Assessment of partial volume effect in small animal cardiac PET imaging using Monte Carlo simulation

Objectives Partial volume effect (PVE), the image blurring caused by the finite spatial resolution of PET imaging system, can cause incorrect quantification of radioactivity especially for small objects. High-resolution small animal PET system has been developed for rodents, but sizes of rodent hearts are still close to the spatial resolution, that might cause significant PVE. The aim is to evaluate PVE in different sizes of rodent hearts in a small animal PET system utilizing Monte Carlo (MC) simulation and realistic rodent digital phantom.

Methods We built a simulation platform using a MC simulation (SimSET+GATE) and reconstruction software (STIR) based on the scanner configuration of a dedicated small animal PET system (Inveon). A line source in a water-filled cylinder was measured and simulated at the center of field of view. Different sizes of realistic rodent digital phantoms (MOBY and ROBY) were simulated with F-18 and Ga-68. Averaged myocardial counts over half maximum at the lateral left ventricular (LV) wall were compared with background on short-axis images reconstructed by 2D OS-EM with attenuation correction and 1.4mm Gaussian post-filtering.

Results Full widths at half maximum measured from the sinogram were about 1.7mm (3.9mm) for F-18 (Ga-68) in MC simulation and 1.8mm (3.7mm) in experiment. Simulation of realistic rodent heart digital phantoms demonstrated thickness-dependent count reduction of 32%, 41%, 51% and 67% at 2.7mm, 1.8mm, 1.0mm and 0.7 mm-thick LV walls for F-18, and even higher reduction of 42%, 56%, 57% and 70% for Ga-68, respectively.

Conclusions We developed a MC simulation platform to evaluate rodent cardiac PET imaging and estimated PVE for different heart sizes and positron ranges. Established combined MC simulation and realistic digital phantom platform for the rodent cardiac PET imaging has the further potential use for developing and evaluating new PVE compensation algorithms.

Research Support This work was supported by the CHFC grants CF 1.6.