J3D1  NSS/MIC Joint Session

Wednesday, Nov. 4  16:30-18:30  Golden Pacific Ballroom

Session Chair:  David Brasse, Institut Pluridisciplinaire Hubert Curien, France; Charles Watson, Siemens Healthcare Molecular Imaging, United States

Show/Hide All Abstracts

(16:30) J3D1-1, Pico-second Jitter Clock Distribution Based on Clock Recovery in Multi-Gigabit FPGA Transceivers

F. T. Abu-Nimeh, W.-S. Choong

Lawrence Berkeley National Laboratory, Berkeley, USA
We develop a new Digital Detector Board (DDB) module in order to extend the current capabilities of the OpenPET platform by providing a new means to acquire and process data from silicon photomultiplier (SiPM) arrays. One of the main challenges in reading out these SiPMs is the high density of readout channels required in a practical system. As a result, application-specific integrated circuits (ASICs) have been developed to meet this requirement. Typically, an ASIC module processes the analog signals from the SiPMs and produces serialized digital information of the processed signal. In time-of-flight applications, a time-to-digital-converter (TDC) is normally employed to generate a time-stamp of the arrival of the signal. Systems with distributed ASIC modules require all components to be synchronized, which is performed by distributing a clock to the ASICs with a time jitter that is less than the resolution of the TDCs -- typically in the range of tens of picoseconds. Traditionally, this clock is distributed through dedicated clock lines to each ASIC module; thus, increasing the complexity, flexibility, and cost of the entire system. In contrast, OpenPET DDB exploits the mutli-gigabit transceivers found in modern FPGAs to distribute the clock and data/control using a single physical medium (line) and a single processing unit. The clock is distributed by recovering it using the FPGA’s transceivers which are equipped with high-performance Clock Data Recovery (CDR) circuitry. The recovered clock is exported from the transceiver to the FPGA fabric/pins using a dedicated global clock path in order to be used in the rest of the design e.g. (TDC or ASICs). We measured the recovered clock jitter to be 7.595ps when using fiber optic SFPs. Additionally, data from ASICs can be sent at a speed up to 5Gsps and processed on the DDB’s FPGA in parallel. Consequently, this reduces the clock distribution complexity of the entire platform while providing scalability that is only limited by bandwidth of the SFP module chosen and the number of resources on the DDB’s FPGA.

(16:45) J3D1-2, Improvement of Purity of Produced 15O Beams for OpenPET

A. Mohammadi1, E. Yoshida1, H. Tashima1, F. Nishikido1, T. Inaniwa2, A. Kitagawa2, T. Yamaya1

1Molecular Imaging Center, National Institute of Radiological Sciences, Chiba, Chiba, Japan
2Research Center for Charged Particle Therapy, National Institute of Radiological Sciences, Chiba, Chiba, Japan
In ion therapy, it is ideal to use radioactive ion beams for in-beam positron emission tomography (PET) imaging in order to visualize the irradiation field for treatment. In cancer treatment with light ions, 15O beam is a good candidate for tumor treating and monitoring the delivery of the dose by (PET) imaging. Moreover, a combination of different positron emitter ions, such as 11C and 15O, might be favorable for a selected clinical situation in the near future for radiation therapy. We have recently produced 15O beams in the Heavy Ion Medical Accelerator in Chiba (HIMAC) using an optimized target of polyethylene. But the purity of the produced beams was around 75% which is not satisfactory for clinical utilization. Therefore in this study we focused on improvement of the purity of the produced beam by adding a wedge-shaped degrader within the beam line. We inserted an aluminum degrader with the thickness of 1.76 cm into the secondary beam line. The production rate and purity of the produced beam for the polyethylene target with thicknesses of 5 cm and 11 cm were measured. The in-beam PET images of the produced beams with and without the aluminum degrader in the beam line were obtained using our whole-body OpenPET system. The purity of the beam produced by inserting the degrader into the beam line was increased from 75% to 97% although the production rate was reduced more than 30%. This purity of 97% for the beam is high enough for clinical applications.

