) when imaging Y-90 with an LSO-crystal scanner. Method: List-mode projection data was acquired for a NEMA94 PET phantom with one insert empty, one filled with water, the third with a solution of Y-90 or F-18 and no background activity (water). Data sets with 80k – 3M net trues were extracted for each isotope. A high-RF F-18 realization was created by adding events to the list mode data to simulate a RF of 0.7, comparable to that observed for the Y-90 data. Images were reconstructed using the clinical reconstruction software, with and without time-of-flight. Results: For the unmodified F-18 images, the total activity in the background was about 3.4% and 2.7% of the activity in the insert for non-TOF and TOF respectively. Artificially increasing the RF of the F-18 projection data increased the amount of activity measured in the cold region of the phantom to 5.3% (non-TOF) and 3.4% (TOF) images, while for Y-90 PET, the activity in the cold region was 9.1% (non-TOF) and 6.8% (TOF) of the activity in the insert. Without TOF, for both the high-RF F-18 and the Y-90 images, the activity in the cold region decreased as the number of net trues increased; this was not observed for the unmodified F-18 data, nor in the TOF reconstructed images. Conclusions: The high RF of Y-90 PET data acquired with an LSO scanner may account for about a third of the increase in activity seen in the cold background region relative to F-18 PET. The contribution of other factors, such as correlated events from the LSO or the bremsstrahlung background still remain to be evaluated.

M5BP-187, The Effect of Anaesthesia on 18F-FDG Uptake in the Rat Brain: a Fully Conscious Dynamic Study Using Motion Correction

M. G. Bickell1, B. de Laat1, R. Fulton2, G. Bormans3, J. Nuyts1

1Department of Nuclear Medicine, KU Leuven, Leuven, Belgium
2Brain & Mind Research Institute and the Faculty of Health Sciences, University of Sydney, Sydney, Australia
3Department of Radiopharmacy, KU Leuven, Leuven, Belgium

Anaesthesia is used in preclinical studies to ensure that no motion of the animal occurs during the scan. The effect of the anaesthesia on the tracer or drug under study is not always known, and this may confound the results of the study and how it is translated to the clinic, where anaesthesia is avoided. Using developed motion correction techniques, we present results from a study of the effect of isoflurane on FDG uptake in the rat brain using fully awake tube-bound rats. The study involved dynamic scans of both awake and asleep rats. Thus far 3 rats have been studied and a consistent decrease in the uptake of FDG due to the anaesthesia was observed.

M5BP-191, Modeling and Analysis of a Physical Tumor Model Including the Effects of Necrotic Core

J. Dey1, S. W. Walker2, J. M. Mathis1, D. Shumilov1, K. M. Kirby3, Y. Luo3

1Dept of Physics and Astronomy, Louisiana State University, Baton Rouge, LA, US
2Dept of Mathematics, Louisiana State University, Baton Rouge, LA, US
3Dept of Comparative Biomedical Sciences, Louisiana State University, Baton Rouge, LA, US

A new comprehensive diffusion-reaction equation for tumor density is presented with the necrotic conditions accounted for. The forward model has been implemented in 3D using FELICITY FEM code and shown on a mouse SPECT/CT dataset. Future work involves inversion of model across time-points to extract parameters from our model to be assessed for potential cancer quantification and management.

M5BP-195, A Simplified Monte Carlo Based Approach for Physical Effects Correction in SPECT

B. Auer, C. Rey, J.-M. Gallone, V. Bekaert, D. Brasse, Z. El Bitar

IPHC, CNRS UMR7178, Universite de Strasbourg, 23 rue du Loess, 67037 Strasbourg, France

In Single Photon Emission Computed Tomography (SPECT), attenuation and scatter introduce important artefacts in the reconstructed images biasing the diagnosis and the follow-up of the imaged patient. Furthermore, by using Monte Carlo Simulation (MCS), physical effects undergone by photons during the SPECT exam can be precisely modeled and accounted for during iterative image reconstruction, which improves the image quality. However, MCS are large time consuming and therefore inappropriate for the rate of daily exams performed in both clinical and preclinical routine. We took advantage that patients are composed of identical biological tissues and that photon propagation in an element volume of a given tissue is similar and reproducible from one patient to another. In this work we propose to accelerate the modeling of the physical effects occurring in the patient making it adequate for daily exam, by using the approach of scatter precalculated database. The efficient patient-dependent attenuation and scatter correction for routine purpose has to be very adequate for implementation on GPU architectures on a state-of-art single-processor workstation. Preliminary results presented in this study are promising in terms of accuracy with full MCS and computation time (speed up factor about 700 was achieved).

M5BP-199, Prompt Gamma Correction for Ga-68 PSMA PET Studies

I. Hong1, H. Rothfuss1, S. Fürst1,2, C. Michel1, S. Nekolla2, B. Bendriem1, M. Casey1

1Siemens Healthcare, Knoxville, TN, USA
2Technische Universität München, München, Germany

Ga-68 Prostate Specific Membrane Antigen also known as PSMA is currently used in prostate cancer PET imaging. The resulting images show high uptakes in kidney and bladder which could produce a photopenic artifact (halo) and potentially mask tumor lesions or bone metastasis at the level of kidney or bladder. The measured contrasts between these organs and background could be as high as 200:1 and 50:1 for kidney and bladder respectively. The correct quantification in these areas requires precise scatter correction which needs to account for the effect of prompt gamma. Ga-68 has a prompt gamma at 1077 keV with a branching ratio of 3.2%. An unscattered prompt gamma ray of 1077 keV in the object has a small probability to be detected. An object scattered prompt gamma has a higher detection probability. When the contrast is low, more accurate quantification can be achieved. On the contrary, when the contrast is very high, halo artifact can be observed around high uptake organs. The purpose of this work is to evaluate the effect of Ga-68 prompt gamma in clinical PSMA studies. The halo artifact around kidney and bladder is strongly reduced by applying a Prompt Gamma Correction. Selected studies were performed on a Siemens mCT and acquired by Technische Universität München, Germany.

M5BP-203, An Investigation of Prompt Gamma Correction on I-124 PET Study

I. Hong, H. Rothfuss, C. Michel, M. Casey

Siemens Healthcare, Knoxville, TN, USA

Prompt emission of gamma radiation degrades the quantitative accuracy and image quality on I-124 PET studies. This prompt gamma generates a relatively flat background compared to object scatter and makes tail fitting of single scatter simulation difficult. Single scatter simulation with tail fitting will be overestimated if this background due to prompt gamma is not accounted for. Several solutions have been investigated to estimate this prompt gamma background such as Monte-Carlo simulation. However, Monte-Carlo simulations take too long to be used in clinical practice. Another approach assumes this background is relatively flat, and uses total measured singles data to approximate it. This assumption was successfully implemented for prompt gamma ray in the cases of Rb-82 and Ga-68 because the branching ratios are small compared to the amplitude of single scatter for tail fitting procedure. However, the larger branching ratio of I-124 generates a non-uniform background which produces error in the single scatter scaling with scatter tail fitting. In this paper, we propose a novel method to generate the non-uniform prompt gamma background and compare it to Monte-Carlo simulations and experimental measurement on I-124 phantom.

M5BP-207, Model Asymmetrical Detector Response Function with a Skew Normal Distribution Function in PET

X. Jin, J. Miao, S. G. Ross, C. W. Stearns

GE Healthcare, Waukesha, WI, U.S.A.

In PET image reconstruction, a point-spread-function (PSF) in the form of normal distribution is commonly used to model the detector response function. The PSF becomes asymmetrical off the center of the scanner's field-of-view. This effect has been modeled with two half normal distribution functions that have different standard deviations on the left and right side. However, this method is subject to unequal noise in the estimated parameters between the two half normal distribution, due to the difference in the number of data points for asymmetrical detector response function. In this work, we present a skew normal distribution model that includes a standard deviation and a skewness parameter to model the asymmetrical detector response function using all the data points from each data acquisition. The skew normal distribution model can noticeably improve the goodness of fit to the raw data over the two-half normal distribution model by an average of 42% in sum of squared difference, thus giving more reliable estimation of the PSF.

M5BP-211, Energy Dependent Normalization Method in Positron Emission Tomography (PET)

M. Aykac, V. Y. Panin, H. E. Rothfuss

Siemens Molecular Imaging, Knoxville, TN, USA

Prior to the reconstruction process, scatter correction and normalization needs to be applied to correct raw PET data. Normalization corrects for all the variations between the detector responses. The components of the normalization describe the calibration factors to be applied to each LOR in PET emission data. One set of normalization components was typically estimated for the clinical energy window between 435keV and 650keV. However, some of the components might change depending on the energy of the detected photons even within the clinical imaging window. In this work, we investigate whether there is need for different sets of norm components based on the energy in our clinical energy window. In order to generalize this investigation, we also study how the norm components change for photons with lower energies.

M5BP-215, Fast Atlas-Based MRI-Guided PET Attenuation Map Generation in Whole-Body PET/MR Imaging

H. Arabi, H. Zaidi

Division of Nuclear Medicine and Molecular Imaging, Geneva University Hospital, Geneva, Switzerland

We propose a one registration multi atlas (ORMA) pseudo-CT generation algorithm for attenuation correction in whole-body PET/MRI based on an optimized atlas-guided bone segmentation procedure. The proposed approach requires only one online registration between the target and reference images regardless of the number of atlas images N, while for the remaining subjects belonging to the atlas dataset, the pre-computed transformation matrices obtained from registration to the reference image is used to align them to the target image. The performance characteristics of the proposed method were evaluated and compared to conventional atlas-based attenuation map generation consisting in direct registration of the entire atlas images to the target. Four different PET attenuation maps were produced using direct registration and ORMA, voxelwise weighted (VWW-Direct and VWW-ORMA) and arithmetic average (AA-Direct and AA-ORMA) atlas fusion strategies. The comparison of validation measures characterizing the accuracy of extracted whole-body bone demonstrated the superiority of VWW-Direct atlas fusion technique resulting in a Dice similarity measure of 0.82±0.04 compared to 0.60±0.02 for AA-Direct. Conversely, the ORMA approach yielded a Dice similarity measure of 0.76±0.05 for VWW-ORMA and 0.55±0.03 for AA-ORMA. Quantitative analysis of PET data revealed good correlation (y=1.01x+0.1, R2= 0.99) between PET images corrected for attenuation using the proposed pseudo-CT and the corresponding reference CT images. The proposed method generates decent attenuation maps and enables to reduce the processing time for atlas-based pseudo-CT generation reduced by a factor N.

