Design and simulations of a self-collimating whole-body SPECT system

2021 
1135 Objectives: Clinical SPECT systems suffer fundamental limitations of low resolution (1~2 cm) and low sensitivity (0.01%~0.02%) from mechanical collimators. To overcome this limit, we propose an innovative self-collimating SPECT design concept. With a multi-layer interspaced mosaic detector (MATRICES) architecture, the spatially distributed sensitive scintillators play the role of collimation instead of absorptive heavy-metal materials. This design concept breaks away from the interdependency of resolution and sensitivity of a conventional SPECT system and simultaneously improves both features. In this work, we propose a modular self-collimating clinical SPECT system design and evaluate its performance through numerical simulations. Methods: The self-collimating SPECT system consists of 6 MATRICES detector heads that form a hexagon with an 800-mm bore size. Each MATRICES detector head has 4 scintillator layers on behind of a high opening ratio (4%) metal plate. The 3 scintillator layers in the middle consist of 75 (tangential) × 37 (axial) GAGG(Ce) crystals with a size of 4.0 mm (tangential) × 4.0 mm (axial) × 6.0 mm (radial). The space between neighboring scintillators is also 4.0 mm (tangential) × 4.0 mm (axial), i.e. the scintillators are interspaced that forms a mosaic pattern. The outermost scintillator layers consist of 150 × 75 GAGG(Ce) crystals with the same size. There is no gap between neighboring scintillators in this layer. We developed a numerical simulation program for system matrix derivation with multi-ray-tracing method. The simulated system has a FOV size of 500 mm (∅) × 300 mm(L) with a voxel size of 0.5 mm × 0.5 mm × 0.5 mm. We simulated a uniform cylindrical source across the FOV to measure the detection efficiency. We evaluated the resolution performance with a planar hot rod phantom. It consists of 6 sectors of hot rods with diameters of 1.25, 1.5, 1.75, 2.0, 2.5 and 3.0 mm respectively. We also simulated a contrast phantom study with 6 hot sphere inserts (1.5, 2.0, 2.5, 3.0, 3.5 and 4.0 mm respectively in diameters) in a 50 mm (∅) × 10 mm (L) cylinder with a warm-background. The concentration ratio of all the hot spheres to the background is 5:1. We define the contrast-to-noise ratio (CNR) to quantitatively evaluate image quality. Projection data for phantom study at different counts levels were generated using forward projection and through sampling Poisson noise. OS-EM algorithm was used for image reconstruction with 300 subsets. We ran the reconstruction until convergence and then performed post-filtering with an empirically chosen Gaussian filter. Results: The peak and average detection efficiency in the full FOV is 0.54% and 0.47% respectively. The planar hot rod phantom study shows that the best achievable trans-axial image resolution is 1.25 mm in noise-free condition and 2.5 mm with 2.6 mCi Tc99m total activity and 5 min scan time. The smallest visible hot sphere in the contrast phantom study is 1.5 mm with a total activity of 2.6 mCi and 100 min scan. With 5 min acquisition, the 3.0 mm hot sphere is clearly visible. Conclusions: The preliminary result shows that the proposed self-collimating MATRICES SPECT design has the potential to yield superb whole-body imaging performance in terms of simultaneous, order-of-magnitude improvement of resolution and sensitivity, as well as excellent image contrast-to-noise performance. It is reasonable to expect a strong clinical application advancement with the self-collimating SPECT technology. Further investigations are on-going to fully characterize the system performance with Monte-Carlo simulations. Research Support: This research was supported by the National Natural Science Foundation of China (No. 81727807), National Key Research and Development (R&D) Plan of China (Grant ID. 2019YFF0302503) and Tsinghua University Initiative Scientific Research Program.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    0
    Citations
    NaN
    KQI
    []