D. Innes

Effects of external shielding on the performance of a 1 mm 3 resolution breast PET camera

We are constructing a 1 mm3 resolution, high sensitivity breast-specific PET camera, consisting out of 2 imaging heads each built out of many layers of 8×8 arrays of 1×1×1 mm3 LYSO crystals coupled to Position Sensitive Avalanche Photo-Diodes (PSAPDs). This paper investigates the effects of shielding on out-of-field background event rate from highly vascular organs like heart, liver, spleen and brain. In order to optimize the performance of the breast specific PET camera, we must address this background by proper external shielding, which reduces the single photon event rate. This avoids data acquisition saturation and reduces the amount of scatter and random coincidences, which enhances the image SNR. Shielding should ideally reduce out-of-field activity while maintaining a high sensitivity for photons emitted from the breast. Shielding design is studied using the Monte-Carlo simulation package GRAY, which supports complex mesh based primitives for phantoms and detector shapes. An anatomically accurate model of the female torso based on the realistic NURBS CArdiac Torso (NCAT) phantom is used, which was manipulated to include a slight breast compression to a width of 7.5 cm. In these simulation studies we exploit the unique feature of GRAY allowing a direct import of the cameras CAD drawings made in Solid Works. As shielding material we use an alloy of 97% tungsten, 2.1% Ni and 0.9% Fe, which has a density of 18.5 g/cm3 and an high atomic number Z. Different shielding thicknesses and locations are investigated and their influence on the system count rate is analyzed. We show that the count rate is reduced by 45% (57 %) when placing 2.54 mm (5.08 mm) of tungsten shielding around the panels. The highest observed event rate per ASIC is 19 kHz (15 kHz), which is well below the 83 kHz maximum estimated count rate of our data acquisition system. The shielding only reduces the counts coming from the breast by 8% (11%).

Thermal regulation of tightly packed solid‐state photodetectors in a 1 mm3 resolution clinical PET system

PURPOSE Silicon photodetectors are of significant interest for use in positron emission tomography (PET) systems due to their compact size, insensitivity to magnetic fields, and high quantum efficiency. However, one of their main disadvantages is fluctuations in temperature cause strong shifts in gain of the devices. PET system designs with high photodetector density suffer both increased thermal density and constrained options for thermally regulating the devices. This paper proposes a method of thermally regulating densely packed silicon photodetectors in the context of a 1 mm(3) resolution, high-sensitivity PET camera dedicated to breast imaging. METHODS The PET camera under construction consists of 2304 units, each containing two 8 × 8 arrays of 1 mm(3) LYSO crystals coupled to two position sensitive avalanche photodiodes (PSAPD). A subsection of the proposed camera with 512 PSAPDs has been constructed. The proposed thermal regulation design uses water-cooled heat sinks, thermoelectric elements, and thermistors to measure and regulate the temperature of the PSAPDs in a novel manner. Active cooling elements, placed at the edge of the detector stack due to limited access, are controlled based on collective leakage current and temperature measurements in order to keep all the PSAPDs at a consistent temperature. This thermal regulation design is characterized for the temperature profile across the camera and for the time required for cooling changes to propagate across the camera. These properties guide the implementation of a software-based, cascaded proportional-integral-derivative control loop that controls the current through the Peltier elements by monitoring thermistor temperature and leakage current. The stability of leakage current, temperature within the system using this control loop is tested over a period of 14 h. The energy resolution is then measured over a period of 8.66 h. Finally, the consistency of PSAPD gain between independent operations of the camera over 10 days is tested. RESULTS The PET camera maintains a temperature of 18.00 ± 0.05 °C over the course of 12 h while the ambient temperature varied 0.61 °C, from 22.83 to 23.44 °C. The 511 keV photopeak energy resolution over a period of 8.66 h is measured to be 11.3% FWHM with a maximum photopeak fluctuation of 4 keV. Between measurements of PSAPD gain separated by at least 2 day, the maximum photopeak shift was 6 keV. CONCLUSIONS The proposed thermal regulation scheme for tightly packed silicon photodetectors provides for stable operation of the constructed subsection of a PET camera over long durations of time. The energy resolution of the system is not degraded despite shifts in ambient temperature and photodetector heat generation. The thermal regulation scheme also provides a consistent operating environment between separate runs of the camera over different days. Inter-run consistency allows for reuse of system calibration parameters from study to study, reducing the time required to calibrate the system and hence to obtain a reconstructed image.

Robust Timing Calibration for PET Using L1-Norm Minimization

Positron emission tomography (PET) relies on accurate timing information to pair two 511-keV photons into a coincidence event. Calibration of time delays between detectors becomes increasingly important as the timing resolution of detector technology improves, as a calibration error can quickly become a dominant source of error. Previous work has shown that the maximum likelihood estimate of these delays can be calculated by least squares estimation, but an approach is not tractable for complex systems and degrades in the presence of randoms. We demonstrate the original problem to be solvable iteratively using the LSMR algorithm. Using the LSMR, we solve for 60 030 delay parameters, including energy-dependent delays, in 4.5 s, using 1 000 000 coincidence events for a two-panel system dedicated to clinical locoregional imaging. We then extend the original least squares problem to be robust to random coincidences and low statistics by implementing

Spatial resolution uniformity, isotropy, and the effect of depth of interaction information in a 1mm3 resolution, limited-angle PET system

\ell _{1}

First measurements of a 512 PSAPD prototype of a sub-mm resolution clinical PET camera

-norm minimization using the alternating direction method of the multipliers (ADMM) algorithm. The ADMM algorithm converges after six iterations, or 20.6 s, and improves the timing resolution from 64.7 ± 0.1s full width at half maximum (FWHM) uncalibrated to 15.63 ± 0.02ns FWHM. We also demonstrate this algorithm’s applicability to commercial systems using a GE Discovery 690 PET/CT. We scan a rotating transmission source, and after subtracting the 511-keV photon time-of-flight due to the source position, we calculate 13 824 per-crystal delays using 5 000 000 coincidence events in 3.78 s with three iterations, while showing a timing resolution improvement that is significantly better than previous calibration methods in the literature.