M5BP-219, 3-D Model-Based Correction of Artifacts Generated by Localized High-Intensities in a PET Scanner

J. Huang, T. Mou, M. Muzi, F. O'Sullivan

School of Mathematical Sciences, University College Cork, Cork, Ireland

The presence of localized high-intensity regions such as the bladder or injection site tubing can introduce Gibbs-like spillover artefacts that distort quantitative accuracy well away from the high-intensity regions in the field of view. This is a familiar problem with the filtered back projection reconstructions used in emission and transmission tomography. A range of techniques have been proposed in the literature to address this. Some involve numerically complex regularization procedures. In recent work, we presented a simplified correction approach based on modelling the 2-D target source distribution as a sum of a high-intensity localized object and a more regular smoothly varying distribution. This method has given promising results in simulations and with real data. The present work refines the efficiency of that algorithm and implements it in a fully 3-D form. In the approach a 3-D mask region (produced by thresholding) is used to guide the parametric approximation of the high-intensity component. By modelling the approximate 3-D point-spread of the scanner, the methodology can be applied to reconstructed images directly, avoiding the heavy computational cost of iterative refinement involved with methods that require projection into the raw detector domain. Our work here specifically reports on a series of dynamic PET imaging studies of acute myeloid leukemia patients using [F-18] fluorothymidine PET. The proposed technique is found to correctly remove local high intensity artifacts associated with the bladder and the injection site tubing. The tissue structures and bone marrow uptake are more easily seen in the corrected PET images. The refined technique is promising and has potential for addressing artifacts associated with a single readily identifiable localized high-intensity object in the field of view.

M5BP-223, New Texture Features for Improved Differentiation of Hyperplastic Polyps from Adenomas via Computed Tomography Colonoscopy

Y. Hu1,2, H. Han1, P. J. Pickhardt3, Z. Liang1

1Radiology, Stony Brook University, Stony Brook, United States
2Applied Mathematics and Statistics, Stony Brook University, Stony Brook, United States
3Radiology, University of Wisconsin School of Medicine and Public Health, Madison, United States

Feature classification plays an important role in computer-aided diagnosis (CADx) of suspicious lesions. While many texture features have been extracted and applied for various clinical purposes, Haralick’s feature extraction method is of great interest, because it gives a series of texture measures on the image intensity correlations among the image pixels across an image slice. Based on the Haralick’s method, we proposed a new set of features for CADx of colonic polyps or differentiation of hyperplastic polyps from adenomas. We evaluated this new feature set by means of random forest (RF) classifiers on a database of 153 polyps, including 116 adenomas and 37 hyperplastic polyps. The classification results were documented quantitatively by the Receiver Operating Characteristics (ROC) analysis and the merit of area under the ROC curve (AUC), which are well-established evaluation criteria to various classifiers. Experimental results demonstrated that the new feature set significantly improved the CADx performance for colonic polyps.

M5BP-227, Maximum a Posteriori Reconstruction of Small-Animal Dynamic Brain PET Images on a Siemens Inveon PET/CT

J. Bini1, S. J. Finnema1, C. R. Lattin1, G. W. Cline2, R. E. Carson1

1PET Center, Department of Diagnostic Radiology, Yale University School of Medicine, New Haven, CT, USA
2Department of Internal Medicine, Yale University School of Medicine, New Haven, CT, USA

Iterative reconstruction algorithms with point-spread function modeling implemented in a maximum a posteriori (MAP) algorithm can potentially provide improved image resolution and tracer quantification in small-animal dynamic brain positron emission tomography (PET). Improving resolution for preclinical imaging of small animals may increase accuracy of time activity curves (TACs) and the resulting kinetic model parameters. We compared 2D filtered backprojection (2D-FBP), 2D ordered subsets expectation maximization (2D-OSEM) and three MAP reconstructions from the Inveon PET/CT system for a variety of small-animal PET brain studies. All acquisitions were acquired in listmode and included: a male low-density lipoprotein receptor knockout mouse injected with 18F-fluorodeoxyglucose, a Sprague-Dawley rat injected with 11C-UCB-J (SV2A ligand which binds selectively in synapses in gray matter) and a house sparrow injected with 11C-raclopride. Each acquisition was reconstructed with 2D-FBP (Ramp filter, cutoff frequency 0.5), 2D-OSEM (4 iterations; 16 subsets), and 3D-OSEM-MAP with 2 OSEM iterations, 16 subsets and 18 MAP iterations with three different ß values (0.5678, 0.0043 and 0.0023). We saw improved quantification, as represented by improved contrast, using 3D-OSEM-MAP reconstruction using ß parameters of 0.0043 and 0.0023 and the potential to improve TACs and kinetic modeling parameters in small-animal dynamic PET imaging.

M5BP-231, An Improved Statistical Approach to the Estimation of Spatial Bias and Variability in Reconstructed PET Data

T. Mou, J. Huang, Y. Zhang, P. Kinahan, F. O'Sullivan

School of Mathematical Sciences,, University College Cork, Cork, Ireland

The spatial bias and covariance characteristics of reconstructed PET data play an important role in characterizing the quantitative accuracy of image data produced by operational PET scanners. In practice physical phantom measurements are used as a basis for such assessments. Our previous analysis of uniform cyclindrical phantom data from 55 sites using the American College of Radiology Imaging Network (ACRIN) protocols for dynamic PET-FDG brain imaging acquisition, showed that both spatial bias and covariance in operational scanners can be empirically decomposed as a product of a temporal, radial and axial patterns in both bias and variance. On the basis of this structure, we develop a penalized maximum likelihood approach for efficient estimation of the bias and variance patterns. The approach is iterative and can be viewed as a variation on the ?-method of function approximation, introduced by Brieman in the Statistical literature in the 80s. The proposed approach is evaluated using a simulation model. The method is also demonstrated on PET data collected on an operationally used clinical PET scanner using the ACRIN PET-FDG dynamic brain imaging protocol. Appropriate residual diagnostics are used to confirm the empirical accuracy of the recovered patterns.

M5BP-235, Imaging Performance Measurements for a 1mm3 Resolution Clinical PET System

D. F. C. Hsu1, D. L. Freese1, D. Innes2, P. D. Reynolds1, A. Vandenbroucke2, C. S. Levin1,2,3,4

1Electrical Engineering, Stanford University, Stanford, CA, USA
2Radiology, Stanford University, Stanford, CA, USA
3Physics, Stanford University, Stanford, CA, USA
4Bioengineering, Stanford University, Stanford, CA, USA

We are constructing a dual-panel PET system dedicated to loco-regional imaging (e.g. breast, head/neck, etc). The edge-on layout of the crystal arrays provides precise 3-D positioning, even when there are multiple interactions, while the dense packing of the 0.9x0.9x0.9 mm3 crystal elements provides high sensitivity and spatial resolution. A 16-layer prototype of our design has been built, with a crystal volume of 160mm by 64mm (area) by 22mm (depth). The National Electrical Manufacturers Association (NEMA) has created standards that several groups have modified to characterize dual-panel PET imaging systems. In this work, we present results from scans of the image quality (IQ) phantom of the NEMA NU-4 standard. We characterize the uniformity, contrast recovery, and spillover ratios of our system, by scanning the IQ phantom with 100µCi for 20 minutes at three different locations. Uniformity was measured with a standard deviation of 44%. Contrast recovery coefficients, when referenced to the maximum pixel value of the 5mm rod, are 0.24, 0.30, 0.75, and 0.77, with standard deviations of 131%, 87%, 43%, and 49%, for the 1mm, 2mm, 3mm, and 4mm rods respectively. The spillover ratios, when referenced to the hot background, are 0.32 and 0.38, with standard deviations of 45% and 50%, for the cold chambers filled with air and water respectively.

M5BP-239, Investigating the Impact of Advanced Instrumentation to Reduce Dose in Pediatric-PET Applications

J. P. Schmall, S. Surti, M. E. Daube-Witherspoon, J. S. Karp

Department of Radiology, University of Pennsylvania, Philadelphia, United States

The clinical utility of PET has been shown for many different types of disease. As the number of PET imaging procedures increases, clinicians remain concerned over the associated radiation dose, and the biological effects are still somewhat a matter of contention. Dose reduction techniques for PET are therefore very useful in broadening the clinical applications of PET, and they will take a heightened significance in pediatric imaging. The diagnosis and treatment of many different pediatric diseases could potentially be improved by the use of PET, but little progress has been made due to concerns over radiation exposure. The Children’s Hospital of Philadelphia has an ongoing clinical trail to assess the impact of 18F-Fluoro-DOPA PET/CT in the treatment of Hyperinsulinemic Hypoglycemia. Our goal is to investigate the degree to which advanced instrumentation will allow for dose reduction, while maintaining acceptable image quality. We use a model emulating pediatric hyperinsulinism and evaluate two different concepts for increasing scanner sensitivity: extending the axial field-of-view (FOV) or improving the system timing resolution. Initial simulations quantify the sensitivity gain associated with having a longer axial FOV. We also investigate the influence of patient attenuation by simulating phantoms of varying size. The system timing resolution was varied from 2ns to 100ps for a scanner having an axial FOV of 18cm and 72cm. Image quality was assessed using a phantom mimicking pediatric hyperinsulinism and count statistics based on an injected activity ranging from 1-0.125mCi. These simulations aim to quantify the TOF benefit in small objects, and compare this to the sensitivity gain obtained by having a longer axial FOV. This study will be used to develop guided instrumentation objectives, which can then be applied to the design of a dedicated pediatric-PET scanner or improve the performance of future scanner designs capable of both adult and pediatric imaging.

