Abstract
1746
Objectives: Detector Response Function (DRF) in PET is spatially variant due to interdetector penetration effect. The introduction in iterative reconstruction methods of a transition matrix (TM) that accurately models the relationship between image and projection spaces, avoids typical errors in both position and extent of reconstructed sources, especially off-center, appearing when ignoring the DRF. Methods: The DRF obtained from the linear attenuation model (Lecomte 1984) was introduced in a TM for a 256x256 OSEM reconstruction. Model DRFs were compared to related experimental sinogram profiles, acquired placing a capillary source 20 cm away from the FOV center. An 'hot spot' phantom was acquired too, and images were reconstructed using both i) OSEM software including the system DRF model (OSEM-M) and ii) OSEM proprietary standard software (OSEM-P). Brain 'in vivo' data were also reconstructed using both OSEM-M and OSEM-P methods and compared to related CT studies. Results: The analysis of the spectral resolution (SR) showed a good agreement between experimental and modeled DRF. A slight underestimation of the modeled DRF SR was highlighted, as inter-crystal scatter is neglected. While OSEM-M images of the capillary source correctly relocate the spot position, OSEM-P images show a mispositioning offset of about 12mm. Comparable shift was found on phantom reconstructions too, where a reliable picture of the phantom was superimposed to both OSEM-M and OSEM-P images. Gaussian fits of OSEM-P 10cm off-center spot profiles showed a compression toward the center of about 5mm. Distances between typical anatomical markers recognized in CT images resulted correctly reproduced by OSEM-M images, while OSEM-P images feature the same spatial mispositioning highlighted in phantom studies. Conclusions: The proposed analytical DRF model can be easily suited to fit the hardware characteristics of any peculiar PET system. Its straightforward implementation in the kernel of an iterative reconstruction software allows restoration of spatial mispositioning, particularly when the image periphery is under evaluation. As it concernes the spatial resolution, while qualitative evaluations seems suggesting an improvement in OSEM-M images, at present accurate quantitative tests are in progress.
- Society of Nuclear Medicine, Inc.