M25  High Resolution and Pre-Clinical Imaging Instrumentation

Saturday, Nov. 2  10:30-12:30  GBR 103

Session Chair:  Yong Choi, Sogang University, South Korea; David Brasse, CNRS - IPHC, France

Show/Hide All Abstracts

(10:30) M25-1, Dual-Resolution MicroSPECT Mouse Imaging Using a Triple-Head SPECT System

S. C. Moore1, M.-A. Park1, D. Xia2, S. D. Metzler2

1Radiology/Nuclear Medicine, Brigham & Women's Hospital and Harvard Medical School, Boston, MA, USA
2Radiology, University of Pennsylvania, Philadelphia, PA, USA

INTRODUCTION: One microSPECT design approach providing both high resolution and high sensitivity is based on imaging a mouse inside a tube with many pinholes; projection images from a small target region, e.g., the heart or brain, can be recorded simultaneously on multiple detectors with little multiplexing since each pinhole aperture's opening angle is restricted to view the target organ; however, this often requires that much of the mouse body be scanned through the target region of the collimating tube to obtain complete data for reconstruction. METHODS: We describe another approach for joint reconstruction of both low- and high-resolution projection data acquired sequentially through many rectangular pinholes embedded in two tungsten tubes of different diameter, placed end-to-end along the axis of a 3-head clinical SPECT scanner. One tube provides a whole-body scout image of good sensitivity, but limited resolution, for locating the organ of interest and acquiring complete data for reconstruction. A second, 'hi-res' tube, which contains many smaller pinholes, is then moved into place axially over the mouse, which is also moved to position its target organ in the center of the gantry. We optimized a scout tube for a Trionix 3-head system to provide minimum mean-squared error of reconstructed voxel counts throughout a ~6-cm axial range, with the constraints of fixed 2.4-mm scout system resolution, limited multiplexing, and no detector motion. RESULTS: The optimal radial distance to the closest scout pinhole and optimal number of MLEM iterations were 4.4 cm and 60 iterations, respectively; radial distances of the 30 pinholes ranged from 4.4 to 4.8 cm. CONCLUSION: After including whole-body mouse reconstructions of simulated 30-pinhole scout data into reconstructions of data from 93, 0.33-mm pinholes in a hi-res tube viewing the heart region, we obtained high-resolution images of the heart, embedded in lower resolution images of the body, with minimal artifacts.

(10:45) M25-2, Model-Based Normalization of a Fractional-Crystal Collimator Prototype for Small-Animal PET Imaging

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

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

Previously, we proposed to use a coincidence-collimator to determine four lines of response (LOR) within a crystal pair and achieve fractional-crystal resolution in PET imaging. We have designed and fabricated a collimator prototype for a small-animal PET, A-PET scanner. To compensate for imperfections in the fabricated collimator prototype, collimator as well as crystal normalization is required to reconstruct quantitative and artifact-free images. In this study, we developed a normalization method for the collimator prototype based on the A-PET normalization using a uniform cylinder phantom with diameter of $10$~cm (and length of $15$~cm). We performed data acquisitions without collimator for crystal normalization first, and then with the collimator from $8$ different rotation views for collimator normalization. After a reconstruction without correction, we extracted the cylinder parameters from which we generated expected emission sinograms with attenuation correction. Single scatter simulation was used to generate the scattered sinograms. The least-squares method was used to generate normalization coefficient for each LOR based on measured, expected and scattered sinograms. The crystal and collimator normalization coefficients are factorized by performing two normalizations separately. We also developed a model-base collimator normalization that can significantly reduce variance and produce collimator normalization with adequate statistical quality at feasible scan time. The normalization methods were verified using the experimental data.