M5BP-243, Development of a Defect Model for Renal Pediatric SPECT Imaging Research

Y. Li1, S. O'Reilly2, D. Plyku1, X. Cao3, F. Fahey3, W. E. Bolch2, S. T. Treves3, G. Sgouros1, E. C. Frey1

1Department of Radiology, Johns Hopkins University, Baltimore, MD, USA
2Department of Biomedical Engineering, University of Florida, Gainesville, FL, USA
3Boston Children’s Hospital, Harvard University, Boston, MA, USA

Children generally are at greater risk from radiation exposure due to their greater sensitivity to radiation and the longer time frame after exposure during which effects can manifest themselves. Thus, reducing radiation exposure is a major concern in pediatric nuclear medicine imaging. Lower administered activities (AA) can reduce patient radiation exposure, but also result in reduced diagnostic accuracy. Typically, the administered activity for diagnostic nuclear medicine imaging of children is based on patient weight, but our previous studies have shown that weight alone may not be sufficient to accurately determine an optimal patient-specific AA. The ultimate goal of this work is to investigate the tradeoff between image quality and radiation risk as a function of AA as a function of patient age, gender and body habitus that can be used to guide dosing of dimercaptosuccinic acid (DMSA) SPECT, a common pediatric nuclear medicine procedure for assessing renal function. In this work we created a model of kidney function defects and used it to generate defects for a population of anthropomorphic reference phantoms with realistic age-dependent variations in anatomy and pharmacokinetic model based organ uptakes. The population includes 90 phantoms modeling both genders, variations in height and weight, and ages from newborn to 15 years. The defects generated with the were projected using an analytic projection code that models physical image degrading effects and were scaled with variations in organ uptake based on a new kinetic model fit to published data and parameters extracted from pediatric SPECT images. Combined with a previously-generated database of projections, the defect database will be used in image quality studies to determine the relationship between image quality, administered activity, and patient height and weight. In conjunction with already-calculated risk estimates, these will provide data to inform guidelines for pediatric dosing of DMSA.

M5BP-251, Comparison of TV Norm Minimization and MLEM for Reduction of Metal Artifacts in Transmission Tomography

H. Guzman, B. Smith

Electrical Engineering, University of Texas San Antonio, San Antonio Texas, USA

Streaking artifacts in Computed Tomography caused by metal objects embedded in the human body can be mitigated by a variety of algorithms. One such method is the total-variation (TV) norm minimization. Another method that could be used is the Maximum Likelihood Expectation Maximization (MLEM). The main objective of this paper is to compare the performance of the MLEM and TV minimization in mitigating these artifacts. To provide a benchmark, these algorithms will be compared with two conventional algorithms; namely ART and ART with a positivity constraint. Mathematical phantoms with simulated region of “metal” pixels were used to produce simulated data. It was found that TV and the MLEM algorithms produced substantially more accurate results than ART and ART with a positivity constraint. The TV algorithm produced more accurate results than the MLEM algorithm. Although the results presented are both the TV-MGD and the MLEM algorithms are promising, further effort is needed to establish their usefulness with real data.

M5BP-255, Scanner Dependent Noise Properties of the Q.Clear PET Image Reconstruction Tool

J. Lantos, A. Iagaru, C. Levin

Radiology, Stanford University, Stanford, CA, USA

In this paper we compare the noise properties of the newly introduced Q.Clear image reconstruction algorithm and conventional OSEM using the ACR phantom measured on the GE Discovery 600 and 690 PET/CT scanners with various count statistics.
In the D600 measurement the SNR decreases from 31.5±1.3 and 26.9±1.6 to 5.3±0.4 and 5.5±0.3 for the Q.Clear (with regularization strength parameter beta=350) and the OSEM images, respectively as the statistics decreases from 100% to 5%. The average mean hot concentration recovery values of the three biggest rods over all time slices are 0.57±0.01, 0.66±0.04 and 0.8±0.02 for Q.Clear (with beta=350) and 0.63±0.03, 0.71±0.05 and 0.83±0.03 for OSEM, respectively. The SUV max related maximum hot activity concentration recovery calculated from the maximum concentration averaged for these 3 rods increased from 0.78 and 0.88 to 1.1 and 1.5, respectively for the two algorithms. In the D690 measurement the SNR decreases from 26.6±1.4 and 31.0±1.3 to 3.8±0.4 and 6.3±0.5 for Q.Clear and the OSEM, respectively. The mean hot concentration recovery values are 0.66±0.04, 0.8±0.02 and 0.92±0.01 for Q.Clear and 0.65±0.04, 0.78±0.02 and 0.9±0.02 for OSEM, respectively. The maximum hot activity concentration recovery increased from 0.98 and 0.93 to 2.3 and 1.4, respectively for the two algorithms.
In summary for the two scanners the SNR of Q.Clear matches the SNR of OSEM with different regularization strength applied. Although the mean concentration recovery values were not significantly different in either measurement (with beta=350) the maximum recovery was affected by noise and was closer to one with Q.Clear in the case of D600 and with OSEM for the D690. Q.Clear resulted in significantly better cold contrast in bone for both scanners (0.053±0.004 vs. 0.106±0.01 and 0.0±0 vs. 0.144±0.004) and in air for the D690 (0.305±0.004 vs. 0.380±0.005).

M5BP-259, Impact of Time-of-Flight Image Reconstruction in PET Parametric Imaging

F. A. Kotasidis1,2, A. Mehranian1, H. Zaidi1,3,4

1Division of Nuclear Medicine & Molecular Imaging, Geneva University Hospital, Geneva, Switzerland
2Wolfson Molecular Imaging Centre, MAHSC, University of Manchester, Manchester, United Kingdom
3The Geneva Neuroscience Centre, Geneva University, Geneva, Switzerland
4Department of Nuclear Medicine and Molecular Imaging, University of Groningen, University Medical Centre Groningen, Groningen, The Netherlands

Kinetic parameter estimation in dynamic PET imaging requires reconstruction of multiple time frames. Due to the limited counting statistics and reduced signal-to-noise ratio (SNR) in each frame, parametric maps suffer from reduced accuracy and precision. Therefore image reconstruction strategies improving upon the SNR are particularly important in the context of parametric imaging. Time-of-flight image reconstruction has been shown to improve the SNR and increase the effective sensitivity. However, so far the benefit of TOF has only been demonstrated in static imaging applications with potentially substantial benefits when used in dynamic pharmacokinetic imaging applications. Using traditional dynamic 3D as well as direct 4D image reconstruction algorithms, we evaluate the benefit of TOF on kinetic parameter estimation using various TOF resolutions, kinetic models and count levels. Initial data suggest that both bias and variance in the kinetic parameters are reduced with improvements depending on the kinetic model and becoming more significant at increased TOF resolutions. Incorporating TOF within direct 4D image reconstruction and combining the corresponding SNR gains results in substantial improvements in parametric maps compared to traditional post reconstruction kinetic analysis.

M5BP-261, The ML-EM Algorithm Is Not Optimal for Poisson Noise

G. L. Zeng1,2

1Engineering, Weber State University, Ogden, UT, USA
2Radiology, University of Utah, Salt Lake City, UT, USA

The ML-EM (maximum likelihood expectation maximization) algorithm is the most popular image reconstruction method when the measurement noise is Poisson distributed. This short paper considers the problem that for a given noisy projection data set, whether the ML-EM algorithm is able to provide an approximate solution that is close to the true solution. It is well-known that the ML-EM algorithm at early iterations converges towards the true solution and then in later iterations diverges away from the true solution. Therefore a potential good approximate solution can only be obtained by early termination. This short paper argues that the ML-EM algorithm is not optimal in providing such an approximate solution. In order to show that the ML-EM algorithm is not optimal, it is only necessary to provide a different algorithm that has better performance. An alternative algorithm is suggested in this paper and this alternative algorithm is able to outperform the ML-EM algorithm.

M5BP-263, The Impact of Contamination with Long-Lived Radionuclides on PET Kinetics Modeling in Multi-Tracer Studies

L. Jødal1,2,3, S. B. Hansen2, S. B. Jensen3,4

1Dept. of Veterinary Disease Biology, University of Copenhagen, Copenhagen, Denmark
2Dept. of Nuclear Medicine & PET Center, Aarhus University Hospital, Aarhus, Denmark
3Dept. of Nuclear Medicine, Aalborg University Hospital, Aalborg, Denmark
4Dept. of Chemistry & Biochemistry, Aalborg University, Aalborg, Denmark

Background: Kinetic modeling of sequential dynamic PET scans in the same subject enhances comparability but introduces the problem of contamination from remains of the first tracer in blood samples, thereby adding a background signal to the input function. This is a problem especially when the first tracer has to be chosen with much longer half-life than the second tracer. Aim: We present a correction based on late re-counting of the blood samples and investigate the effect on kinetic modeling in a two-tissue compartment model with irreversible uptake (3k model). Methods: Our study used dynamic PET scans and blood samples from the recent study by Nielsen et al (2015, www.ajnmmi.us/files/ajnmmi0002237.pdf), involving both 111In and PET tracers. The radionuclide 111In emits two photons in rapid cascade with a combined energy that may be detected within the PET energy window during gamma counting. Samples were re-counted 2 days (~48 hours) later when only 111In (T½ = 2.8 days) was left. We used these data for correction of the original (early) count rates. We performed kinetic modeling using both non-corrected and corrected input functions. Results: The K1 parameter was almost unaffected, in most cases k2 was only slightly affected, while k3 could be underestimated by many percent if non-corrected plasma data were used. Naturally, the correction had the largest impact on the part of the input function with the weakest foreground signal. Conclusions: The used of sequential dynamic PET scans in the same research animal or subject easies comparison, but care must be taken when a background signal from a previous tracer remains, especially in determining a correct input function for kinetic modeling. The more a model parameter depends on the late input function, the more it will be susceptible to background signals. The presented method is general for the situation where the first applied radiotracer has a far longer physical half-life than the following tracer(s).

