目的：正子斷層造影掃描是在核子醫學上一項非侵襲性的檢查技術，提供臨床上功能性生理資訊。目前常用三維擷取模式來提高造影的靈敏度，但隨機與散射同符事件因此提高。一般隨機事件修正是利用即時的延遲時間窗修正法；散射事件修正是利用單次散射光子模擬法，然而以上修正法皆有缺失影響修正後的準確性。為有效矯正正子造影非真實事件，故提出利用蒙地卡羅模擬旋繞式射束阻擋裝置修正法。材料與方法：本研究使用GATE 蒙地卡羅軟體模擬microPET R4 系統以及射束阻擋修正法。利用射束阻擋裝置在不同角度下兩組正弦圖受檔塊遮蔽下的資訊，交叉計算出其非真實事件的貢獻量，因非真實事件分布是緩慢變化函數，所以可根據擋塊遮蔽部份的取樣點利用cubic-spline 內插法得到整張正弦圖中非真實事件分布情形，再進行矯正。結果：本實驗研究優點在提出能窗範圍較大靈敏度提升的同時，掃描時間保持固定，射束修正法矯正的非真實事件分率31.55% 與蒙地卡羅非真實事件分率29.62% 的結果最相近，而且矯正後假體區域活度分佈也較接近實際活度比例的分配。結論：旋繞式射束阻擋法在不額外增加造影掃描時間的情況下，利用射束阻擋裝置轉換位置來提高實驗取樣結果，矯正效果接近蒙地卡羅結果。相信本研究所提出的矯正法是更適合於正子斷層造影臨床上應用。

Purpose : Position Emission Tomography (PET) is a non-invasive technique used in nuclear medicine,which provides physiological information using molecular tracers. As 3D PET data acquisition increases, random and scatter coincidence events increase relevantly. They cause a uniform
distribution of background on the image and degrade the accuracy of quantitative analysis. The conventional corrections for random and scatter coincidences are achieved by the delayed window technique and Single Scatter Simulation (SSS) method. However, they increase noise and reduce image accuracy.
Material and Methods : The aim of this study is to examine the easibility of using a beam stopper (BS) for correcting non-true coincidence events. The BS placed on the line of response (LOR) at two different locations rotating with constant degrees absorbs a particular fraction of the true events.
The non-true coincidence, non-blocked at the LORs by each stopper, can be estimated. Assuming that the non-true radiation is a spatially slow-varying distribution, the whole non-true sinogram can be recovered using cubic-spline interpolation from these local measurements. Beam stopper rotated (BSR) reduces the effective sampling distance without increasing the number of the stoppers. In this study, we performed Monte Carlo (MC) simulations for 3-D PET with the GATE software for a cold phantom to conduct the BSR method.
Result : The estimate of non-true event fraction in a large FOV using BSR method is 31.55% and is close to the result of Monte-Carlo simulation. The activity ratio of region approaches the initial input quantity.
Conclusion : We conclude that the proposed correction method is effective and better than conventional correction schemes without increasing scan time. In sum, the BSR method is a convenient and effective correction for non-true coincidences in 3D PET.
[Therapeut Radiol Oncol 2011; 18(1): 59-67]