Initial results of a truly pixelated APD-based PET scanner for high-resolution preclinical imaging

Objectives: The LabPET II is a new generation APD-based PET scanner designed to achieve sub-mm spatial resolution using fully pixelated detectors and highly integrated parallel front-end processing electronics based on time-to-digital conversion in a dual-threshold Time-over-Threshold (ToT) scheme. The key objectives of this work are (a) to investigate the detector physical characteristics in this ToT scheme, (b) to assess the imaging performance of the mouse scanner version following the NEMA NU 4-2008 standard, and (c) to demonstrate its unique capability for in vivo imaging in small animals. Methods: The basic detector element uses a 4 x 8 array of 1.12 x 1.12 mm2 Lu1.9Y0.1SiO5:Ce (LYSO) scintillator pixels with one-to-one coupling to a 4 x 8 pixelated monolithic APD array at a pitch of 1.2 mm. Four detector arrays are mounted on an interposer carrying two custom flip-chip, 64-channel, mixed-signal, application-specific integrated circuits (ASIC) on the backside interfacing to two detector arrays each. Fully parallel signal processing was embedded on-chip to encode time and energy information using the dual-threshold ToT scheme. The self-contained 128-channel front-end module was designed as a generic OEM component for ultra-high resolution imaging of small to medium-size animals. The LabPET II mouse scanner consists of 48 front-end modules forming a 78.8-mm diameter by 50.4-mm axial length cylinder, with 32 rings of 192 pixel detectors per ring. The energy, time and intrinsic spatial resolution of the detectors were measured as a function of APD bias and ToT thresholds. Imaging performance data include spatial resolution, absolute sensitivity, count rates, scatter and random fractions, and image quality with phantoms. Images of mice and small rats were obtained using 18F-, 11C-, 64Cu- and 89Zr-based tracers in various static, dynamic and gated imaging protocols. Results: ToT spectra show clearly resolved 511 keV photopeak and Compton edge for each individual pixel. After correction for nonlinear ToT response, energy resolution is typically 22 ± 2% FWHM. Average coincidence time resolution between opposing 128-channel modules is typically 3.3 ± 0.6 ns FWHM. The measured intrinsic spatial resolution obtained by sweeping a 0.3-mm diameter 22Na point source midway between opposite coincident detector arrays was 0.82 ± 0.02 mm FWHM and 1.54 ± 0.05 mm FWTM at a pitch of 1.21 ± 0.01 mm, confirming accurate and distortion-free position measurement, in sharp co