M5BP-267, Multi-Bed Tracer Kinetic Imaging of Micro-Parameters from Dynamic Time-of-Flight PET Data

F. A. Kotasidis1,2, N. A. Karakatsanis1, A. Mehranian1, H. Zaidi1,3,4

1Division of Nuclear Medicine & Molecular Imaging, Geneva University Hospital, Geneva, Switzerland
2Wolfson Molecular Imaging Centre, MAHSC, University of Manchester, Manchester, United Kingdom
3The Geneva Neuroscience Centre, Geneva University, Geneva, Switzerland
4Department of Nuclear Medicine and Molecular Imaging, University of Groningen, University Medical Centre Groningen, Groningen, The Netherlands

Compartmental modelling in dynamic imaging requires the full time course of the activity distribution to be sampled. Due to the need for increased early temporal sampling, leading to severely low signal to noise ratio (SNR) in those early frames, such protocols are restricted to a single bed. Time-of-flight (TOF) imaging has been shown to improve the SNR in static imaging applications. When used in dynamic imaging, the effective NEC gain could be used to generate TOF equivalent dynamic data with reduced frame duration and allow interleaving between adjacent bed positions to increase the axial extent from single bed to multi-bed acquisitions. Coupling this with the recent introduction of continuous bed motion acquisition modes, could make such a dynamic protocol feasible. Using simulated data based on the Siemens mCT TOF PET, we propose an axially extended dynamic imaging protocol and demonstrate the feasibility of generating TOF equivalent dynamic data as well as extending kinetic parameter estimation using compartmental modelling from single- to multi-bed dynamic imaging.

M5BP-271, An automated clustering algorithm for reference region extraction of brain 11C-PK11195 studies

L. Presotto1,2, L. Iaccarino1,2, V. Bettinardi3, L. Gianolli3, D. Perani1,2,3

1San Raffaele Vita-Salute University, Milan, Italy
22) In vivo human molecular and structural neuroimaging Unit, Division of Neuroscience, San Raffaele Scientific Institute, IRCCS, Milan, Italy
3Nuclear Medicine Department, San Raffaele Scientific institute, IRCCS, Milan, Italy

Background: A convenient way to analyze quantitatively 11C-PK11195 cerebral PET scans is the simplified reference tissue model (SRTM), which fits Time Activity Curves (TACs) to those of a reference non-pathological region. In this work, we present a fully automatic method which makes use of the expected tracer concentration in 4 predefined tissue classes (gray matter, white matter, blood, high specific binding) to automatically produce the TAC of the reference region. Methods: The proposed algorithm starts by removing irrelevant structures (skin and skull) based on activity and shape information, then it classifies the remaining voxels in the 4 tissue classes based on the L2 distance from pre-defined seeds. Within those classified as “gray matter”, those having the greatest L2 distance from the high specific binding, are selected to build the reference TAC. Seeds were built from scans of 9 healthy controls and from 2 patients with Alzheimer dementia, known to have activated microglia. Binding potential values (BP) were measured in two structures for 9 healthy controls using the SRTM with the output of clustering algorithm or a whole cerebellum reference as input and then compared. Results: The output maps of clusters did not show evident artifacts on visual inspection. The (BP) obtained by the proposed algorithm was close to the one obtained with the cerebellum reference in all cases. Bland-Altman analysis found a mean difference of -0.086 and 95% limits of agreement of -0.17, -0.05. Conclusions: The proposed clustering algorithm is a fully automated one which showed a reliable performance in the healthy controls and was comparable to the reference method.

M5BP-279, Anatomy-Assisted Direct Parametric PET Imaging for Myocardial Blood Flow Abnormality Detection

W. Deng, X. Wang, B. Yang, J. Tang

Oakland University, Rochester, Michigan, United States

Dynamic myocardial perfusion (MP) PET imaging provides quantitative measurement for myocardial blood flow (MBF). The purpose of this study is to incorporate anatomical information in the 4D direct parametric image reconstruction and to evaluate the performance in detecting regional MBF abnormality. To improve the noisy MBF estimation caused by individually reconstructing short dynamic frames, we developed a 4D direct parametric image reconstruction algorithm. The one-tissue compartmental model was formulated in the maximum likelihood (ML) problem to relate the dynamic projection datasets directly to the kinetic parameters. To further improve the 4D direct parametric imaging, we developed a maximum a posterior (MAP) algorithm that incorporates the information-theoretic similarity between the anatomic and parametric images in the reconstruction. The preconditioned steepest ascent algorithm was used to solve the ML and MAP estimation problems. Using the XCAT phantom and patient-based organ time activity curves, we simulated two sets of dynamic MP Rb-82 PET data, one carrying normal MBF and the other with reduced MBF on a region of interest (ROI), each with 20 noise realizations. Corresponding MRI images were simulated with the 3D T1-weighted sequence as specified in a clinical PET/MRI protocol. The reconstructed parametric images from the ML direct and the MAP direct algorithms were compared using the tradeoff between normalized standard deviation versus normalized mean squared error and the signal to noise ratio (SNR), which reflects the separability between the normal and abnormal K1 parameters. On the left ventricle and the ROI, the MAP algorithm results in improved noise versus bias tradeoff compared to the ML algorithm. The SNR on the ROI was 1.88 from the MAP algorithm and was 1.38 from the ML algorithm. We conclude that incorporating anatomical information in the 4D direct parametric image reconstruction improves the detection of regional abnormal MBF.

M5BP-283, Penalized Direct Estimation of Parametric Images in PET

K. Kim, G. E. Fakhri, Q. Li

Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, United States

Direct parametric imaging of positron emission tomography (PET) can compute the voxel-wise kinetic parameters directly from raw data, and thus is able to achieve more accurate physiological information of tracer uptake. Recently, Wang and Qi proposed a direct reconstruction method using a quadratic surrogate function with optimization transfer (OT) method to guarantee the convergence; it contains two steps: updating dynamic images and calculating pixel-wise nonlinear least square (NLS) parameter fitting. Although the cost function converges in this method, the pixel-by-pixel NLS can be noisy due to the high noise variance of dynamic PET measurements. To address this issue, we propose a penalized direct estimation in which the cost function contains the Poisson likelihood, ridge regression and total variation (TV) terms. To obtain stable and reliable kinetic parameters, ridge regression has been used to solve NLS in kinetic parameter extraction from dynamic PET data. We exploit an alternating direction method of multipliers (ADMM) algorithm with separable quadratic surrogate (SQS) to solve the optimization problem. We also demonstrate that the proposed method is more accurate compared to the conventional direct estimation methods using simulation. Furthermore, the proposed method is implemented using GPU, which is approximately 100 times faster than using CPU.

M5BP-287, Dependency of the Quantitative Accuracy on Spatial Resolution and Temporal Sampling in Rb-82 Cardiac PET Imaging: a Monte-Carlo Simulation Study

R. Dai1, H. Peng1,2,3

1School of Biomedical Engineering, McMaster University, Hamilton, Canada
2Department of Medical Physics and Applied Radiation Sciences, McMaster University, Hamilton, Canada
3Department of Electrical and Computer Engineering, McMaster University, Hamilton, Canada

Positron emission tomography (PET) has been widely used to evaluate myocardial blood flow (MBF) using compartmental modeling. However, there exist two major challenges on achieving accurate MBF quantification using Rb-82: spatial resolution and temporal sampling. In this study, we investigated the dependency of quantitative accuracy of MBF (K1) on spatial resolution and temporal sampling for dynamic Rb-82 PET imaging using a two-compartment model. An NCAT phantom including cardiac motion was set up inside a PET scanner (crystal width: 2 mm and 4 mm). An arterial input function was selected based upon a 6-minute Rb-82 patient data and was used to configure the NCAT phantom. The interaction of 511 keV photons with the NCAT phantom and the PET system was performed using GATE Monte-Carlo simulation. Four different temporal protocols were tested to study the effect of temporal sampling. The estimation of kinetic parameter K1 was obtained using a two-compartment model and the bias of K1 was reported. For an input K1 of 2.1 mL/min/g, a high-resolution PET system using 2 mm crystal elements has the potential to reduce the bias in K1 estimation by 9% compared to a PET system using 4 mm crystal elements. Performing cardiac motion correction improves the accuracy of K1 estimation with a bias reduction of 20% for the 4 mm case, while results in less significant improvement (~6.2%) for the 2 mm case. The effect of temporal sampling is found to play a limited role in K1 estimation. For the 4 mm case with cardiac motion considered, the maximum difference in the bias of K1 amounts to ~7.6% among four different temporal protocols. The effect of spatial resolution plays a dominant role in K1 estimation and a high-resolution PET system using 2 mm crystal elements has the potential to reduce the bias in K1 estimation. Performing cardiac motion correction improves the accuracy of K1 estimation. The effect of temporal sampling is found to play a relatively limited role in K1 estimation.

M5BP-291, Quantitative Comparison of 18F-Fallypride PET Binding Potential Estimates Using Different Reference Tissue Models in Rat Brains

D. E. Lee, S. Muthusamy, D. A. Hammoud

Radiology and Imaging Sciences, National Institutes of Health, Bethesda, MD, USA

INTRODUCTION: Diverse methods for the quantification of dopamine D2/3 receptors in primate and rat brains using 18F-fallypride have been reported. The relatively novel compound has substantially higher affinity (33 pmol/L) than other benzamines, including 11C-raclopride, which makes it more sensitive for the detection of extrastriatal D2/3 receptors. However, the same high affinity necessitates lengthy scan times to obtain equilibrium binding estimates. In this study, we calculated and compared binding potential (BP) values obtained from simplified reference tissue approaches with varying lengths of emission recording. METHODS: 23 rodent brains were imaged with 18F-fallypride (~ 1 nmol/kg). The PET scans were acquired dynamically over 90 min using the Bio PET/CT. Striatal and extrastriatal regions were evaluated for differences in BP estimates in the same animal group using standard SRTM60 vs. SRTM90, 2T Reference, and simple binding ratios with the cerebellum as reference. The reliability of BP in all regions was examined by intraclass correlation coefficient (ICC) comparisons. RESULTS: The 2T Reference produced highly reliable BP estimates (ICC of 0.98). SRTM (60 and 90 min) underestimated the BP by up to 20% (ICC of 0.90 and 0.93, respectively) while the simple ratio method seemed to overestimate the BP of 18F-fallypride compared to the 2T Reference model. In contrast the mean 18F-fallypride BP estimates in extrastriatal regions were in close agreement irrespective of the method used. CONCLUSION: The choice of analysis method to derive the BP can influence the mean 18F-fallypride BP in high binding regions thus requiring lengthy scan times to obtain equilibrium binding estimates in the striatum. The 2T Reference with 90 min scan time appeared as the method of choice due to its higher reliability and absence of bias (closer to a linear function). Reference tissue methods for calculation of 18F-fallypride BP in rat brains are suitable for VOI analysis when no plasma input is available.