(17:00) J3D1-3, The 3D Single Photon Counting Module: a Solution to the 10 ps Single Photon Timing Resolution Challenge for Scintillator-Based Instruments

J.-F. Pratte1, F. Nolet1, M.-O. Mercier1, N. Roy1, S. Parent1, L. Maurais1, A. Corbeil Therrien1, M.-A. Tetrault1, F. Dubois1, X. Bernard1, H. Dautet1, R. Lecomte2, S. Charlebois1, R. Fontaine1

1Department of Electrical and Computer Engineering, Universite de Sherbrooke, Sherbrooke, QC, Canada
2Department of Nuclear Medicine and Radiobiology, Universite de Sherbrooke, Sherbrooke, QC, Canada
One of the avenues pursued to improve PET scanner performance aims to reduce the coincidence timing resolution (CTR). A 10 ps CTR would enable preclinical ToF PET and clinical PET with “real-time imaging” since annihilation locations would be readily available. To that end, ongoing research on scintillators seeks to increase the emission of prompt photons originating from Cherenkov radiation, hot intraband luminescence or quantum dots spontaneous emission. With the goal of providing the next generation of PET scanners’ frontend, our group has undertaken the development of 3D single photon counting modules (3DSPCM) achieving ps measurement precision. 3DSPCM consist of an array of single photon avalanche photodiodes (SPAD) individually read out vertically by a matching array of quenching circuits (QC). Processing electronics are also integrated in the 3D ASIC to perform the desired measurements meeting the application requirements. This paper presents a 3DSPCM architecture where each SPAD has a QC and a time-to-digital converter (TDC) with a targeted resolution of 5 ps. The frontend electronics was realized in TSMC 65 nm CMOS. The QC has adjustable threshold for optimum timing resolution. 3 different test SPAD were integrated in the ASIC to measure the QC’s performance in 2D. Using a Ti:Sapphire femtosecond laser, we report a single photon timing resolution of 13 ps FWHM for the SPAD and QC. The TDC is a single stage Vernier ring, where a digital phase locked loop stabilizes the ring oscillators with temperature, process and voltage variations and contributes to even out all the TDC characteristics within the array. For this 1st revision, we measured a resolution of 15 ps with a DNL of 0.34 LSB and an INL of 3 LSB. Recent results on SPAD array optimization in Teledyne-Dalsa HV CMOS will be presented. Finally, each step of the 3D integration microfabrication post-process has been demonstrated and 60 samples are currently (04/2015) being processed.

(17:15) J3D1-4, Evaluation of a Linearly-Graded SiPM (LG-SiPM) Array for Small-Animal PET

J. Du1, A. Gola2, F. Acerbi2, A. Ferri2, C. Piemonte2, S. R. Cherry1

1Biomedical Engineering, University of California, Davis, Davis, CA, USA
2Fondazione Bruno Kessler, Via Sommarive, Trento, Italy
A 2 x 3 array of linearly-graded silicon photomultipliers (LG-SiPMs), a new position-sensitive form of the silicon photomultiplier from FBK (Fondazione Bruno Kessler, Italy), was evaluated for high resolution small-animal PET. Each SiPM in the array has an active area of 5 mm x 7.5 mm, comprising 87,000 20 µm x 20 µm microcells and fabricated using FBK’s red green blue high-density (RGB-HD) SiPM technology. Using a 6 x 6 array of 0.45 mm x 0.45 mm x 6 mm polished LYSO crystals, coupled to the center of one SiPM, we evaluated the detector performance with a focus on applications in very high resolution small-animal PET. Measurements of signal to noise ratio, energy resolution and flood histograms were performed at different over-voltages (from ~1.0 V to 7.0V, in 0.5V intervals) and three different temperatures (0 °C, 10 °C and 20 °C) to find the optimal working conditions. The results showed that all the crystals in the LYSO array can be clearly resolved, even at room temperature (20 °C). The best flood histogram quality was obtained at an over-voltage of ~ 3.5 V and a temperature of 0 °C. Average energy resolution, signal-to-noise ratio and flood histogram quality were 15.3 ± 2.6 %, 404.1 ± 28.5 and 3.51 ± 0.54 respectively, obtained at an over-voltage of 3.5 V and temperature of 0 °C. The flood histogram quality did not get worse at room temperature (20 °C). In summary, the LG-SiPM array appears extremely promising for decoding the small crystal elements needed for very high resolution small animal PET. The next stage will be to multiplex the 30 SiPM signals to 5 (4 position and 1 timing) and evaluate the ability of the LG-SiPM array to encode depth-of-interaction using a dual-ended read out.

