Compact FPGA-Based Data Acquisition System for a High-Channel, High-Count-Rate TOF-PET Insert for Brain PET/MRI

With the development of advanced positron emission tomography (PET) systems, processing a large number of channels in real-time has become common, particularly for applications with one-to-one coupled crystals and silicon photomultipliers (SiPMs). In this study, we present the design and evaluation of a data acquisition (DAQ) system for a radio frequency (RF)-penetrable time-of-flight (TOF)-PET brain-dedicated insert featuring lutetium-yttrium oxyorthosilicate crystals that are one-to-one coupled to SiPMs. Each detector module comprises 768 crystal-SiPM channels, and the PET insert ring consists of 16 such modules. Achieving real-time processing of 12 288 crystal-SiPM channels with a designed maximum 20 kcps singles count rate per channel in a compact design is a significant challenge. We developed a field-programmable gate array (FPGA)-based DAQ system based on a commercial evaluation board and characterized its performance for the full PET ring. Our results demonstrate an energy resolution of 11.99 +/- 0.01% FWHM and a coincidence time resolution (CTR) of 240.5 +/- 0.3 ps FWHM with online DAQ processing. Furthermore, all the parameters, including energy resolution, CTR, coincidence count rate, and random count rate, change by less than 5% compared with an offline algorithm that has access to unlimited resources and no deadtime requirements. A high-dose fluorodeoxyglucose (FDG) experiment showcases the DAQ's high-throughput capability, which can handle up to 5.3 mCi in the field of view with no measurable saturation. Our study demonstrates the capability of our DAQ system for handling a high number of channels and high count rate, and it suggests its potential for this and other PET system applications.