Effects of LSO detector intrisic radioactivity on SPECT imaging

This work was part of a feasibility study of developing SPECT imaging capability on a lutetium oxyorthosilicate (LSO)-based animal PET system. The SPECT aspect of the system was enabled by inserting a collimator assembly inside the detector ring and acquiring data in singles mode. The LSO detectors were used for both PET and SPECT imaging. The intrinsic radioactivity of ¹⁷⁶Lu in the LSO crystals, however, contaminates the SPECT data more severely than for PET imaging. This work was to study the imaging characteristics of the intrinsic LSO activity, the effects of correction, and the minimal activity level required for usable SPECT imaging. To understand the structural and noise patterns of LSO activity in images, measured high counts projections from LSO activity only and simulated multiple-replicates of low counts projections from LSO activity with and without a uniform phantom in the fleld-of-view (FOV) were reconstructed. To identify the minimal activity levels needed for obtaining useful tomographic images with the existence of LSO activity, we derived a formula to estimate the minimal activity in the FOV of one detector crystal ring that would yield a net detectable signal in projection data. A series of phantoms with different activity distribution patterns and total activity levels were scanned for variable length of time through simulation and experiments. The LSO contributions to each study were estimated from a long LSO background acquisition dataset and corrected prior to iterative image reconstruction. The LSO activity showed up as concentric rings in a reconstructed tomographic image. The LSO activity estimation and correction method was effective and it reduced the activity level required for producing images with expected visual patterns to about 1/4 of the level needed without the correction. The minimal activity levels predicted by the projection data based formula correlated well with the activity levels needed for obtaining visible image patterns observed in simulated and experimental studies. Phantom studies demonstrated that while ¡mage quality was dependent on source distribution pattern, activity level, and acquisition time, SPECT imaging on LSO based animal PET is feasible with practical dose level and acquisition time used in animal imaging.