現今高解析度針孔單光子放射電腦斷層掃描系統廣為用於偵測小動物模型，但由於單針孔準直儀接收角度有限致使靈敏度太低，因此本研究針對實驗室建構之單光子放射電腦斷層掃描系統引入多針孔準直儀，以提高系統之靈敏度。 設計多針孔準直儀的過程，首先透過選定 FOV 範圍及最大化使用偵測器面積的方式訂定系統放大率，再藉由共線模型預估投影影像並限制其範圍及多工效應條件限制的方式選定針孔數量及針孔擺放位置，最後以幾何數學模型初估系統解析度及靈敏度來訂定針孔大小，藉由以上的步驟初步得到六組適合的針孔準直儀設置及系統放大率；接下來再透過蒙地卡羅模型軟體 GATE 模擬這六組針孔準直儀，獲取影像系統矩陣；最後以此影像系統矩陣進行偵測任務，比較這六組針孔準直儀的影像品質，並使用傅立葉串擾矩陣計算其空間解析度、重建熱桿假體影像分析空間解析度及計算系統靈敏度，比較這六組針孔準直儀的系統性能。藉由評估結果，最終選用系統放大率 1.52 ， 多工效應 20% 之四個直徑 0.6 mm，對著物空間中心傾斜放置的多針孔準直儀，為我們系統最佳化之多針孔準直儀，搭配偵測器可用面積為 49 x 49 mm2 ，所得最佳系統解析度為 1 mm，靈敏度為$2.2 x 10-4。 High resolution pinhole-SPECT systems are generally applied to small-animal nuclear medicine imaging, but the small acceptant solid angle of the single pinhole will limit the sensitivity. Therefore, multi-pinhole apertures are introduced into the micro-SPECT system developed in our group to raise the sensitivity. This study aims to design the multi-pinhole configuration that optimizes the spatial resolution and the sensitivity simultaneously. The design procedure starts from deciding the multi-pinhole pattern. First, we choose the system magnification based on the predetermined field-of-view (FOV) and the criterion of using maximum detector area. Second, we choose the number of pinholes and the pinhole locations by setting an upper bound for the multiplexing factor and avoiding truncated projections. Third, we choose the pinhole size to have comparable spatial resolution among candidate pinhole patterns, where the sensitivity and resolution are preliminary evaluated by analytical models. After implementing the procedure, we obtain six single- and multi-pinhole patterns and their corresponding system magnifications. The next step is to model the designed pinhole patterns in GATE Monte-Carlo simulations to generate the imaging system matrices. The final step is to compare the Area-Under-Curve (AUC) values, the sensitivities and the spatial resolutions of the designed pinhole patterns. The AUC values are evaluated with their respective imaging system matrices through signal detection tasks. The sensitivities are calculated during the generation of the system matrices as the ratio of detected counts divided by the number of emitted photons. The spatial resolutions are calculated by the Fourier crosstalk approach and visualized by reconstruction images of a hot-rod phantom. According to the resulting AUC, sensitivity and spatial resolution, the four-pinhole pattern with 20% multiplexing, 0.6-mm pinhole diameter, and 1.52 times system magnification is the optimized configuration for our micro-SPECT system with a camera face of 49 × 49 mm2 and a spherical FOV of 14-mm diameter. The corresponding system resolution is 1.0 mm, and the sensitivity is 2.2 ×10-4.
