Scanning multiple mice in a small-animal PET scanner: Influence on image quality

To achieve high throughput in small-animal positron emission tomography (PET), it may be advantageous to scan more than one animal in the scanner’s field of view (FOV) at the same time.\ud \ud However, due to the additional activity and increase of Poisson noise, additional attenuating mass, extra photon scattering, and radial or axial displacement of the animals, a deterioration of image quality can be expected. In this study, the NEMA NU 4-2008 image quality (NU4IQ) phantom and up to three FDG-filled cylindrical “mouse phantoms” were positioned in the FOV of the Siemens Inveon small-animal PET scanner to simulate scans with multiple mice. Five geometrical configurations were examined. In one configuration, the NU4IQ phantom was scanned separately and placed in the center of the FOV (1C). In two configurations, a mouse phantom was added with both phantoms displaced radially (2R) or axially (2A). In two other configurations, the NU4IQ phantom was scanned along with three mouse phantoms with all phantoms displaced radially (4R), or in a combination of radial and axial displacement (2R2A). Images were reconstructed using ordered subset expectation maximization in 2 dimensions (OSEM2D) and maximum a posteriori (MAP) reconstruction. Image quality parameters were obtained according to the NEMA NU 4-2008 guidelines. Optimum image quality was obtained for the 1C geometry. Image noise increased by the addition of phantoms and was the largest for the 4R configuration. Spatial resolution, reflected in the recovery coefficients for the FDG-filled rods, deteriorated by radial displacement of the NU4IQ phantom (2R, 2R2A, and 4R), most strongly for OSEM2D, and to a smaller extent for MAP reconstructions. Photon scatter, as indicated by the spill-over ratios in the non-radioactive water- and air-filled compartments, increased by the addition of phantoms, most strongly for the 4R configuration. Application of scatter correction substantially lowered the spill-over ratios, but caused an over-correction for the recovery coefficients of the FDG-filled rods when the phantom was displaced radially. Image noise was not substantially influenced by scatter correction. In conclusion, when scanning 2 mice, axial displacement (2A) is preferable above to radial displacement (2R) since for axial displacement, the recovery coefficients are higher and spill-over ratios are lower, whereas image noise remains similar. In the case of scanning 4 mice, combined axial and radial displacement (2R2A) is preferable to just radial displacement (4 R)