基于MOA方法的PBRTQC步长寻优机制改进研究

The Research on Improvement of PBRTQC Step Size Optimization Mechanism Based on MOA Method

目的基于患者的实时质量控制(PBRTQC)是一种新型质控技术,近年来受到越来越多的关注。但是由于缺少寻优机制,遍历的繁琐降低了该方法的效率。对原始算法的遍历手段进行改进,以提升该类算法的转化率。方法基于重叠面积最小化(MOA)的参数寻优方法,优化了PBRTQC的执行效率,在遍历参数之前,提前对内部参数进行确定,在统一的硬件化境下,比较二者的寻优结果及执行效率。结果以指数加权移动平均法(ewma)算法,在3个项目下,两种方法对于步长的实验结果吻合,采用了MOA预分析后的PBRTQC,执行时间相对于传统遍历方法提升了近86.8%。结论预分析方法可以改进传统遍历手段,该方法可作为PBRTQC算法的辅助,提升其转化率。

Objective Patient-based real-time quality control(PBRTQC)is a novel quality control technique that has drawn increasing attentions in recent years.However,due to the lack of optimization mechanisms,the tediousness of traversal reduces the efficiency of this method.The traversal means of the original algorithm are currently improved to enhance the conversion rate of this class of algorithms.Methods Based on the minimization of overlapping area(MOA)parameter optimization method,the execution efficiency of PBRTQC was optimized by determining the internal parameters in advance before traversing the parameters.We then compared the optimization results and execution efficiency of the two methods under a unified hardware environment.Results Based on the ewma algorithm through three projects,the experimental results of the two methods for step size were consistent,and the execution time of PBRTQC after MOA pre-analysis was improved by nearly 86.8%,compared to the traditional traversal method.Conclusion The pre-analysis method can improve the traditional traversal means.This method can be used as an auxiliary approach for PBRTQC algorithms to improve their conversion rate.