M6A2  Simulation and Modeling of Imaging Systems

Saturday, Nov. 7  08:30-10:00  Pacific Salon 1&2

Session Chair:  Stefaan Vandenberghe, Ghent University, Belgium; Scott Wollenweber, GE Healthcare, United States

Show/Hide All Abstracts

(08:30) M6A2-1, Simulation Study for Designing a Compact Brain PET Scanner

K. Gong1, S. Majewski2, P. E. Kinahan3, R. L. Harrison3, B. F. Elston3, R. Manjeshwar4, S. Dolinsky4, A. V. Stolin5, J. A. Brefczynski-Lewis6, J. Qi1

1Department of Biomedical Engineering, University of California, Davis, CA, USA
2Department of Radiology and Medical Imaging, University of Virginia, Charlottesville, VA, USA
3Department of Radiology, University of Washington, Seattle, WA, USA
4GE Global Research Center, Niskayuna, NY, USA
5Department of Radiology, West Virginia University, Morgantown, WV, USA
6Department of Physiology & Pharmacology, West Virginia University, Morgantown, WV, USA
The desire to understand normal and disordered human brain of upright, moving persons in natural environments motivates the development of an ambulatory micro-dose brain PET imager (AMPET). An ideal system would be light weight and have high sensitivity and spatial resolution. These requirements are often in conflict with each other. Therefore, we performed simulation studies to search for the optimal system configuration and to assess the improvement in performance over existing scanners. An intuitive design to achieve high sensitivity is to use a tight geometry that covers the brain. However, a tight geometry also increases parallax error in peripheral lines of response, which may increase the variance in ROI quantification. In this study, we first simulated cylindrical PET scanners with different ring diameters. All configurations are subjected to the same maximum weight constraint by restricting the amount of detector materials. We computed the Cramér–Rao variance bound to compare the performance for ROI quantification using different scanner geometries. The results show that while a smaller ring diameter can increase photon detection sensitivity and hence reduce the variance in the center of the field of view, it can result in higher pixel variance in peripheral regions when the length of the detector crystal (LSO) is 15 mm or more. The variance can be substantially reduced by adding depth of interaction (DOI) measurements to the detectors. Our simulation results also show that the relative performance highly depends on the size of the ROI, and a large ROI favors a tighter geometry even without DOI information. Based on these results, we propose a tight helmet PET with DOI detectors for brain imaging. Monte Carlo simulations using GATE show the helmet PET can achieve higher sensitivity and better image quality than traditional cylindrical scanners.

(08:45) M6A2-2, A Monte Carlo Study of Inter-Crystal Scattering and Coincidence Sorting for MADPET4

N. Omidvari1, J. Cabello1, F. Schneider1,2, S. Paul2, S. I. Ziegler1

1Nuklearmedizin, Klinikum rechts der Isar, Technische Universität München, Munich, Germany
2Physik Department E18, Technische Universität München, Garching, Germany
MADPET4 is a high resolution PET insert under development for use in a 7T MR. To fully exploit the capabilities of the insert, the physical interactions of the emitted photons in the scanner need to be understood. Consequently, these processes can be modeled and considered in the image reconstruction. The aim of this study is to investigate the effects of different physical interactions in an accurate model of MADPET4 using Monte Carlo simulations. The main focus of the study was the analysis of different types of coincidences, in order to identify the sources of line of response (LoR) mispositioning and their contribution to image noise. Furthermore, the impact of the different active and passive components of the system on image quality was investigated. Interestingly, noise evaluation results of the image quality phantom showed that by applying a geometrical condition of 40 mm in the coincidence sorting, while the field-of-view was preserved, better image quality and higher sensitivity was achieved at 50 keV threshold compared to a threshold of 350 keV. Including the system passive components in the GATE model resulted in an increase in sensitivity, signal-to-noise ratio, and contrast-to-noise ratio due to the scattering in the plastic structure holding the crystals. Finally, quantification of different types of coincidences showed a potential relative gain of 70% in sensitivity and 30% noise reduction by inclusion of ICS events in the system response matrix (SRM) at 50 keV threshold and 40 mm geometrical condition. On contrary to other studies, where the inclusion of ICS events in the SRM was based on two possible LoRs, in this study a new approach is proposed and is currently being investigated, in which the triples fulfilling a geometrical condition add a new index to the SRM. As a result, the new element ai'j of the SRM represents the probability of detecting an annihilation event emitted from image pixel j by triple i'.