(11:00) M25-3, Characterization of a Trapezoidal Slat Crystal PET Detector

R. S. Miyaoka, A. L. Lehnert, W. C. Hunter

Radiology, University of Washington, Seattle, WA, USA

Objectives: We previously reported on the performance characteristics of a trapezoidal slat crystal (TSC) PET detector. In this work, we report on a new TSC detector with a longer axial extent. In addition the new detector uses multi-pixel photon counter (MPPC) arrays with better photon detection efficiency and smaller pixel elements than the Geiger Mueller avalanche photodiode (GM-APD) arrays used in the initial design. Goals of the TSC detector are to provide sub-millimeter intrinsic spatial resolution, to provide depth of interaction (DOI) positioning, and to support high packing fraction for a compact PET detector system. DOI positioning is achieved by including a model of the depth dependent light response function along the long axis of the crystal. Methods: The TSC detector is comprised of six LYSO trapezoidal, slat crystals each 8 mm tall, 40.0 mm long, and 0.70 mm on the entrance surface and 0.96 mm wide on the exit surface. Mirror film of varying lengths is used to control light sharing within the crystal array. The TSC array is coupled to a 2-by-12 array of MPPC with 3-mm-square sensitive area and with 3.2 mm center-to-center spacing. A statistics-based method was used for event positioning. Data were collected to calibrate the detector and to test its intrinsic spatial resolution and energy resolution performance. Results: Crystal slat decoding of the detector was excellent. The average peak to valley ratio of the crystal map for slat decoding was 6.8. The average intrinsic spatial resolution along the long axis of the crystal was ~1.0 mm FWHM after correcting for source size. The average energy resolution for the detector was 11.3 3.1 % at 511 keV. Conclusions: A new trapezoidal slat crystal, PET detector has been designed and evaluated. The improvement in intrinsic spatial resolution and energy resolution was as expected given the better PDE and more compact photosensor geometry of the new GM-APD array.

(11:15) M25-4, First Measurements of a 512 PSAPD Prototype of a Sub-MM Resolution Clinical PET Camera

A. Vandenbroucke1, P. D. Reynolds2, F. W. Lau2, D. Innes2, D. L. Freese2, D. F. Hsu2, C. S. Levin1,2,3,4

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

We present measurements from a first prototype of a 1 mm3 resolution clinical PET camera dedicated to breast imaging. This first prototype consists of two cartridges of 8 detector layers. Each detector layer comprises 16 dual LYSO-PSAPD modules arranged side by side. The LYSO scintillator is an 8x8 array of 0.9x0.9x1 mm3 crystal pixels. Each array is coupled to a PSAPD. The modules are oriented edge-on with respect to the incoming 511 keV photons. The two cartridges thus contain 256 PSAPDs and 16284 crystals each. The entire camera will have 2 imaging heads of 9 cartridges each. Charge created by the PSAPDs was routed towards a signal conditioning board that was connected to another circuit board that houses the RENA-3 ASIC. The RENA3 chip contains 36 channels each with preamplifier, leading edge discriminator and sample and hold circuitry. The RENA3-boards were connected to a DAQ board that linked to a host PC using 8 USB2 connections. We measured a two-cartridge system energy resolution of 10.62+/-0.04 % FWHM at 511 keV, together with a coincidence time resolution of 12.2+/-0.2 ns FWHM. We determined a crystal misidentification probability of < 0.5% on average. For the intrinsic spatial resolution we measured an average point-spread-function of 0.84+/-0.02 mm, not corrected for finite source width ( 0.25 mm diameter ) , positron range (22Na), or photon acolinearity. We also present a reconstructed Derenzo phantom, obtained with 34 μCi/ml FDG. Although we only used a small subset of the final system and thus had a relatively low count rate, the 1.6 mm spheres are distinctly visible. Randoms fraction was estimated to be 20% of total count rate.

(11:30) M25-5, Continuous Depth-of-Interaction Encoding PET Detector Using Digital Silicon Photomultiplier

M. S. Lee, J. S. Lee

Department of Nuclear Medicine, Seoul National University College of Medicine, Seoul, Republic of Korea

For high resolution and high sensitivity PET system, PET detectors with depth-of-interaction (DOI) encoding capability are required. To measure DOI information from a mono-layer array of scintillation crystals using only a single-ended readout, our group has previously proposed a new method based on light spreading within a crystal array [1]. Our group also has developed a prototype of DOI PET detector using photomultiplier tube (PMT). Consequently, our proposed method showed good DOI resolution which promises for high resolution and sensitivity PET system [2]. In this study, we used Philips digital silicon photomultiplier (dSiPM), since this sensor records its energy and time value digitally, so that this kind of feature is expected to have advantage in developing our DOI PET detector. Here, we present first experimental evaluation of novel DOI PET detector composed with 2x2x20 mm3 pixelated LGSO crystal array and dSiPM (DPC-3200-22-44). Data acquisition process was handled by Philips Digital Photon Counting Technology Evaluation Kit (PDPC-TEK). Moreover, maximum-likelihood estimation was used for decoding DOI information. As a result of DOI PET detector using dSiPM array, each pixelated crystals was well resolved. At the center crystal, the energy resolution averaged over five different depth of crystal was 17.68 %. Furthermore, the DOI accuracy regarding 1 DOI position as correctly estimated was 83.21 % in average.

