A-PET is a quad-head PET scanner designed for small-animal imaging. The detector module consists of 22 × 22 array of 2 × 2 × 10 mm3 Lutetium-yttrium oxyorthosilicate (LYSO) scintillators. The gap between the detector modules is minimized for high sensitivity system. The size of the volumetric field of view (FOV) is 46.1 × 46.1 × 46.1 mm3. The small FOV and the quad-head geometry causes image quality degradation. The main image degrading factor of the quad-head PET is the mis-positioning events caused by the penetration effect in the detector. Thus, a precise system model accounting for the penetration effect in the detector is needed for high spatial resolution.
In this paper, we propose and develop a precise system modeling method including the detection probability of every possible ray path via crystal sampling methods. The sampling points are determined by evenly spaced points along the diagonal line of the crystal, and the diagonal line of a crystal is determined by the relative position of the other crystal pair. Sub-LORs are defined by connecting the sampling points of the crystal pair. We incorporated the detection probability of each sub-LORs by calculating the penetration effect. The LOR-based ML-EM (Maximum likelihood expectation maximization) was used.
As a reference, we used the Monte-Carlo simulation based system model approach, and evaluated the reconstructed image using the National Electrical Manufacturers Association NU 4-2008 standards. We also used the Geant4 Application for Tomographic Emission simulation toolkit (GATE).
The average spatial resolution of different locations is 1.77 mm by the proposed system model and 1.79 mm for the LOR system model, which does not include the penetration effect. The standard deviation of the uniform region in the NEMA image quality phantom is 2.14 % for the proposed method and 14.3 % for the LOR system model.
The proposed approach is also flexible in changing the system and the image parameters. In comparison to the LOR system model, the proposed precise system model showed improved results in spatial resolution and uniformity of the reconstructed image.
In the future, we will evaluate the system model with the experimental data. We will include the physical effects obtained during data acquisition process such as inter-crystal scatter into system model to improve the spatial resolution of the reconstructed image. We further plan to extend the two-dimensional reconstruction algorithm to three-dimensional approach for higher signal to noise ratio.