A comparison between solid angle and joseph line integral reconstruction for small animal PET systems

Animal PET scanners are typically designed to image small subjects, such as mice and rats. This makes the tomographic reconstruction sensitive to parallax affects, and makes resolution recovery essential for good quality images. In this study, data was acquired on a small animal PET scanner. This system processes raw listmode data into quasiparallel Lines Of Response (LORs) to generate 3D sinograms, and geometrical corrections are applied during the image reconstruction. Thus, the reconstruction uses a solid angle (system matrix) approach, such as LOR based Point Spread Function Order Subset Expected Maximum (PSF-OSEM3D). In this study, the data is geometrically binned into uniformly sampled, parallel projection, 3D sinograms, and the reconstruction uses the Joseph line integral. The data is normalized and reconstructed with four iterations and sixteen subsets. No attenuation or scatter corrections are applied during the reconstruction to eliminate any bias that they may contribute. The resulting images are compared in resolution recovery and uniformity, and the results show that the Joseph line integral using image based PSF-OSEM3D resulted in a similar signal to noise ratio, but with a better resolution recovery.