M5BP-295, Continuous Bed Motion Vs. Step-and-Shoot Acquisition on Clinical Whole-Body Dynamic and Parametric PET Imaging

N. A. Karakatsanis1, V. Garibotto1, O. Rager1, H. Zaidi1,2,3

1Division of Nuclear Medicine and Molecular Imaging, University of Geneva, Geneva, Switzerland
2Geneva Neuroscience Centre, University of Geneva, Geneva, Switzerland
3Division of Nuclear Medicine and Molecular Imaging, University of Groningen, Groningen, Netherlands

Continuous bed motion (CBM) has been recently introduced in the clinic as an alternative PET acquisition mode with respect to the traditional time-discretized step-and-shoot (SS) multi-bed acquisitions. In CBM mode, each slice can be considered as a different bed, since it is scanned over a different acquisition window. By reducing scan time discretization from a whole bed down to a single slice, the CBM acquisition offers additional degrees of freedom when designing acquisition protocols and thus larger margins for their optimization. Therefore, CBM mode could be particularly important for protocols of higher complexity, such as whole-body (WB) dynamic PET acquisitions. However, only a few studies have quantitatively validated the 2 modes on clinical data. In this work, we evaluate systematically CBM vs. SS quantitative performance on a set of clinical data acquired over a 0-60min period using a WB dynamic PET acquisition protocol on a Biograph mCT TOF scanner. The TOF resolution (580ps) of the scanner allowed selecting faster CBM speeds (4.2mm/sec) and equivalent SS frames (30sec/bed) eventually permitting acquisition of 12 WB total passes of which half were performed in CBM mode and half in SS mode by alternating the modes between successive passes to allow a balanced interleaving of post-injection time between the 2 modes to ensure their objective comparison. The evaluation of the 2 modes in terms of target-to-background (TBR contrast) and contrast-to-noise ratio (CNR) has been conducted for various noise levels by gradually adding all respective passes from each mode as well as in the parametric images after applying post-reconstruction WB Patlak analysis. Our results indicate a bias of ~10% between the 2 modes for the low count frames, reducing to <3% as we gradually add all respective frames. In addition, a small bias of <10% was observed in Patlak images at bed overlaps, which is attributed to the non-uniformity of the axial sensitivity profile of SS mode.

M5BP-299, Clinical Evaluation of Direct 4D Whole-Body PET Parametric Imaging with Time-of-Flight and Resolution Modeling Capabilities

N. A. Karakatsanis1, A. Mehranian1, M. Lodge2, M. E. Casey3, A. Rahmim2,4, H. Zaidi1,5,6

1Division of Nuclear Medicine and Molecular Imaging, University of Geneva, Geneva, Switzerland
2Division of Nuclear Medicine, Johns Hopkins University, Baltimore,MD, USA
3Siemens Molecular Imaging, Knoxville, TN, USA
4Department of Electrical and Computer Engineering, Johns Hopkins University, Baltimore,MD, USA
5Geneva Neuroscience Center, University of Geneva, Geneva, Switzerland
6Division of Nuclear Medicine and Molecular Imaging, University of Groningen, Groningen, Netherlands

Whole-body (WB) PET parametric imaging has recently become clinically feasible thanks to the introduction of multi-bed dynamic PET acquisition protocols, benefiting from the latest technologies in clinical PET scanners. Currently, the Time-of-Flight (TOF) capabilities of modern clinical PET systems allow for more accurate localization of the annihilation position along the line of response (LOR). As a result, TOF utilization can prevent propagation, during image reconstruction, of various resolution degrading factors and noise beyond their origin and across the image space, thus providing i) an inherent correction or contrast recovery mechanism and ii) an effective sensitivity gain relative to non-TOF acquisitions. In addition, the incorporation of the PET system’s point spread function (PSF) within the reconstruction system matrix has also recently resulted in i) similar or better contrast and ii) considerably lower image roughness. Recently, we explored the effect of TOF and PSF on WB indirect Patlak imaging. In this work, we systematically investigate the additional benefit of TOF and PSF on clinical studies when reconstructing WB Patlak images directly from projection data. Therefore, we developed a nested direct 4D Patlak WB reconstruction algorithm capable of i) utilizing TOF information, ii) modeling PSF with an effective space-invariant Gaussian resolution kernel and iii) supporting both standard and generalized Patlak analysis. Our clinical evaluation on a set of WB clinical studies acquired on the Siemens Biograph mCT TOF scanner indicated a 15-30% target-to-background (TBR) and contrast-to-noise ratio (CNR) enhancement in all examined regions and for both Patlak methods, compared to non-TOF non-PSF, when only TOF feature is enabled with an additional 5-10% improvement when combined with a 4mm FWHM Gaussian PSF kernel demonstrating the clinical benefits of TOF and PSF features when combined with highly efficient direct 4D WB Patlak reconstruction

M5BP-303, Gamma Emission in Hadron Therapy - Towards New Tools of Quality Assurance

A. Wronska1, P. Bednarczyk2, D. Boeckenhoff3, A. Bubak4, S. Feyen3, A. Konefal4, K. Laihem3, A. Magiera1, A. Stahl3, M. Zieblinski2, G. Obrzud1, K. Rusiecka1, L. Kelleter3, J. Leidner3

1Jagiellonian University, Cracow, Poland
2H. Niewodniczanski Institute of Nuclear Physics PAN, Cracow, Poland
3RWTH Aachen University, Aachen, Germany
4University of Silesia, Katowice, Poland

The Gamma-CCB experiment is focused on investigation of gamma emission in experiments modeling the course of hadron therapy. The main goal is to search for manifestation of the Bragg peak in the prompt gamma spectra. Experimental program comprises a series of measurements at different proton beam energies and for various phantom materials. In the two beam times performed by the group so far in CCB Cracow and HIT Heidelberg, data were taken with proton beam energy of 70 and 130 MeV and with three phantom materials: graphite, poly(methyl methacrylate) PMMA and polyoxymethylene POM, in order to study effects from different elements forming human tissue. Two different measurement modes were tested. In the first mode the gamma spectrum integrated over the whole beam penetration path in the phantom was registered, in the other mode only the gamma quanta originating in a phantom slice at a certain depth. In both measurement modes we observe strong correlation of the intensity of the carbon and oxygen excitation lines (4.44 and 6.13 MeV, respectively) with the Bragg peak position. Moreover, the correlation was found to be far more pronounced at backward angles than at the 90∘ polar angle studied so far.

M5BP-307, Detailed Requirements for a Laser-Based Proton/Ion Accelerator for Radioisotope Production

M. Seimetz1, P. Bellido2,1, A. Aguilar1, P. Conde1, A. J. Gonzalez1, A. Iborra1, L. Moliner1, J. P. Rigla1, M. J. Rodriguez-Alvarez1, F. Sanchez1, S. Sanchez1, A. Soriano1, R. Lera3, A. Ruiz-de la Cruz3, S. Torres-Peiro3, L. Roso2, J. M. Benlloch1

1Instituto de Instrumentacion para imagen molecular (I3M), Valencia, Spain
2Centro de Laseres Pulsados, Villamayor, Spain
3Proton Laser Applications S.L., Olerdola, Spain

Background and objectives: Laser-plasma acceleration of protons and ions is often considered a promising technique for compact applications of highly intense beams of multi-MeV particles. A remarkable example is the on-site production of short-lived radioisotopes for medical and preclinical interventions. We study quantitatively the activity of four important PET isotopes which may be obtained by irradiation of suitable target nuclei with laser-accelerated protons and deuterons. These simulations allow for confining the range of useful parameters of a laser-based production system. Methods: We choose a total of ten p- and d-induced reaction channels for the production of F-18, C-11, O-15, and N-13 from suitable target nuclei. We calculate the activity yield as a function of projectile energy starting from the corresponding, known cross sections. In order to simulate typical laser-plasma particle spectra we generate exponentially decaying distributions spread over a wide range, up to a maximum energy between 6 and 16 MeV. From the yield curves and the spectra we obtain the single-shot activation and the total activity after a realistic production time at 100 Hz pulse rate, taking into account saturation effects due to decay during irradiation. Results: We present numerical results for ten reaction channels and six realistic projectile spectra. With single laser shots, the highest activities are generated for O-15 (up to 11 (20) kBq for 16 MeV maximum p (d) energy). After prolonged irradiation at 100 Hz pulse rate, useful quantities of C-11 and O-15 may be obtained from spectra with 10 MeV maximum energies. The production of N-13 and F-18, to the contrary, requires much higher energies and/or shot rates. Conclusions: 10 MeV particle energy and 100 Hz pulse rate are realistic benchmarks for a laser-based PET isotope production system. Experimental work to achieve these demanding objectives is in progress.

M5BP-311, Influence of Proton Scattering Angles on Energy Radiograph in Proton Radiography: a Simulation Study

A. K. Biegun1, J. Takatsu2, M. van Beuzekom3, E. R. van der Graaf1, M.-J. van Goethem4, T. Klaver3, J. Visser3, S. Brandenburg1

1KVI-Center for Advanced Radiation Technology (KVI-CART), University of Groningen, Groningen, The Netherlands
2Department of Radiation Oncology, Graduate School of Medicine, Osaka, Japan
3National Institute for Subatomic Physics (Nikhef), Amsterdam, The Netherlands
4Department of Radiation Oncology, University of Groningen, University Medical Center Groningen, Groningen, The Netherlands

Proton radiography is one of the novel imaging modalities in which the proton stopping powers can be determined directly. The quality of the proton radiography image may be reduced by multiple Coulomb scattering and energy loss straggling of protons, especially in more dense materials. These will affect the position resolution in the object and will lead to blurred images. To avoid this scenario, scattering angles of protons have to be studied with a special care. We study energy loss and scattering angle radiographs. We simulate with Geant4 4.9.6.p03 a proton radiography detection system with two ideal 10 x 10 cm2 position sensitive detectors, and energy detector of 15 cm diameter and 15 cm thickness. A small cylindrical phantom of 2.5 cm diameter and a length of 2.5 cm, filled with different (also tissue-like) materials is placed between two position detectors. We calculate the energy loss radiograph for proton beam energy of 150 MeV and we analyze the influence of proton scattering angle on the obtained energy loss images. Only protons traveling in a straight line should be evaluated to recognize accurately neighboring materials. Few scattering angle conditions have been applied. Energy radiograph for protons with scattering angles below ?<26.2 mrad shows the biggest blurring, while for protons that scattered only up to ?<8.7 mrad the image is more sharp. By applying cuts for proton scattering angles the energy loss radiograph can give important information about positions of neighboring materials, especially when the density difference of these materials is very small. This can help to determine more accurately the type of material in the proton beam in proton therapy, crucial for delivering a proper dose in the correct location. The analysis for other materials is ongoing.