(17:30) J3D1-5, Radiation Hardness of dSiPM Sensors in a Proton Therapy Radiation Environment

F. Diblen1,2, T. Buitenhuis1, T. Solf3, P. Rodrigues4, M.-J. van Goethem5, S. Brandenburg1, P. Dendooven1,6

1KVI-Center for Advanced Radiation Technology, University of Groningen, Groningen, The Netherlands
2MEDISIP, Department of Electronics and Information Systems, Ghent University-iMinds Medical IT-IBiTech, Ghent, Belgium
3Philips Digital Photon Counting, Aachen, Germany
4Oncology Solutions, Research, Philips Group Innovation, Eindhoven, The Netherlands
5University of Groningen, University Medical Center Groningen, Dept. of Radiation Oncology, Groningen, The Netherlands
6Helsinki Institute of Physics, Helsinki, Finland
Due to the localized high dose deposit in the Bragg peak, proton beam radiotherapy is quite sensitive to a variety of differences between the actual and planned treatment situation. A method for in-vivo dose delivery verification is thus needed and can be based on fast scintillation detectors for prompt gamma or PET imaging. The digital silicon photomultiplier (dSiPM) allows excellent scintillation detector timing properties in the 300 ps range and is thus being considered for such verification methods. We present here the results of the first investigation of radiation damage to dSiPM sensors in a proton therapy radiation environment. Radiation hardness experiments were performed at the AGOR cyclotron at KVI-Center for Advanced Radiation Technology, University of Groningen together with Philips. A 150 MeV proton beam was fully stopped in a water target. In a first experiment, bare dSiPM tile sensors were placed at 25 cm from the Bragg peak, perpendicular to the beam direction, a geometry typical for an in-situ implementation of a PET or prompt gamma imaging device. In a second experiment dSiPM-based PET detectors with LYSO scintillator crystal arrays were placed at 2 and 4 m from the Bragg peak, perpendicular to the beam direction; resembling an in-room PET implementation. Radiation damage is assessed by the increase in dark count rate (DCR). It was found that neutron damage in the first experiment (in-situ) leads to an excessive increase in sensor dark count rate. For PET detectors in an in-room location (2 - 4 m distance), detector performance was almost unchanged - even after an irradiation equivalent to 3 years of use in a treatment room (3 � 1015 protons). The reason is based on a unique feature of the dSiPM, disabling damaged SPAD cells to effectively mitigate the effect of the neutron damage as long as the damage is affecting less than 5%-10% of the cells. Therefore, extended lifetime of dSiPM detectors in a proton therapy radiation environment is expected.

(17:45) J3D1-6, The Strip Silicon Photo-Multiplier: an Innovation for Enhanced Time and Position Measurement

M. C. S. Williams1,2, K. Doroud2, K. Yamamoto3

1INFN, Bologna, Italy
2PH, CERN, Geneva, Switzerland
3Solid State Division, Hamamatsu Photonics K.K., Hamamatsu, Japan
There is considerable R&D to enhance the time resolution of various particle detectors, and in particular for the Silicon PhotoMultiplier. Here, we will present a new geometry for the SiPM in the form of a strip. A strip can be read out at each end, with each end coupled to an individual TDC (time to digital converter). The time difference is related to the position of the firing SPAD along the length of the strip, while the average of the two times gives the time of the hit. We will present and discuss the results from the testing of the first prototype of a strip MPPC array that consists of five strips.