A cost-effective modular programmable HV distribution system for photodetectors

Limited-angle tomography PET systems use configurations that result in incomplete and irregular sampling of projection data. In a two paneled, 1mm resolution PET system dedicated to breast imaging with 1mm depth-of-interaction (DOI) resolution, we measure both the uniformity and isotropy of the spatial resolution. Uniformity measures the variation in spatial resolution over the field of view (FOV), and isotropy measures the difference in resolution in the x, y, and z dimensions. Reconstructing data from point sources in air placed across the FOV with MLEM, we measure spatial resolution across the FOV in 3 dimensions. At the center of the FOV, the resolution was 0.775 mm, 1.296 mm, and 0.782mm in x, y, and z, respectively. We examine the impact of MLEM iterations on the measured uniformity and isotropy of spatial resolution. We compare MLEM resolutions to metrics in literature, see some correspondence, and discuss their limitations. We step a point source in 100 μm steps to show the ability of the system to resolve closely spaced points. The system separates point sources by more than one FWHM at 0.700mm spacing. DOI resolutions impact is studied by varying the binning resolution of interactions in crystals along the direction away from the region of interest. Resolution in all directions is substantially impacted by DOI rebinning. Resolution in the DOI direction (orthogonal to the panels) degrades from 1.296mm FWHM at 1mm DOI resolution to 5.908mm with 16mm DOI resolution showing the importance of DOI and sampling from oblique lines of response in limited-angle systems.

3D position-sensitive scintillation detectors for a high-resolution loco-regional PET imaging system (Conference Presentation)

We present measurements from a first prototype of a 1 mm<;sup>3<;/sup> 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 8×8 array of 0.9×0.9×1 mm<;sup>3<;/sup> 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 16,284 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 an energy resolution of 10.62 + 0.04% FWHM at 511 keV, together with a coincidence time resolution of 15.7 + 0.2 ns FWHM. We determined a crystal misidentification probability of <; 0.5% on average. In order to determine 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 cylinders are distinctly visible. Randoms rate was estimated to be 20 %.

Contrast recovery performance for a 1mm 3 resolution clinical PET system

In this paper, we present a scalable HV biasing circuit that is able to control 128 channels between FGND and FGND + 125 Volts (with FGND usually at -1800 Volts). The circuit features two boards: a floating HV board that references power and digital signals to a common floating ground (outside of the imaging head), and a HV bias board that houses four 32 channel DAC and a microcontroller elevated to -1800 V. We were able to bias dual PSAPD modules and obtain high quality flood histograms. We measured the ripple of the HV lines to be less than 20 mV RMS.

Hardware setting optimization for a 1mm 3 resolution clinical PET system

We are building a dual-panel positron emission tomography (PET) system prototype that enables focused imaging of local regions of interest of the body with 1 mm3 spatial resolution, The detector panels are built from novel 3D position sensitive scintillation (3DPSS) detectors comprising arrays of 1x1x1 mm3 LYSO scintillation crystal elements coupled to position-sensitive avalanche photodiodes (PSAPD). At the present state of system construction, the measured energy resolution over 98,304 crystal elements coupled to 1,536 PSAPDs is 11.34%, and 76.2% (74,938) of the system LYSO crystal elements are found to have greater than 99% event positioning accuracy. Imaging studies performed with a high-resolution phantom demonstrate resolution of the smallest (1.2 mm diameter) features. Besides enabling 1 mm resolution clinical PET studies, we describe other novel ways we are planning to exploit the 3DPSS detector information.

Feasibility of software-based real-time calibration of multi-gigabit PET data

We are constructing the worlds first 1 mm3 resolution clinical PET system dedicated to locoregional imaging (e.g. breast, head/neck, etc). The edge-on layout of densely-packed 8×8 arrays of 0.9×0.9×1 mm3 LYSO crystal elements provides high sensitivity and the capability to position the 3-D coordinates of one or more interactions per 511 keV photon event. Contrast recovery (CR), which measures the accuracy of focal tracer accumulation in the field-of-view (FOV), must be evaluated in order to understand the ability of this system to visualize and quantify smaller lesions. The constructed dual-panel system prototype has a sensitive area of 160mm by 33mm, separated by 64mm. Relative CR was evaluated with the different rod sizes of a micro-Derenzo phantom, yielding CR coefficients of 1.00, 0.96, 1.03, 0.88, 0.67, 0.54 for rod sizes of 4.8, 4.0, 3.2, 2.4, 1.8, and 1.2 mm respectively. A micro hollow spheres phantom scan, with sphere diameters of 7.9, 6.2, 5.0, and 4.0 mm suspended in warm background with an activity concentration ratio of 5:1, resulted in CR coefficients of 0.65, 0.65, 0.61, and 0.54 respectively. Finally, an array of activity-filled capillary tubes were suspended in warm background, with an activity concentration ratio of 20:1, resulting in CR ranging from 0.082 at the center of the FOV to 0.074 at 5 cm away from the center of the FOV. These results show the outstanding performance of our 1mm3 resolution locoregional clinical PET imaging system at recovering extremely small contrast structures with realistic background uptake values.