(09:00) M6A2-3, Simulation for CaLIPSO PET Scanner Project

O. Kochebina1,2, S. Jan1, V. Sharyy2, X. Mancardi2, P. Verrecchia2, E. Ramos2, D. Yvon2

1IMIV-SHFJ, CEA, Orsay, France
2SPP-IRFU, CEA, Gif sur Yvette, France
The aim of the CaLIPSO project (French acronym for Liquid Ionization Calorimeter, Scintillation Position Organometallic) is to develop the proof of concept for new gamma detector for Positron Emission Tomography (PET) scanner for small animals and human brain imaging. The objective is to have a spatial resolution about 1 mm3 without losing an efficiency. Moreover, excellent time resolution (~100 ps) important for coincidence selection is also provided. Such performances are possible thanks to the concept of double detection of the signal created by the electron from the photon conversion in trimethyl bismuth, innovative liquid filling a PET cell. This electron emits Cherenkov photons and ionizes the medium. Both, light and free charges, are collected simultaneously and used for the reconstruction of the time and 3D position of the interaction as well as deposited energy. The ongoing work is focused on prototypes developments for light and ionization detection systems. We also develop GATE Monte Carlo simulation to design a full PET scanner and to compare simulation results with performances of other high resolution PET systems. The CaLIPSO is promising ongoing project for a PET scanner with a high potential dedicated to small animal and brain imaging that should outperform the other detector technologies proposed for PET imaging such as scintillating crystals, high-Z semiconductors and liquid Xenon scanners.

(09:15) M6A2-4, Monte Carlo Study for Pinhole X-ray Fluorescence Imaging of Gadolinium Nanoparticles

S. Jung, W. Sung, S.-J. Ye

Seoul National University, Seoul, Korea
This work aims to develop a Monte Carlo (MC) model of pinhole x-ray fluorescence imaging for gadolinium (Gd) nanoparticles in a cylindrical water phantom and to derive the relationship between imaging dose and quality for different x-ray spectra and incident directions. The image quality and doses for the bench-top setting using polychromatic diagnostic x-ray sources were compared with those using synchrotron x-ray sources. We created a MC model for a pinhole x-ray fluorescence imaging system using Monte Carlo N-Particle Version 6.1 (MCNP6.1). We simulated two different sets of Gd concentrations in the three columns: (A) 0.05%, 0.075%, and 0.1% by weight of Gd in order, and (B) 0.01%, 0.025%, and 0.05% by weight of Gd in order. We also assumed two sets of column locations w.r.t. an incident direction of x-rays at the center of the phantom: (a) 0° and ±90° and (b) ±90° and 180°. The water phantoms with and without the Gd columns were irradiated by fan beams of 51 and 55 keV monochromatic photons. Polychromatic x-ray fan beams of 110 kVp extracted from Spektr code and 110 kVp beams with a 0.3 mm tungsten (W) filter were also used to irradiate the phantoms. The images were evaluated using the contrast to noise ratio (CNR). The imaging dose to water of each x-ray source was also calculated by using the MCNP6.1 code. Finally we estimated the imaging dose to the water required to achieve the same level of CNR for different x-ray sources, since CNR is proportional to vN, (N = No.of histories). Comparing the doses required to achieve CNR=5, the filtered 110 kVp could be suitable for bench-top setting of pinhole x-ray fluorescence imaging. The MC simulations demonstrated the feasibility of pinhole x-ray fluorescence imaging for Gd nanoparticles. In addition, we found that CNR of fluorescence images varied with incident directions of x-ray.

