Implementation and Image Quality Benefit of a Hybrid-Space PET Point Spread Function

High resolution PET imaging is being driven by recent advances in both scanner hardware and image reconstruction. As a result, images are being reconstructed with smaller pixel sizes and little or no post-reconstruction filtering. In certain circumstances, this combination can lead to distracting image artifacts due to high frequencies introduced by the projectors. We present a method that utilizes PET point spread function (PSF) processing to mitigate these artifacts while preserving image resolution. Our approach includes a "hybrid-space PSF," with components in both sinogram-space and image space. Radial smoothing of the sinogram data (or a broadened line of response if performing list mode reconstruction) is combined with a spatially invariant image smoothing. Unlike fully-image-based PSF approaches, this method allows a PSF kernel specific to each crystal pair, even when corrected for patient motion. The hybrid-space PSF approach also opens the door to isotope-dependent positron range imaging, by allowing the use of different image-based kernels for isotopes with different positron range, while preserving the same projection-based kernel (which is a function of detector design only). The effectiveness of the hybrid-space PSF approach is demonstrated with a brain
¹⁸
F-FDG dataset and spatial resolution point sources.