Characterization of a compressed sensing PET detector and data acquisition system: Effects of multiplexing and sampling rate

Reducing the number of readout channels of PET detectors with multiplexing can save the cost and complexity for developing a scanner. However, a high multiplexing ratio can also degrade the signal-to-noise (SNR) ratio of the detector since the noise of the multiplexed channels is summed together. Compressed sensing (CS) was recently shown to be a promising multiplexing method for PET detectors that can achieve high multiplexing ratio while resolving simultaneous hits on multiple pixels with high SNR recovery for PET detectors. In this work, we study the effects of multiplexing and sampling rate on the performance of the CS detectors and the corresponding DAQ system we have developed for a PET insert for simultaneous PET/MR brain imaging. The coincidence time resolution acquired by a pair of single-crystal detectors degraded from 1.4 ns to 3.2 ns when the sampling rate dropped from 20GHz (sampled by the digital oscilloscope) to 65MHz (sampled by the 256-channel DAQ system developed in house). The coincidence time resolution further degraded to 3.7 ns when the coincidence events were acquired by two CS detectors (with a multiplexing ratio of 8:1) and sampled by the DAQ system, indicating the effect of multiplexing on the timing performance. Possible future studies to diagnose and mitigate the causes of timing performance degradation are discussed.