M5BP-315, Simulation of Heterogeneous Microsphere Distribution using Hepatic Arterial Tree Model in Yttrium-90 Microsphere Therapy and Imaging

G. S. K. Fung, N. Crookston, A. K. Jha, Y. Du, E. C. Frey

Department of Radiology, Johns Hopkins University, Baltimore, MD, United States

The aim of this work is to develop a hepatic arterial tree (HAT) model to simulate the heterogeneous microsphere (MS) distribution in Y-90 MS therapy and imaging. It is known the MSs have a non-uniform distribution inside the liver depending on the size and number of injected MSs. Accurate prediction of liver toxicity and realistic simulation of Y-90 MS PET and SPECT require realistic models of the MS distribution. A physiological-realistic HAT model provides the anatomical details, including the lumen diameters and the flow distribution in the vasculature, that are essential for simulating the spatial distribution of MSs. A macrocell-based approach was developed to generate the HAT through growth cycles. Conservation of blood, Murray’s Law for the relationship between parent and daughter vessels’ radii, and Poiseuille’s Law for the pressure change at the ends of the vessel were enforced at each bifurcation. The model started with the liver shape and the large hepatic arterial tree, e.g., segmented from patient CT images or obtained from digital phantom. This was then scaled to 1/20 of the adult size. Macrocells were uniformly generated inside the scaled liver and connected to the vasculature with optimal bifurcation angles. In subsequent cycles, macrocells proliferated or died according to the probabilities defined by macrocell type and cycle number. When the liver volume was filled to normal density, the liver surface was expanded and the macrocells spread out uniformly inside the new volume. Cycles continued until an adult liver size was reached. We used XCAT phantom to initialize the liver shape and HAT. An adult size HAT and liver, with a volume of 1767ml, and containing 13807 macrocells and 27612 vessel segments, was obtained after 15 growth cycles. The generated HAT and macrocell models were analyzed in terms of 3-D MS distribution maps and histograms. This HAT model provides an important tool for optimizing and evaluating MS therapy planning and imaging.

M5BP-319, Total Variation Reconstruction of 12C Beams Measured with the Whole Body Dual Ring OpenPET Scanner

J. Cabello1, H. Tashima2, E. Yoshida2, S. I. Ziegler1, T. Yamaya2

1Nuklearmedizinische Klinik und Poliklinik, Klinikum rechts der Isar der Technischen Universität München, Munich, Germany
2National Institute of Radiological Sciences, Chiba, Japan

Beam monitoring in hadron therapy is of paramount importance to accurately determine whether the delivered dose is within the prescribed limits. In-beam (ib)-PET is an attractive candidate to estimate the distribution of ß+-emitters during or immediately after irradiation, while the patient is still lying on the treatment bed, to maximize the amount of detected annihilation photons. Total variation (TV) is commonly used in CT to reduce truncation artifacts due to gaps, which ib-PET scanners also have. In this work we investigated the experimental 12C beam range error measured at the HIMAC facility using TV and a quadratic function (QF) as priors for a MAP-EM reconstruction algorithm (one-step-late), measured with the whole body dual ring OpenPET. The whole body OpenPET consists of two rings separated by a 90 mm gap for the beam irradiation. Resulting ß+-emitters distributions produced by 12C beams, applied to PMMA blocks placed at different distances from the beam entrance, were reconstructed and analysed in terms of image noise and range accuracy. Results showed that using ML-EM the reconstructed images were too noisy to measure the beam range. In comparison, TV and QF produced reconstructed images with low noise, enabling the possibility of measuring the beam range. TV in general produced lower noise images than QF and more robustness to the hyperparameter required in the OSL implementation of MAP-EM. The proper choice of the hyperparameter was a critical part of the present study, hence a wide range of values was analysed. The beam range error measured with QF was =2 mm while the range error measured with TV was in the range of 1-2 mm.

M5BP-323, A Bi-Modal System for the on-Line Verification of Particle Therapy Treatments

M. G. Bisogni1,2

1Department of Physics, University of Pisa, Pisa, Italy
2INFN, pisa, italy

On behalf of the INSIDE Collaboration

The INSIDE (INnovative SolutIons for DosimEtry in hadrontherapy) project is a research program aiming at the development of an integrated system for the on-line quality control of particle therapy treatments. In this project the in-beam PET and the prompt particle detection techniques have been combined in a unique imaging device to reconstruct the actual particle range within the target volume to be compared with the planned one. PET imaging is a well assessed monitoring technique that exploits the correlation between the particle range and the distribution of the positron emitters induced by nuclear reactions in the irradiated tissues. The INSIDE PET system is integrated in the gantry and operated during the irradiation so as to provide an immediate feedback on the dose deposition. The scanner, tailored for the monitoring of head and neck tumors, is composed of two planar detectors, 10 cm (transaxially) x 25 cm (axially), each to be placed at 25 cm from the isocenter. The detectors are made of arrays of LSF pixel crystals, 3x3x20 mm^3 each, coupled one to one to Silicon Photomultiplers (SiPM). A custom designed FE and DAQ system are being developed to cope with the rates typical of a therapeutic beam. The PET activity map created in the target is complemented by the beam profile, obtained by tracking the prompt secondary particles (mainly protons) coming from the interaction of the primary beam with the target nuclei and from projectile fragmentation in case of carbon beams. The tracking system is composed of 6 planes of orthogonal squared scintillating fibers coupled to 1 mm^2 SiPMs. An electromagnetic calorimeter made of LYSO crystal arrays coupled to Position Sensitive PMTs is placed downstream of the tracker and provides the residual energy of the charged particles. This paper describes in detail the INSIDE project and summaries its achievements.

M5BP-327, Pencil Beam Approach to Proton Computed Tomography: a Performance Study

R. Rescigno1,2, C. Bopp1,2, M. Rousseau1,2, D. Brasse1,2

1Universit� de Strasbourg IPHC, Strasbourg, France
2CNRS UMR 7178, Strasbourg, France

Proton computed tomography (pCT) has the potential to improve the accuracy for proton treatment planning. It will also be useful for pretreatment verification of patient positioning relative to the proton beam. The current approach to pCT requires that individual protons are tracked one by one. This requirement implies the development of fast detectors able to sustain a mean data rate of about 2 MHz. However accelerators used in the framework of hadrontherapy deliver beams of particles with a specific temporal structure, usually optimised for therapy purposes. In some cases a prohibitive data rate has to be handled if this parameter is included in the pCT design specification. Recently the pencil beam (PB) approach to pCT was proposed. Protons are not tracked one by one but a beam of particle is considered instead. The mathematical formalism of such approach was previously defined however the achievable performances were not yet estimated. In this study these performances were evaluated and a comparison with the classical proton by proton approach is presented.

M5BP-331, Real-Time Magnetic Electron Energy Spectrometer for Use with Medical Accelerators

P. E. Maggi1, K. L. Matthews II1, R. Carver2, K. R. Hogstrom1,2

1Louisiana State University, Baton Rouge, LA, USA
2Mary Bird Perkins Cancer Center, Baton Rouge, LA, USA

For cancer centers with multiple linear accelerators (linacs), matching of treatment beams allows patients to be treated on any linac without needing to recalculate machine-specific treatment plans. Beam matching is performed at the time of installation, typically using a scanning water tank to measure depth-dose curves. We have developed a prototype real-time magnetic electron energy spectrometer for tuning the electron beams of linacs. A real-time spectrometer has the potential to simplify and hasten accelerator tuning, and also allows additional characterization of a linac not possible through depth-dose measurements alone. Our spectrometer uses a 0.54 T permanent magnet block as the dispersive element, and scintillating fibers coupled to a CCD camera as the detector. A key component of the real-time spectrometer is the algorithm that unfolds the energy spectrum from the spectrometer’s output image. This work describes the real-time spectrometer system, the detector response model, and the spectrum unfolding algorithm. Examples of reconstructed spectra from both simulated and measured electron beams are presented.

M5BP-335, Testing a Pre-Clinical Proton-CT Head Scanner

R. P. Johnson, Member IEEE1, T. E. Plautz, Member IEEE1, H. F. W. Sadrozinski, Senior Member IEEE1, A. Zatserklyaniy1, V. Bashkirov, Member IEEE2, R. F. Hurley2, R. Schulte, Member IEEE2, B. Schultze3, K. Schubert, Senior Member IEEE3, V. Giacometti, Member IEEE4

1Physics, University of California, Santa Cruz, Santa Cruz, CA, USA
2Radiation Research Labs, Loma Linda University, Loma Linda, CA, USA
3School of Engineering & Computer Science, Baylor University, Waco, TX, USA
4Centre for Medical Radiation Physics, University of Wollongong, Wollongong, NSW, Australia

We are exploring low-dose proton radiography and computed tomography (pCT) as techniques to improve the accuracy of proton treatment planning and to provide artifact-free images for verification and adaptive therapy at the time of treatment. Here we report on comprehensive beam test results with our pre-clinical (Phase 2) pCT head scanner. The system consists of two silicon-strip telescopes that track individual protons before and after the phantom or patient, and a novel multistage scintillation detector that measures a combination of the residual energy and range of the proton, from which we derive the water equivalent path length (WEPL) of the protons in the scanned object. The detector system and data acquisition attain a sustained rate of more than a million protons individually measured per second, allowing a full CT scan to be completed in 5 to 10 minutes of beam time. The set of WEPL values and associated paths of protons passing through the object over a 360 degree angular scan is processed by an iterative parallelizable reconstruction algorithm that runs on modern GP-GPU hardware. In order to assess the performance of the scanner for proton radiography and computed tomography, we have performed scans using both the synchrotron of the Loma Linda University Medical Center and a modern cyclotron at the Northwestern Medicine Chicago Proton Center. The synchrotron beam was scattered by lead foils into a cone that encompassed the instrument aperture, while the cyclotron beam was modulated across the aperture. We performed scans on a series of phantoms, including a custom phantom designed to assess the spatial resolution, a CATPHAN 404 phantom to assess the measurement of relative stopping power, a dosimetry phantom, and a pediatric head phantom. Performance and dosimetry results from those phantom scans are presented together with a description of the instrument and data acquisition system.

