Quantitative Analysis of PET Reconstruction Techniques over a Wide Activity Range with 2D and 3D Acquisition Modes

Quantitative assessment of activity uptake within features seen on positron emission tomographic (PET) images is an important property of this modality. Methods: a phantom containing anatomically relevant features was imaged on a dedicated PET-CT system using F-18 over a wide range of activity within and outside the imaging field-of-view (FOV). A series of short 2D and 3D acquisitions was obtained over a 12-hour period; images were reconstructed using 2D filtered backprojection (FBP), 2D OSEM, 3D reprojection and fully-3D iterative methods. Mean values of activity concentration were recorded within regions-of-interest drawn on single slices through tissue background, liver and lung. Estimates of activity concentration from each acquisition and reconstruction condition were compared to values extrapolated from a a set of low-activity 2D FBP images. Results: All PET scan modes and reconstruction methods provided similar quantitative imaging performance, with differences less than 3.5% in average ROI value between scan modes or reconstruction methods in background tissue, and less than 7% for this same comparison in liver. The ratio of lung to tissue ROI activity was lower using analytic reconstruction compared to that found using iterative methods. In general, the iterative reconstruction methods demonstrated less variability in mean ROI value over the wide activity range as compared to the reconstruction using analytic methods.