Please note! All times in the online program are given in New York - America (GMT -04:00) times!

New York - America ()
Jan 29, 2022, 7:53:12 AM
Your time ()
Tokyo - Asia ()
Jan 29, 2022, 9:53:12 PM
Our exhibitors and sponsors – click on name to visit booth:

To search for a specific ID please enter the hash sign followed by the ID number (e.g. #123).

Simulation and Modeling of Medical Imaging Systems

Session chair: Roncali , Emilie (University of California, Davis, Department of Biomedical Engineering, Davis, USA); Tashima , Hideaki (National Institutes for Quantum and Radiological Science and Technology, Department of Advanced Nuclear Medicine Sciences, Chiba, Japan)
Shortcut: M-02
Date: Wednesday, 20 October, 2021, 7:00 AM - 8:45 AM
Room: MIC - 2
Session type: MIC Session


Click on an contribution to preview the abstract content.

7:00 AM M-02-01

A collimatorless CZT detector using Compton and proximity imaging for in vivo 225Ac tomography: a feasibility study (#87)

J. Caravaca Rodriguez1, Y. Zheng2, Y. Huh1, G. T. Gullberg1, Y. Seo1

1 UCSF, Radiology & Biomedical Imaging, San Francisco, California, United States of America
2 UC Berkeley, Nuclear Engineering, Berkeley, California, United States of America


We evaluated the performance of a novel collimatorless detector and show that it can reach the demanding sensitivities required to image in vivo biodistributions of 225Ac, a potent alpha-emitting radionuclide, in mice. To achieve a high-sensitivity detection of the emitted high energy gamma rays (218keV from 221Fr and 440keV from 213Bi), we propose to use a cadmium zinc telluride gamma camera that exploits Compton and proximity imaging. A two-parallel planes design enables both approaches at the same time: Compton imaging with a scatterer-absorber configuration, as well as proximity imaging by having the high-Z scatterer placed very close to the subject. Our Geant4 Monte Carlo simulations, including a complex mouse anatomy and a realistic detector model, show outstanding sensitivities of ~30% and ~5% for 218keV (221Fr) and 440keV (213Bi), respectively, for proximity imaging, and of 1% and 0.5%, respectively, for Compton imaging. To reconstruct an image, we consider a standard Maximum Likelihood Expectation Maximization for proximity imaging and a list-mode Ordered Subsets Expectation Maximization for Compton imaging, which yield spatial resolutions of ~20mm FWHM and ~3mm FWHM, respectively. This is more than 2 orders of magnitude higher sensitivity than that of state-of-the-art SPECTs (typically below 0.01% for energies above 200keV), and a comparable spatial resolution. We show images of 225Ac biodistributions for each independent radionuclide (221Fr and 213Bi) in a simulated mice phantom for typical doses and organ activities, and also discuss the detector design and the reconstruction algorithms that leverage our collimatorless approach.

Keywords: SPECT, theranostics, actinium-225, targeted alpha therapy, ultra-high sensitivity
7:15 AM M-02-02

Simulation Study of GAGG-based Scatterer Detectors for High Resolution Compton Imaging in WGI (#291)

S. Takyu1, H. Tashima1, T. Nishina1, 2, T. Yamaya1

1 National Institutes for Quantum and Radiological Science and Technology (QST), Institute for Quantum Medical Sciences, Chiba, Japan
2 Chiba University, Chiba, Japan


Whole gamma imaging (WGI) is a new concept that combines PET and Compton imaging (camera). Inserting a scatterer detector ring into a PET ring enables various kinds of gamma-ray imaging with one system. We developed a WGI prototype based on scintillation detectors and demonstrated three-dimensional tomographic Compton imaging of the 909 keV gamma rays emitted from 89Zr-injected mouse. However, improving position resolution and energy resolution of the scatterer detector were required for high resolution Compton imaging. Therefore, in this study, we optimized scatterer detector configuration (crystal pitch and energy resolution) by Geant4 simulation. Scatterer detector rings with different GAGG crystal sizes (0.9 × 0.9 × 6 mm3, 1.45 × 1.45 × 6 mm3 and 2 × 2 × 6 mm3) were modeled. The absorber detector ring, which was composed of 4-layer DOI GSO detectors with the crystal size of 2.9 × 2.9 × 7.5 mm3, was modeled according to the prototype. Energy resolution assumed in the simulation was derived from experiment data, and the energy resolution was virtually halved and 1.5 times degraded for comparison. The list-mode OSEM algorithm was used for Compton image reconstruction. In the point source simulation, 1.7, 1.9 and 2.3 mm spatial resolutions (FWHM) were obtained at the center in 0.9, 1.45 and 2 mm pitch GAGG scatterers, respectively. Spatial resolution of the 0.9 mm pitch GAGG scatterer was improved to 1.4 mm by halved energy resolution and deteriorated to 2.1 mm by 1.5 times degraded energy resolution. The spatial resolution of the radial direction was degraded at the peripheral region because of the parallax error, while that of the tangential direction was improved due to the reduced angular uncertainty by getting close to the detector. In the hot-rod-like phantom simulation, 2 mm diameter rods were separated in the 0.9 mm pitch GAGG scatterer. The separation became clearer by halved energy resolution.