M5BP-339, Characterization and Simulation Results of a Two/Three-Layer Compton Telescope with LaBr3 and SiPMs

E. Muñoz, J. Barrio, A. Etxebeste, C. Lacasta, J. F. Oliver, C. Solaz, P. Solevi, M. Trovato, G. Llosa

Instituto de Fisica Corpuscular (CSIC/UVEG), Valencia, Spain

A three layer Compton telescope for dose monitoring in hadron therapy has been developed at IFIC - Valencia. It consists of three detector layers and each layer is made of a continuous LaBr3 crystal coupled to four silicon photomultiplier arrays. A dedicated image reconstruction method has also been developed. The telescope has been characterized and tested with radioactive sources and in-beam. Simulations with GATE have been carried out in order to optimize the detector geometry and response. Ongoing work is mainly focused on performance optimization and on the simultaneous acquisition and processing of two- and three-layer interaction events.

M5BP-343, Development and Evaluation of a Simultaneous PET/MR Scanner Model Using GATE

F. Monnier1, H. Fayad1,2, J. Bert1, H. Schmidt3, D. Visvikis1

1LaTIM, INSERM UMR 1101, Brest, France
2Université de Bretagne Occidentale, Faculty of Medicine, France
3University Hospital of Tübingen, Tübingen, Germany

Innovative processing methods for simultaneous PET/MR acquisition need a reliable tool to validate their efficiency. The assessment of such methods benefit from controlled experiments that allow obtaining reproducible ground truth raw data such as the ones provided by the GATE open source software. The purpose of this work was to develop and validate the model of the Siemens Biograph mMR PET/MR system in GATE. The full system geometry and detection process of the Siemens Biograph mMR was modeled using GATE. The accuracy of the developed model was assessed by comparing simulated PET datasets of physical and anthropomorphic phantoms to corresponding acquired raw datasets and reconstructed images using a Siemens Biograph mMR system. Attenuation maps from both a spherical phantom having variable levels of activity concentrations and a patient study using a 68Ga DOTATATE PET protocol were exploited. Simulation parameters, such as simulation time, number of detected coincidences, etc were extracted from the clinical acquisitions (. Several evaluation indices were used to compare the reconstructed images, including a qualitative profile comparison and quantitative tissue activity ratios. Good agreement of profiles and contrast indices were obtained between simulated and corresponding acquired PET images. The activity concentration ratios between tissues showed a close agreement (differences <4%) between the simulated and acquired datasets of both the phantom and patient PET reconstructed images. The scatter fraction agreement was within 3%. Our results suggest that the numerical model of the Biograph mMR reproduces accurately the performance of a clinical Biograph mMR PET scanner. Such model could be used to evaluate PET/MR developed reconstruction/correction algorithms and be combined with a corresponding MR simulation of the MR imaging component.

M5BP-347, Modeling the GE Signa PET-MR with Monte-Carlo Simulations Using GATE

M. Khalif?1, S. Stute1, A. Wagadarikar2, S. Jan1, C. Comtat1

1IMIV, UMR 1023 Inserm/CEA/Universit? Paris Sud - ERL 9218 CNRS, CEA/I2BM/SHFJ, Orsay, France
2GE Global Research, Niskayuna, NY, USA

PET-MR is an emerging technique that has been studied in the last decade for its numerous advantages in medical imaging. Combining nuclear imaging with MRI allows acquiring anatomical, functional and molecular images, thus providing a good anatomical reference for a better diagnosis, a shorter time spent in the scanner for the patient as well as a reduced radiation dose. However, the combination of PET and MR imaging raises new methodological challenges, like photon attenuation correction, but also offers new opportunities such as motion correction, MR-guided PET image reconstruction and multi-parametric imaging. To address these issues, it is necessary to have a realistic numerical model of the system PET components. For this, we used GATE for Monte Carlo simulations and we built a realistic model of the GE Signa integrated PET-MR. We applied the NEMA protocols to estimate absolute sensitivity, spatial resolution and scatter fraction and compared the results with the published measured values. Sensitivity was simulated for the NEMA sensitivity phantom while considering energy readout levels at block and at sub-block. Spatial resolution was computed for 6 positions of a point source in the field of view (FOV) and the scatter fraction was determined for a line source inserted in cylindrical phantom. Results showed good agreement with the measured data for sensitivity at block level (22.3kcps/MBq vs 22.9kcps/MBq) and for spatial resolution (4.2mm vs 4.2mm radially and 5.7mm vs 5.7 mm axially at 1 cm; 5.9 mm vs 5.8 mm radially, 4.6 mm vs 4.4 mm tangentially and 5.7 mm vs 7.0 mm axially at 10 cm). The simulated scatter fraction (38%) underestimates the measured one (43%). Future work will include the modeling of the inter-block scatter recovery inside GATE . M. Khalif?, S. Stute, S. Jan and C. Comtat are with IMIV, UMR 1023 Inserm/CEA/Universit? Paris Sud - ERL 9218 CNRS, Orsay, France. A. Wagadarikar (now at Apple) was with GE Global Research, Niskayuna, NY, USA.

M5BP-351, Dose calculation from photoneutrons emitted in Radiotherapy Treatments by means of MCNP6 simulation and unstructured mesh

S. Morató, B. Juste, R. Miró, G. Verdú

Instituto de Seguridad Industrial, Radiofísica y Medioambiental (ISIRYM), valencia, Spain

Medical linear accelerators (LINAC) often use high energies that are able to generate neutrons. Most of 3D conformal radiotherapy treatments use linac emitting photons and electrons to impart absorbed dose in the patient's tumor. When the LINAC operates with higher voltages than 8 MeV, undesirable neutrons are generated by photonuclear reactions (gamma, n) with collimators, flattening filters, MLCs , etc; which are composed of high atomic number materials, such as lead and tungsten, whose cross section presents Giant Dipole Resonance (GDR). Thus, the therapeutic external beam from Linac is contaminated by neutrons. In this work, MCNP6 Monte Carlo code is used to quantify the neutron absorbed dose received by a patient, and Abaqus / CAE software to generate the geometry of RANDO phantom by means of unstructured mesh. The purpose of this work is lay bare the possible dose received by a patient during treatment due to induced neutron which are not considered in most of current radiotherapy treatments planners.

M5BP-355, Monte-Carlo Simulation of a Dual-Layer Argon-Xenon Compton Gamma Camera

A. Berdnikova, A. Bolozdynya, V. Belyaev, L. Dubov, Y. Shtotsky

National Research Nuclear University MEPhI, Moscow, Russian Federation

A cylindrical Ar-Xe Compton gamma camera for radionuclide imaging Geant4/Gate simulation results are presented in this study. The goal of modeling was to optimize the parameters of the gaseous scattering and absorption detectors to obtain the highest possible registration efficiency of useful events. The considered geometry of the Compton camera is the coaxial system of two cylindrical detectors. The optimal design of the scattering detector is the Ar-gaseous chamber of R ˜ 35 cm thickness with an argon pressure P = 15 atm, which provides the necessary scattering efficiency ec ˜ 10%. Xenon was chosen as a working gas for the absorption detector, and the optimal design is gaseous chamber of R ˜ 30 cm thickness with a xenon pressure P = 15 atm, which will provide the efficiency of photoelectric absorption ep ˜ 72%. Geant4/Gate simulation of the Compton camera in cylindrical geometry with a point Tc-99m gamma source (140 keV) was carried out to estimate the contribution of different types of coincidences and develop the coincidences selection algorithm. For evaluation of the image quality dependence on the gamma camera geometry the ellipse-stacking method (ESM) and the maximum likelihood estimation method (MLEM) were implemented for the image reconstruction of 2D projections on the central cross-section of the camera of two point Tc-99m sources, placed at the distance of 40 mm between each other. FWHM spatial resolution is ~ 40 mm with the ESM method, and ~ 20 mm with MLEM algorithm. The coincidences registration efficiency was ~ 3 % while selecting the events with the total energy released (ESD+EAD) > 100 keV, and ~ 0,5 % with the energy window for the SD detector ESD = 10 ÷ 29 keV, energy threshold for the AD detector EAD > 70 keV. The results of Monte-Carlo simulations proved that proposed design of the Compton camera provides the sensitivity gain by approximately two orders of magnitude in comparison with an Anger gamma camera with mechanical collimation.

M5BP-359, Brain PET Imaging Based on the Axial PET Concept: a Simulation Study

P. Solevi1, G. Reyéns-Llompart1, J. Cabello1,2, J. E. Gillam1,3, C. Joram4, J. F. Oliver1, M. Rafecas1,5

1IFIC (CSIC/UV), Valencia, Spain
2now at Klinikum rechts der Isar, Technische Universität München, Munich, Germany
3now at Brain & Mind Research Institute, University of Sydney, Sydney, Australia
4CERN, Geneva, Switzerland
5now at Institute of Medical Engineering, University of Lübeck, Lübeck, Germany

Brain PET imaging is used for regional localization of tumors and through the use of novel radio-tracers to study the diagnostics, pathogenesis, and staging of several diseases. However novel radio-tracer development should be accompanied by equal advancements in instrumentation. The Axial PET (AX-PET) concept based on long axially oriented crystals was originally conceived for brain imaging. It decouples spatial resolution and sensitivity thus representing an ideal detector concept for small diameter PET scanners. Two AX-PET modules have been fully assembled and characterized at CERN and a detailed Monte Carlo model of the demonstrator was developed and extensively validated against experimental data. In this work, we have extended the model to simulate a brain scanner based on the AX-PET concept. Two detector arrangements are investigated and compared, one based on the standard detector design and the other on a slanted crystal positioning that provides a compact design with reduced gaps between modules. An important feature of AX- PET is the large fraction of Inter-Crystal Scatter (ICS) events whose impact at count-rate and image level is studied in details. A reformulation of the Noise Equivalent Count-rate (NEC) is proposed in order to take into account ICS inclusion into the process of image reconstruction. Spatial resolution and image quality phantoms have been simulated and reconstructed in order to estimate the performance of the brain AX-PET scanner and to study the influence of the two different detector arrangements. Figures of merit point towards a better performance of the standard detector arrangement, while for both systems ICS inclusion significantly improves image noise in reconstructed images.