(11:45) M25-6, Light-Sharing Interface for dMiCE Detectors Using Sub-Surface Laser Engraving

W. C. J. Hunter1, R. S. Miyaoka1, L. R. MacDonald1, W. McDougald1, T. K. Lewellen1,2

1Radiology, University of Washington, Seattle, WA, USA
2Electrical Engineering, University of Washington, Seattle, WA, USA

We have previously reported on dMiCE, a method of resolving depth or interaction (DOI) in a pair of discrete crystals by encoding light sharing properties as a function of depth in the interface of this crystal-element pair. A challenge for this method is the cost and repeatability of interface treatment for a crystal pair. In this work, we report our preliminary results on using sub-surface laser engraving (SSLE) as a means of forming this depth-dependent interface in a dMiCE detector. A surplus first-generation SSLE system was used to create a partially reflective layer 100-microns thick at the boundary between two halves of a 1.4-by-2.9-by-20 mm^3 LYSO crystal. The boundary of these paired crystal elements was positioned between two 3-mm wide Geiger-Mller avalanche photodiodes from Hamamatsu. The responses of these two photodetectors were acquired for an ensemble of 511-keV photons collimated to interact at a fixed depth in just one crystal element. Interaction position was then varied to measure detector response as a function of depth, which was then used to maximum-likelihood positions events. Despite use of sub-optimal SSLE processing we found an average DOI resolution of 3.4 mm for front-sided readout and 3.9 mm for back-sided readout. We expect DOI resolution can be improved significantly by optimizing the SSLE process and pattern.

(12:00) M25-7, Enhancing Spatial Resolution of Timepix Positron Camera Based on Classification of Primary Interactions using SVM

Q. Wang, K. Shi, Z. Liu, S. Ziegler

Nuclear Medicine, Technische Universitaet Muenchen, Muenchen, Germany

Position-sensitive positron camera may be advantageous for some preclinical and intraoperative clinical applications. However, the resolution of a positron camera is limited by the scattered interactions within the detectors. It is still impossible to detect the primary interaction along a particle track by hardware. Here, we proposed a novel data-driven method to improve spatial resolution (SR) based on classification of primary interactions of positrons within the detector using support vector machine (SVM). The method is tested and assessed on 18F-FDG imaging of an absorbing edge protocol. A classification model is constructed by learning the interactions of a positron track based on the data derived from Monte-Carlo simulation using Geant4. Four features about the topology and relative energy of interactions generated by 18F positrons were considered for the training. The measurements were acquired with a Timepix positron camera operated under time-over-threshold mode. After applying the classification model on the measurements, the primary interaction of the positron tracks in the silicon detector were estimated. The spatial resolution is thus improved from 181.5 Ám and 188.2 Ám (centroid method with and without energy weighting) to 126.9 Ám and 149.3 Ám (our method with and without energy weighting on primary interaction). The positive preliminary results support the further investigation of the proposed algorithm, which may provide the potential to promote the applications of positron camera in clinical and preclinical imaging.

(12:15) M25-8, Performance Evaluation of the PET Component of Novel Preclinical PET/CT Scanner Using NEMA NU-4 2008 Standard

N. Belcari1,2, N. Camarlinghi1,2, M. Cecchetti1, S. Ferretti1,2, M. Hohberg1, D. Panetta3, P. Salvadori3, G. Sportelli1,2, K. Straub2, A. Del Guerra1,2

1Department of Physics, University of Pisa, Pisa, Italy
2Sezione di Pisa, INFN, Pisa, Italy
3Institute of Clinical Physiology, CNR, Pisa, Italy

A novel preclinical PET/CT system for mice and rats has been developed featuring state of the art technology. The scanner comprises a full ring PET and a high resolution CT system placed sequentially like in clinical PET/CT scanners. The scanner has a high sensitivity of 9% at CFOV and a high spatial resolution. The optimized PET detector module design allows to reach and excellent energy resolution of about 12%. This article presents the performance assessment study of the PET component of the system following the National Electrical Manufacturers Association (NEMA) NU-4 2008 standard. The PET component of the scanner consists of 16 modular detectors arranged in two octagonal rings. The field-of-view has 95 mm axial coverage and a diameter of 80 mm. Each module comprises a LYSO:Ce matrix of 702 crystals of 1.6 mm x 1.6 mm x 12 mm with a pitch of about 1.7 mm directly coupled to a 64 anodes PMT (Hamamatsu H8500). Spatial resolution, sensitivity, counting rate capabilities, and image quality were evaluated in accordance with the NEMA NU-4 standard. Energy and timing resolution were also evaluated. The PET system also features the unique possibility to perform rotational acquisitions where data are acquired at several angular positions while the PET ring spins around the object (step-and-shoot mode) similarly to the four head PET configuration we have previously developed for the YAP-(S)PET scanner. Even if not strictly necessary in this case due to the full angular coverage of the detectors, the results obtained in rotational mode show a significant reduction in the image noise in comparison to the non-rotational thus making this acquisition mode particularly well suited for those cases where image quality is more important than the tracer dynamic information.