Feasibility of in vivo small animal imaging using a clinical total-body PET/CT system

Clinical PET scanners have long been explored for preclinical imaging, but limited spatial resolution and sensitivity have restricted their use for preclinical studies. The recent availability of total-body (TB) PET/CT scanners with extended axial fields of view (FOVs) has largely overcome sensitivity limitations, enabling potential new opportunities for small-animal imaging. This study evaluated the feasibility and performance of the Biograph Vision Quadra TB-PET/CT for rodent imaging compared to the dedicated small-animal PET scanner Inveon DPET.

Recovery coefficients (RC), image noise, and optimum image reconstruction parameters were assessed using the preclinical NEMA NU 4-2008 image quality phantom and a sub-cohort of three anesthetized mice as a proof-of-concept demonstrating the feasibility of the setup. In vivo quantification accuracy was evaluated by scanning nine frozen mice simultaneously in three different arrangements with the Quadra compared with individual scans at the Inveon. To ensure comparability, all mice were snap-frozen after 1 h uptake of [¹⁸F]FDG, scanned sequentially and individually at the Inveon (90 min p.i.), and subsequently scanned at the Quadra with decay-corrected acquisition times. SUV_mean and SUV_max values were determined for liver, muscle and brain regions on both systems. To evaluate potential position-dependent effects within the extended axial FOV, a single frozen mouse was scanned at multiple positions.

Phantom rods ≥ 2 mm could be resolved with the Quadra, showing a comparable RC for larger structures, e.g. for the 5 mm rod of 1.17 compared to 1.09 (Inveon) when using point-spread-function modeling, whilst having lower noise of 5.1%SD vs 9.0%SD. No substantial position-dependent effects were detected in the phantom or single-mouse scan across the axial FOV. SUV_mean values were consistent between both scanner across all investigated organs, with liver and muscle uptake remaining comparable for frame durations down to 5 s. SUV_max values exhibited greater variability, with significant differences observed in muscle and brain regions.

Despite the lower spatial resolution of the clinical TB-PET/CT scanner (~ 3-4 mm) compared to the dedicated preclinical scanner (~ 1.5 mm), robust SUV_mean quantification was achievable. Together with successful in vivo imaging of anesthetized mice, these findings support the feasibility of using clinical TB-PET/CT for preclinical research, acknowledging spatial resolution as a limiting factor.