质心引导的最大似然定位算法的模拟

The Simulation of Centroid Guided Maximum Likelihood Positioning Algorithm

目的:单光子发射计算机断层成像系统(SPECT)的一个关键问题是:根据光电倍增管(PMT)的输出信号反推出探测器晶体上γ射线击中晶体内的作用位置,从而获取精确的投影数据。本文在系统分析用于SPECT事件定位的质心算法和最大似然算法的基础上,模拟并实现了质心导引的最大似然定位算法。方法:首先用质心法计算出事件发生的位置,并对其进行非线性校正;然后在质心法经过校正后确定的事件击中位置的可能的范围内,用最大似然法进行γ光子作用位置的精确估计。结果:质心引导的最大似然法估算的γ光子作用位置,可以和原始位置精确地吻合,定位误差能控制在0.2mm以内;运算时间较普通最大似然法降低30倍以上。结论:质心算法的应用,大大缩小了搜寻范围,可以降低一般最大似然法对计算机硬件和存储空间的要求,加快运算速度。综合来讲,质心引导的最大似然定位算法可用于探测器的设计。

Objective： A key problem in SPECT （Single Photon Emission Computed Tomography） is to work out the location of hit by gamma rays in the crystal, from the output signals of the PMTs （Photomultiplier Tubes） in order to acquire the accurate projection data. Based on the systematic analysis of the centroid weighted （CW） and the maximum likelihood （ML） positioning algorithms for SPECT, a method called centroid guided maximum likelihood （CGML） algorithm was realized by simulation. Methods： CW algorithm is used firstly to calculate event position cursorily, and the positions of the events are accurately determined by ML algorithm between the positioning error regions after non-linear calibration. Results： The position estimated by CGML algorithm can accurately inosculate with the primary position, the positioning error can be controlled in 0.2mm, and the computation time was decreased by 30 thales than the general ML algorithm. Conclusion： The searching region is greatly reduced by using the CW algorithm which can reduce the memory requirement and computation burden for the general ML calculation, In conclusion, this algorithm can be used in the design of the detector.