Keywords: Whole gamma imaging, Compton camera, scatterer detector, spatial resolution, angular resolution measure
7:30 AM M-02-03

Effect on the Spatial Resolution of PET Event Recovery from Compton Events in C-shaped Compton-PET (#1205)

T. Nishina1, 2, H. Tashima2, S. Takyu2, F. Nishikido2, M. Suga1, 2, T. Yamaya2

1 Chiba University, Chiba, Japan
2 National Institutes for Quantum and Radiological Science and Technology (QST), Institute for Quantum Medical Sciences, Chiba, Germany


To meet the demand for C-shaped PET, we have proposed C-shaped Compton-PET, in which projection data truncation of C-shaped PET is compensated for by applying the Compton camera technique. The C-shaped Compton-PET can measure both PET and Compton events by making the C-shaped PET detectors work as an absorber for a scatterer inserted at the location opposing the open space. Image reconstruction combining PET and Compton events reduces truncation artifacts. This study focused on Compton events that can be treated as PET events because PET event position estimation accuracy is superior to that of Compton events. To assess the effect of an event selection method to extract such Compton events as PET events, we conducted an imaging simulation using the Geant4 Monte Carlo toolkit. The spatial resolution was evaluated with point sources on a cylindrical background. The results showed the PET event extraction method could slightly improve the spatial resolution of the C-shaped Compton-PET.

Keywords: PET, Compton camera, Event selection, Monte Carlo simulation, C-shaped Compton-PET
7:45 AM M-02-04

Imaging Performance of AdaptiSPECT-C for 99mTc/123I Single- and Dual-Isotope imaging (#1081)

B. Auer1, K. S. Kalluri1, C. Lindsay1, J. De Beenhouwer2, R. G. Richards3, M. May3, M. A. Kupinski3, P. H. Kuo3, L. R. Furenlid3, M. A. King1

1 University of Massachusetts Medical School, Department of Radiology, Worcester, Massachusetts, United States of America
2 University of Antwerp, imec- VisionLab, Antwerp, Belgium
3 University of Arizona, Dept. of Med. Imaging & Wyant College of Optical Sciences, Tucson, Arizona, United States of America


We have designed for cerebral imaging a next-generation multi-pinhole system, AdaptiSPECT-C. This prototype incorporates 24 detector modules and can provide a total of up to 120 pinholes in multiple combinations around the brain. The system can adapt each aperture to one of three sizes and whether it is open or closed via a shutter mechanism. AdaptiSPECT-C provides high-performance patient-personalized imaging in various brain imaging procedures. In this work we initially determined the pinhole size and combination of AdaptiSPECT-C for optimal single- and dual-isotope 99mTc-HMPAO perfusion/123I-Ioflupane DaT imaging. We found through quantification and visual inspection that the medium aperture size in a central plus obliques pinhole combination for the adaptable AdaptiSPECT-C system is the most suited for high performance single- and dual-isotope perfusion and DaT imaging. We observed that the use of the largest pinhole size degrades visually and in terms of NRMSE and %SBR, the reconstructions despite higher sensitivity due to lower spatial resolution compared to that of the medium aperture diameter. For simultaneous imaging, we investigated an extreme case for which the two symmetric photopeak windows slightly overlap with each other and the crosstalk contamination (~20% and ~10% for the 99mTc and 123I photopeaks) was not corrected. Our initial results suggest that the reconstructions obtained from the dual-isotope study remain close quantitively and visually to that of the single-isotope simulations. We observed that the obliques or obliques plus central combination is more affected by crosstalk contamination than the central-only configuration.


Research reported in this publication was supported by the NIBIB/NIH under award number R01 EB022521. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Keywords: 99mTc-HMPAO/123I-Ioflupane SPECT dual-isotope imaging, cerebral blood-flow perfusion, DaT imaging, GATE Monte-Carlo simulation, AdaptiSPECT-C
8:00 AM M-02-05

In silico comparison of additive and subtractive charge sharing correction algorithms at medically relevant fluxes in pixelated x-ray photon counting multispectral detectors. (#809)

O. L. P. Pickford Scienti1, 2, D. G. Darambara1, 2

1 Institute of Cancer Research, Radiotherapy and Imaging, London, United Kingdom
2 Royal Marsden NHS Trust, Multi-modality Molecular Imaging team, London, United Kingdom