(09:30) M6A2-5, Optimization of Parallel-Hole Collimators for Intraoperative Localization of Iodine-125 Seeds

B. Arsenali, K. G. A. Gilhuijs, H. W. A. M. de Jong

University Medical Center Utrecht (UMCU), Utrecht, The Netherlands
Optimization of parallel-hole collimators for intraoperative localization of iodine-125 seeds B. Arsenali, K. G. A. Gilhuijs, and H. W. A. M. de Jong Radioactive seed localization is a well-known two-step method which is used clinically to reduce the number of incomplete excisions. First, an iodine-125 seed is injected in the center of the tumor under mammographic or ultrasonic guidance. Second, this seed is localized with the help of a hand-held gamma probe. It seems likely that the number of incomplete excisions can be reduced if (in addition to intraoperative information about the location of the tumor) preoperative information about the extent of the tumor is transferred to the operating room. To realize this, we are developing a system for computer-assisted resection of non-palpable breast lesions. The purpose of this system is to overlay preoperative information about the extent of the tumor onto an optical image of the patient in the operating room. To achieve this, radioactive seed is localized during surgery from large distances by two gamma camera heads and two parallel-hole collimators. In this study, the impact of different parallel-hole collimators on the localization accuracy (defined as trueness and precision) of the system was investigated using Geant4 simulations. The results from the simulations indicate that high-resolution low-sensitivity collimators are less accurate for the localization of the radioactive seed than low-resolution high-sensitivity collimators if imaging times of 2.5 s and collimator-source distances of 75 cm are used. The collimator which minimized the localization trueness had a sensitivity of 1.54 × 10-4 and resolution of 4.25 cm at a collimator-source distance of 75 cm.

(09:45) M6A2-6, Sensitivity Comparison of the Helmet-chin PET with a Cylindrical PET: a Simulation Study

A. M. Ahmed, H. Tashima, E. Yoshida, F. Nishikido, T. Yamaya

Molecular Imaging center, Biophysics Program, National Institute of Radiological Sciences, Chiba, Japan
Dedicated brain PET scanner becomes an important tool for early diagnosis of Alzheimer's disease and brain function studies. The spatial resolution of the PET system can be improved by use of advanced depth-of-interaction (DOI) detectors, and its sensitivity can be enhanced by increasing its solid angle. Conventionally, PET scanners are designed based on a cylindrical geometry, and the sensitivity is limited due to the small solid-angle coverage. On the other hand, we have proposed a dedicated brain PET scanner based on a hemispheric-shaped and a chin detector (here after referred to as helmet-chin PET). The chin detector helps to increase the number of possible lines-of-response in the hemisphere. In this study, we evaluate the sensitivity of the helmet-chin PET under a realistic scanner design using Geant4 simulation toolkit. The scanner was constructed from 54 four-layer DOI block detectors arranged in a hemisphere (helmet part) of radius 126.5 mm and a chin detector. The helmet part had three rings with different radii and number of block detectors, and a top cover with 5 block detectors arranged in a cross-shaped geometry. Each block detector consisted of 16 x 16 x 4 array of GSO crystals with a dimension of 2.8 x 2.8 x 7.5 mm3. The　sensitivity of the helmet-chin PET was compared to that of a cylindrical PET constructed from the same number and type of block detectors. The point source sensitivity of the helmet-chin PET was 1.4 times higher around the bottom part of the helmet and 4 times higher at the top part of the helmet compared to that of the cylindrical PET. The results showed that the helmet-chin PET can significantly improve sensitivity at the cerebellum and cerebrum regions, and could be promising for early diagnosis of Alzheimer's disease and accurate brain function studies. Recently, we have developed the first helmet-chin PET prototype.