M5BP-363, Radiotherapy Treatment of Mouse Tumor and Impact of Beam Energy on Nearest Organs: a Monte Carlo Investigation

M. Hamdi1, M. Mimi1, M. Bentourkia2

1Department of Electrical Engineering, University of Mostaganem, Mostaganem, Algeria
2Nuclear Medicine and Radiobiology, Université de Sherbrooke, Sherbrooke, Quebec, Canada

Purpose: Radiotherapy treatments to local tumors are always associated with dose deposit in surrounding tissues and even in distant tissues not traversed by the radiation beams. Also, depending on the location of the tumor, a beam energy and orientation should be determined prior to radiation treatment. In the present work, we report absorbed doses in a lung tumor and in secondary distant organs not on the beam path in a digital mouse irradiated by different radiation energies and orientations using Monte Carlo simulations. Methods: We simulated a set of seven radiation energies of 50, 100, 150, 200, 250, 350 and 450 keV, each in seven irregularly incremented beam angles targeting a lung tumor of 1.4 mm in diameter, and we analysed energy transfer as a function of photon interaction types and related absorbed doses in 8 volumes in the lung, the heart and the spine. We also simulated the energies in form of energy spectra with the same kilovoltage peaks. Results: The results showed an increase of absorbed dose as a function of energy even in the organs not intersected by the beams with lowest effects for the 100 keV radiation energy. The tumor received doses (Gy) as a function of radiation energies: 0.91 1.44 2.44 3.56 4.68 6.81 8.57. The spinal cord, of comparable size to the tumor and excluding the spinal bones, which was not directly irradiated by the beams, received a dose representing in average 1% of that of the tumor, while the spinal bone received doses of 6.6 and 0.12 times those in the tumor at 50 and 450 keV, respectively. As the spectral radiations contained different energies, we observed different shapes of the absorbed doses in the organs as a function of energy, in comparison to the monoenergetic radiations. Conclusions: Such Monte Carlo simulations could be necessary to select the appropriate beam energy, photon flux and beam angles to efficiently treat tumors and to reduce the impact of the radiations in the other organs in small animal models.

M5BP-367, High Linearity 3D Scintillation Positioning Method for Gamma Scintillation Detector

B. Wang1, D. Larsson1, P. Bennati1, I. Valastyán1,2, M. Colarieti-Tosti1,3

1School of Technology and Health, KTH Royal Institute of Technology, Stockholm, Sweden
2Institute for Nuclear Research, MTA-Atomki, Hungarian Academy of Science, Debrecen, Hungary
3Department of Clinical Science Intervention and Technology, CLINTEC, Karolinska Institute, Stockholm, Sweden

In many Single Photon Emission applications, accurate 3D positioning over the entire volume of the scintillation crystal is important. Even though such work has been done, positioning linearity over the whole detector has been less studied. The aim of this study was therefore to evaluate 3D positioning in a front and back light (double) readout SiPM-based gamma detector. A detector consisting of a LaBr3:Ce crystal with evaluated setups of single/double readout as well as white/black reflector sides was created in Monte Carlo simulation, with realistic electronic-noise added a posteriori. A library of light distributions from known scintillation positions was created and from this, two library-based positioning methods were evaluated: Maximum Similarity (MS, minimizing the Euclidean distance to the reference library) and standardized Maximum Likelihood (ML) positioning. The setup and algorithms were tested with regards to energy resolution as well as 3D spatial resolution and linearity. Results indicated a clear increase in energy resolution with white reflector sides. For 3D positioning, both ML and MS showed significantly higher resolution and linearity compared to conventional Anger logic. For identical setups, MS proved slightly better than ML with regards to spatial positioning: a fact probably due to the extreme similarity peak experienced in MS with small Euclidean distances. These results suggest that a gamma detector with double readout and white reflectors in combination with MS positioning may provide a valuable combination of maximized energy output and highly linear spatial resolution in 3D. With this work being exclusively simulation-based, future expansions should include experimental validation as well as evaluation of the mapping of a simulated 3D library to a real detector.

M5BP-371, Optimization of a High Resolution Focussed Molecular Breast Tomosynthesis Device

J. van Roosmalen1, M. C. Goorden1, F. J. Beekman1,2

1Section of Radiation, Detection and Medical Imaging, Delft University of Technology, Delft, The Netherlands
2MILabs B.V., Utrecht, the Netherlands

Objectives: Molecular breast imaging (MBI) with dedicated gamma cameras is rapidly gaining in popularity. We recently launched a novel Molecular Breast Tomosynthesis (MBT) device. The device uses linearly moving multi-pinhole collimators, with the patient in prone position. Gamma cameras with multi-pinhole collimators are positioned on both sides of the slightly compressed breast. The aim of the present paper is to optimize the geometry of the system for focussed and whole breast imaging. Methods: Using analytical models that include photon penetration of pinholes, detector distance from the breast was optimized to maximize photon sensitivity at fixed system resolution. Moreover, we used full system simulations to determine the effect of intrinsic detector resolution, and detector discretization. For comparison, simulations of an existing high-performance planar molecular breast imaging device were included for micro-Derenzo resolution phantoms differently oriented in the breast. Results: We found that a detector at approx. 4 cm from the breast yielded optimal resolution sensitivity trade-off. Given the modest system resolution requirements, it turned out that improving the intrinsic resolution of the detector beyond the current 2.8 mm had little benefit, nor did the use of a pixelated detector show any significant difference. In Derenzo phantoms, the optimal system managed to reach a resolution of 3.0 mm in the transverse plane. The planar breast imaging system resolved the 3.5 mm rods at best. In contrast to the planar system, which does not provide any depth information, MBT was found to be able to resolve rods with a 4.0 mm diameter in the depth direction. Conclusions: Our simulations show that relative close by detector placement lead to highest performance and that pixel sizes of less than 3 mm are sufficient. Overall, our results indicates that multi-pinhole MBT can significantly increase tumour-to-background CNR and resolution compared to planar MBI.

M5BP-375, Advantage of Pinhole Collimators over Parallel Hole Collimators in Reducing Downscatter for I-123 Imaging

A. Konik1, J. De Beenhouwer2, M. A. King1

1Radiology, Umass Medical School, Worcester,MA, USA
2iMinds-VisionLab, University of Antwerp, Antwerp, Belgium

With the usage of multiple-pinhole (MPH) collimators, improvements in resolution and sensitivity can be achieved over parallel and fan-beam collimators. Earlier, we proposed an inexpensive method to improve the performance of the existing general-purpose dual-camera SPECT systems for brain imaging with I-123 (DaTscan) by using a specifically designed multi-pinhole (MPH) collimator on one of the detector heads. We present here our preliminary work investigating the downscatter (backscatter and collimator scatter) of the low-abundance high-energy photons of I-123, for our MPH collimator in comparison to pinhole and parallel hole collimators. These investigations were conducted with GATE simulations and validated with experimental studies with a Philips Prism2000 SPECT system (1.905 cm crystal) for which we have pinhole and parallel-hole collimators. The downscatter fractions (DSF) obtained from the Prism2000 experiments were 29.2% for the LEUHR and 3.9% for the pinhole collimators, which are similar to the simulation results: 28.4% for the LEUHR and 4.6% for the pinhole collimators. DSF for the MPH (0.9525 cm crystal) was even lower than the Prism2000 pinhole results, where total scatter was only 3.7% (1.5% Compton and 2.2% Rayleigh). These quantitative results were in well agreement with the energy spectra obtained from the simulations of the SPECT acquisitions using parallel LEUHR, single pinhole and MPH collimators. The spectrum obtained from Prism2000 LEUHR system showed a much larger scatter component compared to the Prism2000 pinhole and MPH SPECT systems. Thus, our study shows that an additional benefit of the MPH is the negligible downscatter for I-123 brain imaging.

M5BP-379, The Simulation Study on Capacitive Charge-Dividing Multiplexing for SiPM Based High Resolution Brain PET

H. Park, G. B. Ko, J. S. Lee

Seoul National University, Seoul, Korea

In this study, we evaluated three kinds of 8 × 8 capacitive charge-dividing multiplexing methods, which were modified by substituting resistors into capacitors based on the well-known resistive multiplexing schemes (Anger logic, symmetric charge division [SCD], and dual-split circuit). For the simulation, Hamamatsu 3 mm × 3 mm SiPM model were used and the single cell response of SiPM coupled with LSO crystal was modeled. The encoded position signals from the multiplexing circuit (let us denote them as A, B, C, and D) were amplified by fast current-feedback operational amplifiers (AD8000; Analog Device, USA) with 50 O input impedance. We tested the multiplexing schemes while changing capacitor values and rise time was measured. Furthermore, for the sake of investigating the position decoding accuracy which is determined by both dynamic range and charge collection efficiency of multiplexing method, Gaussian noise were fed into each output position signal A, B, C, and D. For the fixed DWR level, 50, the largest dynamic range and the best pulse rise time were obtained in the case of Anger scheme, followed by the Dual-split and SCD scheme in descending order. From the simulation results, we found out that capacitive multiplexing schemes show promising results for the high performance PET applications in terms of rise time and flood map quality. Among them, Anger scheme exhibited higher performance in comparison of other two schemes. Therefore, we can conclude that Anger logic based capacitive multiplexing scheme can be a good candidate for highly sensitive brain-dedicated PET system. However dual-split scheme also can be the alternative in terms of complexity, since the Anger scheme need much more passive components (i.e. capacitors).