Reconstruction of a Dual-Head Small-Animal PET System: An SVD Study

We have proposed the use of two large-area detector panels with a small detector spacing for achieving high sensitivity for small-animal PET imaging. The stationary geometry results in missing views and data truncations, creating challenges in image reconstruction. The proposed system geometry also provides great flexibility in the imaging configuration, such as permitting changes in the detector spacing and field-of-view (FOV) definition, and therefore creates more variables to consider in reconstruction. In this work, we employ the singular value decomposition (SVD) analysis to obtain theoretical analysis on how changes in the FOV definition of the dual-head small-animal imager affects imaging. We study three FOV definitions and identify the characteristic reconstruction patterns for them as revealed by SVD. Our results suggest that change in the FOV definition can considerably affect the sensitivity to noise and the characteristic patterns of the resulting image artifacts.