(18:00) J3D1-7, 100ps Coincidence Time Resolution with LYSO Coupled to NUV-HD SiPMs

A. Ferri1, F. Acerbi1, A. Gola1, G. Paternoster1, G. Zappalà2, N. Zorzi1, C. Piemonte1

1Fondazione Bruno Kessler, Trento, Italy
2Department of Physics, University of Trento, Trento, Italy
The coincidence timing resolution (CRT) is one of the key parameter to improve the image quality in the next generation of PET scanners. The most used scintillator is the L(Y)SO that combines high density, fast light emission and absence of hygroscopicity. On the detector side, the best candidate is represented by the SiPM technology, which is evolving rapidly to exploit the performances of the scintillator materials. In order to optimize the SiPM technology for the readout of LYSO scintillators, we developed the NUV-HD technology that combines the High Density layout and fabrication process (including trenches), with the NUV p-on-n junction type, more suited for the detection of blue light. We fabricated devices having 25x25 µm2 micro-cells (with a fill factor of 73%) and two active area dimensions: 1x1 mm2 and 4x4 mm2. We performed the electro-optical characterization on the smaller devices and tested the timing and energy resolutions of the larger when coupled to LYSO scintillators. The primary dark count rate is below 200 kHz/mm2 at 9 V over-voltage and 20 °C, which is similar to the NUV technology without trenches. At the same bias, the excess noise factor is below 1.5, and the PDE reaches a peak of 53% at 400 nm. For the timing resolution measurements we tested different scintillator dimensions (3x3x5 mm3 and 3.8x3.8x22 mm3) and temperatures: +20, 0 and –20 °C. The low correlated noise allowed to bias the SiPMs at high over-voltages (above 12 V), even when coupled to the scintillators. The best measured CRTs are: 100 ± 2 ps FWHM and 146 ± 3 ps FWHM for the 5 mm and 22 mm crystal, respectively. No significant dependence on the temperature was observed. To our knowledge, for the first time the NUV-HD allows to reach, with commercial LYSO, timing resolutions that were only available with LaBr3 or calcium co-doped LSO.

(18:15) J3D1-8, Scintillator Performance Enhancement Through Imprinted Nanostructures

B. Singh1, A. R. Knapitsch2, M. S. J. Marshall1, J.-G. Kim3, S. Li3, G. Barbastathis3, P. Lecoq2, V. V. Nagarkar1

1Radiation Monitoring Devices, Inc., Watertown, MA, USA
2CERN, Geneva, Switzerland
3Massachusetts Institute of Technology, Cambridge, MA, USA
The ability to extract scintillation light from a crystal with efficiency and speed has great potential to realize the much needed gains in the performance of numerous radiation detection and imaging instruments. Such high performance spectrometers/imaging instruments are vital in medical imaging (e.g., PET), industrial, and homeland security applications for the detection, localization, and energy classification of X-rays, gamma-rays or neutrons. We report on the fabrication of nanostructures imprinted in custom-developed, high refractive-index polymers for improving the performance of scintillators. Nanostructures such as photonic crystals permit effective utilization of the scintillation light which is otherwise lost when a high refractive index scintillator is coupled to a photodetector with low refractive index window, thereby substantially improving the light extraction efficiency, and energy and timing resolution. We have fabricated these nanostructures using low-cost, scalable, nanoimprinting technique in novel high refractive index polymers. Silicon master templates have been fabricated for large area imprinting of these nanostructures, which are currently optimized for LYSO emission wavelength. The templates range in size from 7 mm × 7 mm to 20 mm × 15 mm. The nanoimprinted photonic crystals have demonstrated excellent reproducibility, and have improved the light collection efficiency from LYSO by 18% compared to the current state-of-the-art grease coupling. Refinements are currently underway to further improve the performance of the scintillators through the use of these periodic structures. Details of the design, fabrication and characterization of the nanostructures/photonic crystals, and their impact on scintillator performance will be presented.