X-ray photon counting spectral imaging (x-CSI) is a technique of great interest as it could allow x-rays to be used for truly molecular imaging provided good energy resolution and detection efficiencies can be achieved. Currently the limiting factor in fully developing x-CSI detectors for medically relevant x-ray fluxes is the counting speed of the electronics and for a given semiconductor detector there is a tradeoff between spectral resolution and operational fluxes. Charge sharing correction algorithms (CSCA) can be implemented to prevent spectral deterioration at smaller pixel sizes. In this work we investigate the effect that correction mechanism (CM) has on CSCA performance as a function of x-ray flux by modelling the response of 143 different systems, varying in their pixel pitch (100-600 µm) and CM scheme (additive vs subtractive) when exposed to a range of medically relevant x-ray fluxes (~106 - ~109 photons mm-2 s-1). The work was performed using our in-house simulation suite, which combines Monte Carlo and finite element methods to simulate the chain of events in x-CSI imaging as well as MATLAB custom codes to model the processes of signal generation, CSCA application and energy thresholding. 4 different metrics were applied to assess each detector-flux configuration: absolute detection efficiency, absolute photopeak bin efficiency, relative coincidence counts and binned spectral efficiency. Results demonstrate that additive CSCAs drop in spectral performance much more rapidly than their subtractive counterparts with increasing x-ray flux due to their increased susceptibility to pulse pileup. Attractive CSCAs have other advantages, such as higher overall count rates, which can result in lower patient doses being needed. A detailed discussion regarding how additive and subtractive CSCAs differ in performance with respect to the 4 metrics, and the impact this has in informing x-CSI detector design for medical applications, will be presented.

Keywords: Charge Sharing Correction, Finite element analysis, Monte Carlo Methods, Photon counting, x-ray detector
8:15 AM M-02-06

Advantages of Mixed Coincidences between a Breast PET Insert and the Siemens Biograph mMR (#1106)

C. Pommranz1, 2, F. Schmidt1, J. Mannheim1, S. Diebold2, C. Tenzer2, A. Santangelo2, B. Pichler1

1 Eberhard Karls University Tuebingen, Werner Siemens Imaging Center, Tuebingen, Baden-Württemberg, Germany
2 Eberhard Karls University Tuebingen, Institute for Astronomy and Astrophysics, Tuebingen, Baden-Württemberg, Germany


Operating a breast PET insert inside a whole-body scanner with PET modality can in principle enable the acquisition of mixed coincidences between both devices in addition to the common insert only and whole-body only coincidences. Due to the typically smaller crystal sizes of PET inserts compared to whole-body PET systems, the mixed coincidences can improve the spatial resolution outside the field of view of the insert, what is referred to as the virtual pinhole effect. In this work, simulation studies were performed, which show great potential of utilizing mixed coincidences in terms of sensitivity and spatial resolution for image reconstruction outside the insert field of view.
Three different breast PET insert geometries were previously proposed for integration into an existing MR breast coil (Breast Biopsy Coil, NORAS MRI products), to be used inside a 3T PET/MRI scanner (Siemens Biograph mMR, Siemens Healthineers). The detectors are made of 16 x 16 Lutetium Oxyorthosilicate (LSO) crystals with dimensions of 1.51 x 1.51 x 10 mm^3.
Single event list mode data from both PET scanners were acquired from a multi system Monte Carlo simulation. Mixed coincidences were retrieved from the single events using the single event timestamps, which were derived from a global virtual clock at simulation time.
The sensitivity for all three types of coincidences was determined using a point-like 22-Na gamma source with an activity of 100 kBq. With absolute mixed sensitivities between 1.6 % and 3.3 % for the investigated source positions, mixed coincidences show an important contribution to the whole-body only sensitivity between 5.1 % and 10.3 %.
The spatial resolution was determined by simulating a point source placed in a warm background and using iterative reconstruction methods. The radial and axial spatial resolution using only mixed coincidences show an improvement of up to 1.9 mm and 1.1 mm, respectively, compared to the Biograph mMR.

Keywords: PET, Monte Carlo Simulations, PET Image Reconstruction, Mixed Coincidences, GATE
8:30 AM M-02-07

Modeling the GE Discovery MI scanner in GATE: optimization of the digitizer (#1303)

A. Merlet1, J. Salvadori2, A. Cochet1, 3, B. Presles1, J. - M. Vrigneaud1, 3

1 University of Burgundy, ImViA EA 7535, Dijon, France
2 Institut de cancérologie Strasbourg Europe (ICANS), Strasbourg, France
3 Department of Nuclear Medicine, Georges-François Leclerc Cancer Center, Dijon, France


We present the preliminary results of the validation of the General Electric Discovery MI PET scanner model in GATE. The validation was performed using the latest NEMA guidelines by comparing GATE-simulated tests with the experimental ones. The 3D geometry of the system was implemented according to the manufacturer specifications. The phantoms were accurately modeled and the patient bed was taken into account. The digitizer module in GATE aim at modeling the electronic treatment of photonic hits in the detectors without explicitly describing all the mechanisms involved in the detection process. Consequently, a dedicated method was used to match the simulated counting rates to the ones observed. Good agreement was obtained between simulations and experiments for single events detection, randoms rate estimation and prompt coincidences. Simulated sensitivity was 12.98 cps/kBq versus 12.81 cps/kBq for the measured one. No significant differences were distinguishable on axial sensitivity profiles. A complete validation of the system according to NEMA guidelines is on-going.

Keywords: GATE, PET, Monte Carlo simulation, NEMA

Our exhibitors and sponsors – click on name to visit booth: