Assessing PET performance, image quality, and attenuation correction in the presence of RF coils within a 9.4 T MRI for preclinical simultaneous imaging.

BACKGROUND

This study evaluates the performance of a positron emission tomography (PET) insert combined with a 9.4 Tesla magnetic resonance (MR) imaging system using three different radiofrequency (RF) coils. The impact of the RF coils within the PET field of view (FoV) on key PET performance parameters was carefully examined. Additionally, the accuracy of MR-based attenuation correction (AC) on the quantitative PET results was assessed.

PROCEDURES

The performance of the PET scanner was assessed according to the NEMA NU 4-2008 standards, focusing on sensitivity, spatial resolution, noise-equivalent count (NEC) rate, and image quality metrics (homogeneity, spillover ratios, and recovery coefficients). These evaluations were conducted with three different RF coils within the PET FoV: a mouse-dedicated coil, a rat-dedicated coil (not optimized for PET), and a big-rat-dedicated coil. Quantitative testing of AC accuracy was also performed using homogeneously filled phantoms of varying sizes and activity levels, comparing images reconstructed with and without AC.

RESULTS

The peak absolute sensitivity varied depending on the RF coil used, though spatial resolution was unaffected by the presence of the coils. The NEC curve for the mouse phantom peaked at 441.2 kcps at 29.3 MBq with the mouse-dedicated coil. For the rat phantom, NEC was only tested with the rat-dedicated coils, and the NEC peak reached 203.1 kcps at 27.5 MBq with the big-rat-dedicated coil. Activity concentrations in the NEMA image quality phantom were underestimated when standard MR-AC was applied, likely due to the absence of phantom walls in the generated µ-maps. However, for uniformly filled phantoms with wall thicknesses under 1 mm, standard MR-AC provided accurate quantification.

CONCLUSIONS

Overall, the presence of RF coils had only a minor effect on the performance of the PET/MR insert, with the most significant variation observed for the non-PET-optimized coil. Attenuation correction based on generated µ-maps performed well for phantoms with minimal wall thickness.